article
stringlengths 0
745k
| abstract
stringlengths 0
92.7k
|
|---|---|
* samples . * both als and flash experiments used samples microfabricated on silicon nitride membranes supported in a silicon window frame .
the ura and test pattern for the als experiments were fabricated using center for x - ray optics nanowriter , lawrence berkeley national lab .
polymethyl methacrylate , a positive resist , was patterned on si@xmath6n@xmath7 membrane substrate and subsequently gold electroplated ( 10 nm au ) .
patterning dose was achieved in a fraction of a second , enabling potential mass production .
following glutaraldehyde fixation , spiroplasma cells were air dried from a solution containing s. melliferum on a si3n4 window covered by a 10 nm poly - l - lysine .
the ura was fabricated next to a spiroplasma cell using focused ion beam milling at the national center for electron microscopy .
* at the als , diffaction patterns were collected using a coherent portion of a soft x - ray beam ( @xmath2=2.3 nm ) from an undulator source selected by a 5 @xmath1 m pinhole .
the hologram was collected with an in - vacuum back illuminated ccd ( 1300x1340 20 @xmath1 m pixels ) placed at 200.5 mm from the specimen .
the direct beam is blocked by a beamstop placed in front of the detector to prevent damage to the camera .
the beamstop is moved during data collection to recover a large portion of the diffraction pattern .
total collection time was 5 seconds .
at flash the same ccd is placed at 54.9 .
mm from the specimen and collects elastically scattered x - rays filtered by a graded multilayer mirror .
a hole in the mirror allows the direct beam through , removing the need for a beamstop . * reconstruction . * the autocorrelation map , obtained by fourier transform of the diffraction pattern ,
was multiplied by a binary mask 0 in the region outside the cross correlation between the object and the ura ( the hologram ) . ) .
the reconstruction was performed by applying the same processing used for pinhole camera images , by replacing the recorded intensity image with the masked complex valued cross correlation term . a cyclical convolution with a ura decodes the hologram .
the ura autocorrelation is a delta function in periodic or cyclical systems , not when surrounded by empty space . to mimic the cyclical correlation
, a mosaic of 3x3 binary uras is convolved with the holographic cross term .
the reconstruction procedure retrieves the hologram of the object to a resolution determined by the size the ura elements ( at sub - array spacing resolution ) .
however when the spacing between dots is larger than the dots , the array no longer scatters optimally the available light , decreasing the snr .
the signal is proportional to the number of elements in the array , and the noise is proportional to the square root of the number of elements used to deconvolve the image .
only 2x2 uras contribute to the reconstructed image since the holographic term is at most twice the size as the array itself ( the object is smaller than the array ) .
the mosaic uras are made of + 1 and -1 ( instead of 1 and 0 ) terms , yielding an additional factor of @xmath8 the noise .
the signal to noise ratio therefore increases as @xmath9 . for a phase ura
the signal to noise ratio would increase by a factor of 2 .
* phase retrieval .
* the reference points in the ura and a few dust particles where fixed in space throughout the optimization process , facilitating phase retrieval optimization@xcite .
the addition of a redundant linear constraint yields more reliable reconstructions , de facto increasing the resolution of the retrieved complex valued images .
the rest of the illuminated sample was reconstructed with dynamic support@xcite .
the reproducibility of the image as a function of spatial frequency@xcite drops at 75 nm resolution .
99 stroke , g. w. _ introduction to coherent optics and holography _ , ( academic press , new york , ny 1969 ) .
mcnulty , i. et al .
high - resolution imaging by fourier transform x - ray holography
. _ science _ * 256 * , 1009 - 1012 ( 1992 ) .
bajt , s chapman , h. n. spiller , e. a. alameda , j. b. woods , b. w. frank , m. bogan m. j. , barty a. , boutet s. , marchesini s , hau - riege s. p. , hajdu j. , shapiro d. a camera for coherent diffractive imaging and holography with a soft - x - ray free electron laser , appl .
opt . in press . | advances in the development of free - electron lasers offer the realistic prospect of high - resolution imaging to study the nanoworld on the time - scale of atomic motions .
we identify x - ray fourier transform holography@xcite , ( fth ) as a promising but , so far , inefficient scheme to do this .
we show that a uniformly redundant array@xcite ( ura ) placed next to the sample , multiplies the efficiency of x - ray fth by more than one thousand ( approaching that of a perfect lens ) and provides holographic images with both amplitude- and phase - contrast information . the experiments reported here demonstrate this concept by imaging a nano - fabricated object at a synchrotron source , and a bacterial cell at a soft x - ray free - electron - laser , where illumination by a single 15 fs pulse was successfully used in producing the holographic image .
we expect with upcoming hard x - ray lasers to achieve considerably higher spatial resolution and to obtain ultrafast movies of excited states of matter . , the ideal microscope for the life and physical sciences should deliver high - spatial - resolution high - speed snapshots possibly with spectral , chemical and magnetic sensitivity .
x - rays can provide a large penetration depth , fast time resolution , and strong absorption contrast across elemental absorption edges .
the geometry of fourier transform holography is particularly suited to x - ray imaging as the resolution is determined by the scattering angle , as in crystallography . the problem with conventional fourier holography is an unfavourable trade off between intensity and resolution .
we show here how recent technical developments in lensless flash photography can be used to reduce this effect . around the @xmath0 century
, european painters used the camera obscura , a dark camera with a pinhole to form an image on a canvas .
even earlier , pinhole cameras had been used to image solar eclipses by chinese , arab and european scientists .
scientists and painters knew that a small pinhole was required for reaching high resolution , but the small pinhole also dimmed the light and the image@xcite .
it was eventually discovered that lenses could collect a larger amount of light without degrading resolution .
shutters and stroboscopic flash illumination allowed recording of the time evolution of the image .
the development of the computer allowed a resurgence of pinhole techniques and random arrays of pinholes were used , initially in x - ray astronomy @xcite .
each bright point of a scene deposits a shadow - image of the pinhole array on the viewing screen .
depth information about the object is encoded in the scaling of the shadow image of the object points@xcite .
knowledge of the geometry of the pinhole array ( the `` coded aperture '' ) allowed numerical recovery of the image .
eventually the pinholes were replaced by binary uras which were shown to be optimal for imaging@xcite .
their multitudes of sharp features contain equal amounts of all possible spatial frequencies , thereby allowing high spatial resolution without sacrifice of image brightness .
ura coded apertures are now commonly used in hard x - ray astronomy@xcite , medical imaging@xcite , plasma research@xcite , homeland security@xcite and spectroscopy@xcite to improve brightness where lenses are not applicable .
the forerunner of our x - ray holography with a ura reference source is conventional visible - light fth @xcite .
the interference pattern between light scattered by an object and light from a nearby pinhole is recorded far downstream ( fig .
1 ) . when this recording ( the hologram ) is re - illuminated by the pinhole reference wave , the hologram diffracts the wavefront so as to produce an image of the object .
a second inverted ( `` twin '' ) copy of the object is produced on the opposite side of the optical axis . under far - field measurement conditions ,
a simple inverse - fourier transform of the fth recording produces an image of the specimen convolved with the reference pinhole source . as in the pinhole image of the camera obscura , the brightness of the image ( or equivalently
, the signal to noise ratio ) increases as the reference pinhole increases in size , at the expense of image resolution . in general the solution to
the problem of weak signal from a single pinhole ( and resulting long exposure time ) is to use multiple reference sources .
for example , a unique mesoscale object has been imaged by x - ray fth using 5 pinholes sources@xcite .
no two pairs of pinholes were the same distance apart , so that each holographic term could be isolated in the autocorrelation map .
this geometry blocks much of the available light and limits the number of pixels available to image the specimen . hitherto efforts to produce strong signal have been pursued using complicated reference objects@xcite .
the possible improvement in signal to noise of fourier - transform holograms with a strong reference saturates@xcite at 50% of that for an ideal lens - based amplitude image with loss of phase information .
however there is still the difficulty of deconvolution due to the missing frequency content of the reference .
the flat power spectrum of the ura is designed to optimize the reference to fill the detector with light uniformly . in summary ,
the optimum method of boosting the holographic signal is to use a ura as the reference object .
the gain in flux compared to a single pinhole is the number of the opaque elements in the ura ( twice as much for phase uras ) , in our case , n=2592 and n=162 . the signal to noise ratio ( snr ) of the ura - produced image increases initially with the square root of the number of pixels in the ura@xcite ( see methods ) , in our case by a factor of 18 and 4.5 for n=2592 and n=162 respectively .
the x - ray fth experiment is conceptually simple : a coherent beam of x - rays impinges on a specimen and the coded array , and the diffraction pattern is recorded far downstream ( fig 1 ) .
we report here two experiments both using an area detector fitted to an existing experimental end station@xcite , and already used in several recent coherent x - ray diffractive imaging experiments@xcite .
the first experiment was carried out at beam line 9.0.1 at the advanced light source ( als ) at the lawrence berkeley national laboratory .
an x - ray beam defined using a 4 @xmath1 m coherence - selecting pinhole was used to image a test object placed next to a 44 nm resolution array ( figure 2 ) .
the flux advantage of the ura method is illustrated by the calculations shown in figs .
2 ( ) and ( f ) and explained in the caption is made evident from our results .
73 array of 43.5 nm square gold scattering elements , imaged by scanning electron microscopy .
scale bar is 2 @xmath1 m .
( b ) diffraction pattern collected at the advanced light source ( @xmath2=2.3 nm , @xmath3 photons in 5 s exposure , 200 mm from the sample ) .
( c ) real part of the reconstructed hologram ( linear grayscale ) , the smallest features of the sample of 43.5 nm are clearly visible .
( d ) simulation with @xmath3 photons .
( e ) simulation with the same number of photons , but a single reference pinhole .
( f ) cut through two dots separated by 120 nm .
[ fig:2 ] , scaledwidth=48.0% ] the second series of experiments was carried out at the flash soft x - ray free electron laser ( @xmath2=13.5 nm ) in hamburg .
a 15 fs pulse of @xmath4 photons traverses the sample and ura , and is diffracted just before they both turn into a hot plasma@xcite , and become vapourised .
a bacterium was imaged to demonstrate that the experiment was feasible with the lower scattering strength of biological material ( fig .
3 ) .
more information on these experiments is given in the figure captions and methods section .
in fth , the final image is produced by an especially simple one - step calculation .
a fourier transformation of the measured intensity pattern delivers the autocorrelation of the wavefield that exits the object plane . in standard fth with a point reference source
such an autocorrelation would already contain the image .
for fth with a ura reference source , it includes the convolution of the object and the ura .
positioning the sample and the array with a sufficient spacing between them ensures that this information - containing term will be separated from the autocorrelations of the array and the object . to extract the final image ,
we convolved the holographic term ( with the rest of the autocorrelation map set to zero ) using a mosaic of @xmath5 uras , that is using the same delta - hadamard transform used to reconstruct the pinhole camera image intensities , but replacing the intensity image with the complex valued cross correlation term .
the reconstructed images made by this method are shown in figure 2 using 44-nm ura elements .
the forward or small angle scattering was discarded during data collection , yielding edge enhancement in the image . for the biological image additional refinement was demonstrated . by using the fourier - hadamard - transform image provided by a 150 nm resolution ura as the starting point for a phase retrieval algorithm , we refined the 15-fs flash image of a small helical bacterium ( spiroplasma melliferum ) to the full extent of the recorded diffraction pattern at half the pinhole - size resolution ( 75 nm ) ( fig . 3 and methods section ) .
m ) .
( b ) diffraction pattern collected at flash in a single 15 fs ( @xmath2=13.5 nm ) pulse .
( c ) reconstructed image by phase retrieval methods@xcite .
( d ) the reproducibility of the recovered image as a function of spatial frequency drops at 75 nm resolution .
[ fig:3 ] , scaledwidth=48.0% ] the resolution of the holograms corresponds to the resolution of the fabricated uras : 44 nm for the lithographic pattern used at als , and 150 nm ( refined to 75 nm ) for the bacterium used at flash .
these values are among the best ever reported for holography of a one - micron - sized object , and we believe resolution will improve in the future with the development of nano - arrays . uras with 25 nm resolution and 9522 elements have already been produced by conventional methods . in conclusion ,
we have successfully demonstrated holographic x - ray imaging with uras and obtained amplified x - ray holographic images at attractive resolutions .
these images were orders of magnitude more intense than those from conventional fourier transform holography , enabling the potential use of novel tabletop sources@xcite . since ura diffraction uniformly filled the detector with light , image reconstruction could be performed by a fourier - hadamard transform in a single step without iterations .
the technique shows good prospects for improvements in the spatial resolution , and the results verify the predicted high performance values with respect to high brightness and ultrafast time resolution .
imaging with coherent x - rays will be a key technique for developing nanoscience and nanotechnology , and or massively parallel holography will be an enabling tool in this quest .
we are grateful to the staff of flash and als for help , and to d. a. fletcher for the spiroplasma samples .
this work was supported by the following agencies : the u.s .
department of energy by lawrence livermore national laboratory under contract w-7405-eng-48 and de - ac52 - 07na27344 ; the advanced light source ; national center for electron microscopy ; center for x - ray optics at lawrence berkeley laboratory under doe contract de - ac02 - 05ch11231 ; the stanford linear accelerator center under doe contract de - ac02 - 76-sf00515 ; the european union ( tuixs ) ; the swedish research councils , the dfg - cluster of excellence through the munich - centre for advanced photonics ; the natural sciences and engineering research council of canada to m. b. , and the sven and lilly lawskis foundation of sweden to m.m.s . |
recently there has been considerable interest in models of spacetimes with relatively large extra dimensions , in which the familiar standard model fields are confined to a `` brane '' , a 3-dimensional defect in the larger space , while gravity propagates in all the dimensions ( the `` bulk '' ) . in some models@xcite , the extra dimensions can be as large as the current direct experimental gravitational probes of the order of a millimeter@xcite .
the apparent ( usual ) planck mass in 3 + 1d , @xmath3 , is related to the true fundamental scale @xmath4 by @xmath5 where @xmath6 is the volume of @xmath7 extra large dimensions . in another class of models ( `` nonfactorizable geometries '' , @xcite )
, the extra dimensions can be even larger , but have curvature @xmath8 which traps gravitons in a bound state close to a brane .
the curvature radius @xmath9 is again on a mesoscopic scale which may be as large as @xmath10 mm .
the cosmology of these models might well include radical departures from the orderly evolution of the standard model , including episodes of violent mesoscopic geometrical activity at relatively late times .
new classical geometrical effects remain important long after inflation , until the hubble length @xmath11 is larger than the size or curvature radius @xmath12 of the largest extra dimensions@xcite .
of particular interest are two new geometrical degrees of freedom common to many of these models : `` radion '' modes controlling the size or curvature of the extra dimensions@xcite , and new nambu - goldstone modes corresponding to inhomogeneous displacements of the brane in the extra dimensions@xcite .
previous treatments have focused on microscopic thermal and quantum emission in these modes , but cosmological symmetry breaking can also create large - amplitude , coherent classical excitations on much larger scales ( of order @xmath13 ) as the configuration of the extra dimensions and the position of the brane settle into their present state .
this highly dynamic geometric activity generically produces a conspicuous detectable relic : an intense , classically generated background of gravitational radiation . in this paper
i explore the possibility of a direct detection of a gravitational wave background generated by coherently excited radion and nambu - goldstone modes on the threshold of standard cosmology .
the main features of the background spectra can be estimated from general scaling considerations , without reference to particular models .
[ [ frequency ] ] _ frequency : _ + + + + + + + + + + + + the characteristic gravitational ( `` hubble '' ) frequency@xcite redshifted to the present day is @xmath14 or : @xmath15 the estimate is valid back to the threshold of the extradimensional dynamics , @xmath16 .
the mapping between observed frequency , temperature @xmath17 and hubble length is shown by axis labels in figure 1 for frequencies in the lisa band . and @xmath18 ) . even above 100 gev
we expect standard relativistic cosmology derived from relativity and thermodynamics to hold(e.g.@xcite ) , @xmath19 and @xmath20 . ]
extra dimensions between @xmath21 mm and @xmath22 mm produce backgrounds peaked in the lisa band ( @xmath0 to @xmath1 hz ) ; observations with ligo ( up to @xmath23hz ) could detect , to detect waves with @xmath24 , which may be possible with future instrumentation . ] activity from dimensions down to @xmath25 mm . [ [ amplitude ] ] _ amplitude : _ + + + + + + + + + + + + the mechanisms considered here excite geometrical degrees of freedom in approximate equipartition of energy density with the thermal relativistic matter . a stochastic background of gravitational radiation with rms metric perturbations @xmath26 over bandwidth
@xmath27 contributes a fraction of the critical density @xmath28 . in an experiment such as lisa
a stochastic background can only be distinguished from other sources of noise and astrophysical wave sources by resolving the background in frequency ( and to some extent , direction and polarization ) .
the rms strain produced in a lisa frequency resolution element is@xcite @xmath29^{1/2}.\ ] ] the reference density is set@xcite by the mean energy density in all relativistic species ( photons and three massless neutrinos ) , @xmath30 , where @xmath31 refers to the hubble constant . since the energy density of gravitational waves redshifts like relativistic matter , this gives a maximal bound for primordial broad - band backgrounds , shown in figure 1 along with projected lisa [email protected] ( corresponding to @xmath32 , millisecond pulsar timing gives a limit at 90% confidence@xcite , @xmath33
the radion can be a significant source of mesoscopic activity if its potential has a first - order phase transition .
the radion is stuck initially in a metastable state with some initial value @xmath34 and thermal or quantum nucleation leads to randomly nucleated regions corresponding to the final value @xmath35 , accompanied by a release of internal energy .
the gravitational waves from this mode are easiest to describe if for the largest extra dimension , @xmath36 : the final dimensional stabilization then happens within the context of an approximately 3 + 1-dimensional cosmology . on our brane
this process resembles nucleation of bubbles or vacuum domains of higgs scalars@xcite , with the extra complication of modifications in gravity . in
an expanding universe@xcite the bubbles collide and overlap , creating flows of energy with velocities of the order of unity .
the coherence scale @xmath37 of the flows is determined by nucleation dynamics which generally yields@xcite @xmath38 .
a number of models have been used to estimate the gravitational radiation in similar situations , from colliding bubbles and the resulting energy flows in the context of qcd and electroweak phase transitions@xcite . in the absence of a definite model of radion bubbles
, we estimate the maximal gravitational wave spectrum , based on establishing a sustained relativistic turbulent cascade up to the scale @xmath37 for a time @xmath13 .
the power output of a system in gravitational waves@xcite , @xmath39 , is determined by the ( changing quadrupolar ) flow of mass - energy @xmath40 , where @xmath41 and the numerical factor @xmath42 is typical of simple asymmetric geometries .
flows of mass - energy on the hubble scale produce a gravitational wave power per volume close to @xmath43 , and integrated over time @xmath13 produce broad - band @xmath44 at characteristic frequency @xmath45 ( now shifted to @xmath46 ) .
flows on smaller scale @xmath37 create a spectrum peaked at higher frequency @xmath47 and with a smaller amplitude .
relativistic motions in a volume @xmath48 with density perturbations of order unity , @xmath49 create a mass - energy flow @xmath50 which if sustained for time @xmath13 gives @xmath51 . in a narrow band
this translates to amplitude @xmath52 . to estimate the low - frequency spectrum consider motions of smaller velocity on the same length scale , @xmath53 .
the mass - energy flow is @xmath54 , hence the gravitational wave power @xmath55 .
the maximal tail of low - frequency waves then has @xmath56 , hence @xmath57 .
these estimates lead to the maximal spectrum shown in figure 1 for @xmath58tev and @xmath59gev .
( in the quieter case where turbulence is not sustained , the amplitude is less than this estimate by @xmath60 . )
in many scenarios there is another degree of freedom , the position of the brane .
spatial inhomogeneities in this displacement correspond to nearly - massless nambu - goldstone modes@xcite . in the cosmological formation of the brane / defect ,
the position is in general a random variable uncorrelated on large scales , and large - scale modes are excited by the kibble mechanism . since the modes also represent curvature of the brane , they are efficiently coupled into gravitational waves as viewed on the brane .
this mode is generally excited if for the largest extra dimension , @xmath61 : the 3-brane condenses as a defect before cosmology enters the 3 + 1-d era .
suppose our brane forms at some early time as a defect in @xmath62 space , spontaneously breaking the poincar symmetry of the full theory .
the position of the brane in each higher dimension @xmath63 can be described by a field @xmath64 which depends on the coordinate @xmath65 on the brane.s are analogous@xcite to group - transformation angles @xmath66 in the goldstone description of the pion or axion , but have dimensions of length .
they are related to the canonically - normalized physical fields @xmath67 by @xmath68 ( rather than @xmath69 ) where @xmath70 is the brane tension . when a @xmath71 is `` eaten '' it gets a mass @xmath72 or @xmath73 for @xmath74tev . ]
these fields represent propagating nambu - goldstone bosons , new modes in addition to the standard model fields on the brane and gravitation propagating in the bulk . in a cosmological
setting , the initial values of the @xmath75 s are not generally correlated on large scales but are random for points with large separation , since the position of the brane , when it condenses as a defect , is determined by local accidents .
a topologically stable 3-wall forms for example if vacua in a 4-bulk fall into two discrete degenerate minima ; the spontaneous choice of one minimum or the other is a random variable at large separation , so the wall is not formed initially in its ( flat ) ground state , but in an excited ( wrinkled ) state@xcite .
this is the same `` kibble mechanism '' which leads in other contexts to formation of cosmic strings , axion miniclusters , or goldstone excitations of scalar fields ( which are also gravitational wave sources , e.g. @xcite ) .
if the cosmology thereby excites these modes , the brane is wrinkled on all scales up to the dimensional size or curvature radius @xmath76 ( which we take to be fixed here ) , curving through the extra dimensions . ignoring the subtleties of the cosmological metric and imagining the brane in a flat spacetime
, displacements directly induce an effective 4-metric on the brane @xmath77 , where the perturbed 4-metric is @xmath78 .
although these initial perturbations are scalars , no symmetry prevents transverse - traceless tensor components@xcite from being excited dynamically for waves with @xmath79 , creating classical gravitational waves .
kibble excitations corresponding to perturbations of magnitude @xmath80 on scale @xmath81 thereby lead to wave amplitudes of the order of @xmath82 in the cosmological context , the kibble excitation is regulated by the expansion .
the brane location is initially uncorrelated on large scales , so on the hubble scale variations in @xmath83 are of order @xmath13 , producing waves with amplitude @xmath84 , comparable in total energy with the other forms of energy on the brane ( @xmath85 ) which now appear in @xmath86 .
these waves show up today as gravitational waves at the redshifted hubble frequency @xmath87 .
the maximal spectrum corresponds to a background of the order of the energy density in brane fields ( that is , @xmath88 on each scale as it enters the horizon ) , leading to constant @xmath89 over a range of higher frequencies .
the amplitude of excitation is reduced after the universe enters the classical 3 + 1-d era , when @xmath90 ( or after @xmath91 ) ; variations in @xmath83 are at most of order @xmath35 ( or @xmath92 ) , leading to @xmath93 and to a spectrum with a corresponding @xmath94 rolloff at lower frequencies .
the damping scale and low - frequency spectrum contain information on the scale of extra dimensions and/or the stabilization of the brane .
the spectrum at higher frequencies is a probe of the cosmological model during the multidimensional / preclassical period , including the excitation mechanism .
the predicted backgrounds can be distinguished from other expected signals by their isotropy , and by the distinctive spectra with rolloffs at both low and high frequencies . under some circumstances these backgrounds
could be stronger than previously contemplated sources@xcite .
i am grateful for useful discussions with e. adelberger , p. bender , d. kaplan , a. nelson , s. sigurdsson , e. witten , and especially g. fuller .
i thank the aspen center for physics , the max - planck - institut fr astrophysik , the isaac newton institute for mathematical sciences and the ettore majorana centre for scientific culture for hospitality .
this work was supported at the university of washington by nasa , and at the max - planck - institute fr astrophysik by a humboldt research award .
arkani - hamed , n. , dimopoulos , s. , and dvali , g. phys .
lett . b429
( 1998 ) 263 antoniadas , i. , arkani - hamed , n. , dimopoulos , s. , and dvali , g. , phys .
lett . b436
( 1998 ) 257 arkani - hamed , n. , dimopoulos , s. , and dvali , g. phys .
d59 ( 1999 ) 086004 hoyle , c. d. , heckel , b. , adelberger , e. , schmidt , u. , gundlach , j. , swenson , e. , & kapner , d. , in preparation ( 2000 ) randall , l. and sundrum , r. , phys.rev.lett .
83 ( 1999 ) 3370 randall , l. , and sundrum , r. , `` an alternative to compactification '' , hep - ph/9906064 lykken , j. and randall , l. `` the shape of gravity '' , hep - ph/9908076 csaki , c . ,
graesser , m. , randall , l. and terning , j. , `` cosmology of brane models with radion stabilization '' , hep - ph/9911406 arkani - hamed , n. , dimopoulos , s. , kaloper , n. , and march - russell , j. 1999 , `` rapid asymmetric inflation and early cosmology in theories with sub - millimeter dimensions '' , hep - ph/9903224 goldberger , w. d. and wise , m. , `` phenomenology of a stabilized modulus , '' hep - ph/9911457 sundrum , r. , phys.rev .
d59 ( 1999 ) 085009 hogan , c. j. , `` cosmological gravitational wave backgrounds , '' in _ second international lisa symposium on the detection and observation of gravitational waves in space _ , aip conference proceedings 456 , ed . w. m. folkner , pp . 79 - 86 ( 1999 ) , astro - ph/9809364 maggiore , m. , `` gravitational wave experiments and early universe cosmology '' , phys . rep . in press
( 2000 ) , gr - qc/9909001 lisa study team , _ lisa pre - phase a report _ ( 1998 ) ; this document and others are available at http://lisa.jpl.nasa.gov/documents/ armstrong , j. w. , estabrook , f. b. , and tinto , m. , astrophys .
j. 527 ( 1999 ) 814 kolb , e. w. and turner , m. s. , _ the early universe _ , ( addison - wesley,1990 ) heckler , a. and hogan , c. j. , phys .
d 47 ( 1993 ) 4256 kainulainen , k. , kurki - suonio , h. , and sihvola , e. phys.rev .
d59 ( 1999 ) 083505 carr , b. j. , ann .
astrophys . 32
( 1994 ) 531 thorsett , s. and dewey , r. , phys .
d53 ( 1996 ) 3468 coleman , s. , phys .
d. 15 ( 1977 ) 2929 peierls , r.e . , singwi , k. s. and wroe , d. phys .
87 ( 1952 ) 46 ignatius , j. , kajantie , k. , kurki - suonio , h. and laine , m. phys.rev .
d49 , 3854 ( 1994 ) kurki - suonio , h. , and laine , m. , phys.rev.lett . 77
( 1996 ) 3951 hogan , c. j. phys .
b133 ( 1983 ) 172 witten , e. , phys .
d 30 ( 1984 ) 272 hogan , c. j. , mon . not .
218 ( 1986 ) 629 kamionkowski , m. , kosowsky , a. , and turner , m. s. , phys.rev .
d 49 ( 1994 ) 2837 gleiser , m . and roberts , r. , phys.rev.lett . 81 ( 1998 ) 5497 - 5500 misner , c. w. , thorne , k. s. , and wheeler , j. a. , _ gravitation _ ( freeman , 1973 ) thorne , k. , in the proceedings of the snowmass 95 summer study on particle and nuclear astrophysics and cosmology , eds . e. w. kolb and r. peccei ( world scientifc , singapore ) , gr - qc/9506086 ; see also gr - qc/9704042 zeldovich , ya . b. , kobzarev , i. yu . and okun , l. b. , zh . eksp . teor .
67 ( 1974 ) 3 [ sov .
jetp 40 ( 1975 ) 1 ] .
kibble , t. w. b. , j. phys .
a 9 ( 1976 ) 1387 vilenkin , a. , phys .
48 ( 1982 ) 59 krauss , l. m. , phys .
b 284 ( 1992 ) 229 hogan , c. j. , phys .
74 ( 1995 ) 3105allen , b. , in _
les houches school on astrophysical sources of gravitational waves _ , eds .
jean - alain marck and jean - pierre lasota , ( cambridge university press , 1996 ) ; gr - qc/9604033 | recent models which describe our world as a brane embedded in a higher dimensional space introduce new geometrical degrees of freedom : the shape and/or size of the extra dimensions , and the position of the brane .
these modes can be coherently excited by symmetry breaking in the early universe even on `` mesoscopic '' scales as large as 1 mm , leading to detectable gravitational radiation .
two sources are described : relativistic turbulence caused by a first - order transition of a radion potential , and kibble excitation of nambu - goldstone modes of brane displacement .
characteristic scales and spectral properties are estimated and the prospects for observation by lisa are discussed .
extra dimensions with scale between 10 and 1 mm , which enter the 3 + 1-d era at cosmic temperatures between 1 and 1000 tev , produce backgrounds with energy peaked at observed frequencies in the lisa band , between @xmath0 and @xmath1 hz . the background is detectable above instrument noise and astrophysical foregrounds if initial metric perturbations are excited to a fractional amplitude of @xmath2 or more , a likely outcome for the nambu - goldstone excitations . |
colossal magneto - resistive ( cmr ) materials have provided an opportunity to understand the intricate relationship between their structural , electronic and magnetic degrees of freedom .
the magnetic order in these materials is essentially ferromagnetic which is due to parallel alignment of the spin of the hopping @xmath7 electron between mn@xmath8-o - mn@xmath9 pairs due to double exchange mechanism @xcite .
study of these materials becomes more complex at low temperatures owing to the rich magnetic , electronic and structural phases exhibited by them .
recent finding of coexistence of ferroelectricity and magnetism in manganites has only added to the existing complexity @xcite .
khomskii and coworkers @xcite have pointed out that coupling between magnetic and charge - ordering in charge - ordered and orbital - ordered perovskites can give rise to ferroelectric magnetism .
although , the cmr properties have been investigated extensively , the physics behind their varied low temperature ground states is still unclear .
the search for a microscopic picture of the ground state especially in half - doped manganites remains a very active research field .
compounds belonging to the series sm@xmath0pr@xmath1sr@xmath2mno@xmath3 ( @xmath10 ) exhibit significantly different low temperature ground states with @xmath4 and provide an opportunity to understand the causes for attaining such varied ground states . in this series , end member sm@xmath2sr@xmath2mno@xmath11 undergoes paramagnetic - insulating ( pmi ) to ferromagnetic - metallic ( fmm ) phase transition below 95 k @xcite .
the other end member pr@xmath2sr@xmath2mno@xmath3 exhibits pmi phase at rt , fmm phase below 255 k and a orbitally ordered antiferromagnetic - insulating ( afi ) phase below 140 k @xcite .
however , at intermediate doping level of @xmath5 , pr@xmath12sm@xmath13sr@xmath2mno@xmath3 exhibits ce type charge ordered ( co ) state below 155 k starting from pmi phase at rt and fmm phase in the temperature range 215 k to 155 k @xcite .
presence of co state , for the intermediate doping , evokes interest in understanding the role of rare - earth ions in inducing the co ground state in the material .
role of the weighted average radius ( @xmath14 ) of the a - site cation and the size - mismatch of a - site cations is well known in co materials @xcite .
smaller a - site cation size is reported to support charge - ordered state due to the smaller e@xmath15 electron bandwidth ( w ) whereas greater size mismatch in a - site cations has been reported to destroy charge - ordering and render the manganite ferromagnetic @xcite . in all these investigations ,
the role of mn - o - mn bond angle and also mn - o bond distance have been the centre of focus in deciding the ground state of these materials .
ferromagnetic double - exchange , antiferromagnetic superexchange , jahn - teller ( jt ) distortion of manganese octahedra and charge - ordered state invariably depend upon these quantities .
the multiferroic properties of manganites have also been attributed to tilting of polyhedra or frustrated magnetism @xcite .
some studies on charge - ordered samples have also indicated the role of covalent mixing of o(2p@xmath16 ) and a - site cations , on t@xmath17 and t@xmath18 @xcite .
it has been argued that the covalent mixing of the anion o(2p@xmath16 ) orbital with the a - site cation 4f orbitals , competing against its covalent mixing with the mn 3d@xmath7 orbital , plays an important role in deciding the properties of manganites .
mizokawa et al have also expressed the possibility of a 4f - o(2p@xmath19 ) band to accommodate some holes leading to reduction in the effective hole concentration in the mn 3d - o(2p@xmath20 ) band @xcite .
hence , local structure around mn sites needs to be probed carefully to understand the role of mno@xmath6 octahedra in various phases exhibited by such samples .
such investigations using mn k - edge exafs have been undertaken earlier in some cmr materials @xcite . however , similar mn k - edge exafs study on sm@xmath13pr@xmath12sr@xmath2mno@xmath3 is not possible as mn k - edge exafs is severely affected by exafs of pr / sm l - edges .
recently it has been shown that @xcite that information about mno@xmath6 octahedra could be deduced from the study of rare - earth k - edge exafs .
hence it will be interesting to investigate the local structure of a - site cations in various phases of a co compound in order to glean an insight on the role of mno@xmath6 octahedra in stabilizing a particular ground state .
we report herein exafs of a - site cation ( pr / sm ) in pmi and ce - type afi charge - ordered phases of sm@xmath13pr@xmath12sr@xmath2mno@xmath3 .
polycrystalline sm@xmath13pr@xmath12sr@xmath2mno@xmath3 sample was synthesized using the conventional solid - state reaction method using the preparation technique mentioned in @xcite .
the samples were characterized by xrd , resistivity and ac susceptibility measurements .
x - ray powder diffraction patterns were recorded with cu k@xmath21 radiation ( @xmath22 = 1.5418 ) at room temperature between @xmath23 = 20@xmath24 to 100@xmath24 with a step 0.02@xmath24 on a siemens d5000 diffractometer
. low field a.c .
susceptibility measurements were carried out during warming from liquid nitrogen temperature to 300 k. standard four - probe method was used to measure resistivity of the samples in the temperature interval of 80k to 300k .
exafs spectra at pr and sm k - edges were recorded using bl01b1 xafs beam - line at spring-8 .
si(311 ) crystal plane served as the monochromator .
the first mirror and the monochromator were fully tuned in order to get optimal resolution .
the slit width of the monochromator exit was 0.3 mm vertical and 6 mm horizontal to ensure good signal to noise ratio .
the photon energy was calibrated for each scan with the first inflection point of pr k - edge in pr metal ( 42002 ev ) .
both the incident ( i@xmath25 ) and the transmitted ( i ) synchrotron beam intensities were measured simultaneously using ionization chambers filled with a mixture of 15% ar and 85% n@xmath26 gases and 100% ar gas respectively . the absorbers were made by pressing the samples into pellets of 10 mm diameter with boron nitride as binder .
the thickness of the absorber was adjusted such that @xmath27 was restricted to a value in the range 0.5 to 1 , where @xmath28 is edge step in the absorption coefficient and @xmath4 is the sample thickness .
data analysis was done using the athena and artemis @xcite .
room temperature xrd patterns of the sample , presented in fig.[fig : xrd ] , shows primarily a single phase sample .
no crystallographic data is available in literature on this sample .
however , the diffraction pattern is very similar to that of sm@xmath2sr@xmath2mno@xmath11 which is reported to have orthorhombic structure @xcite .
indexing our diffraction pattern with pnma space group identifies all the reflections and gives lattice parameters as @xmath29 = 5.4301 , @xmath30 = 7.6471 , @xmath31 = 5.4733 .
susceptibility as a function of temperature presented in fig.[fig : sus ] and the four - probe resistivity measurements ( not shown ) performed on these samples closely match with those reported in literature @xcite .
weighted exafs spectra , @xmath32 versus @xmath33 plots of sm@xmath13pr@xmath12sr@xmath2mno@xmath3 at room temperature ( r.t . ) at pr and sm k edges are presented in fig.[fig : chik ] .
it can be seen that exafs oscillations are visible upto 16 @xmath34 indicating usable data range .
fourier transforms ( ft ) ( not corrected for phase shift ) of pr k - edge exafs in pr@xmath12sm@xmath13sr@xmath2mno@xmath3 at r. t. and 50k are presented in fig.[fig : fouripr ] .
new features appear at around 2.4 and 3.25 at 50 k as compared to that at room temperature .
this is perhaps due to redistribution of pr - o coordination numbers at 50 k. in addition , a slight shift of the main peak in ft towards higher r - side indicates longer pr - mn bonds at 50 k. it reflects a gradual increase in pr - mn bond lengths from rt to 50 k. pr / sm k - edge exafs were fitted with k@xmath35 weighting in k - range of 2 to 16 @xmath34 and in the r - range between 1 to 3.5 to include pr / sm - o and pr / sm - mn bond lengths . as no structural report is available for this compound , the theoretical exafs function ( @xmath36 ) including amplitude and phase , coordination numbers and bond distances ( given as calculated bond lengths in tables [ prexafs ] and [ smexafs ] ) were calculated using the crystallographic data available for isostructural and charge ordered compound , nd@xmath2sr@xmath2mno@xmath11 @xcite with lattice parameters obtained from xrd .
the value of k@xmath37 = 16 @xmath34 gives a resolution in r - space of about 0.1 .
therefore all the pr / sm - o bond lengths with difference less than 0.1 were grouped together .
for example , pr has one oxygen neighbour at 2.418 ( written as 2.418 @xmath38 1 ) and 2 oxygens at 2.443 .
these were clubbed togather as one bond distance at 2.43 with a coordination number 3 .
thus three pr - o bond lengths at 2.43 , 2.69 and 3.06 with coordination numbers of 3 , 6 and 3 respectively and two pr - mn bond lengths at 3.23 and 3.35 with coordination number 2 and 6 respectively were realized .
using these constraints the theoretical exafs function was fitted to the experimental data with the bond length and mean square relative dispalcement ( @xmath39 ) along with amplitude reduction factor and @xmath40 as variable parameters .
the coordination numbers was kept fixed to the above distribution .
the final fitted parameters for pr k - edge exafs are presented in table [ prexafs ] and fitting in k - space is shown in fig .
[ fig : prqfits ] .
it can be seen that the one of the three pr - o bond length obtained from exafs analysis is much longer than the one calculated from xrd .
it must be noted here that one obtains an average picture from xrd and hence local distortions in the structure are masked in comparison with exafs which is essentially a local probe . here ,
since there are three cations ( pr , sm and sr ) occupying the a site , such a distortion could be expected .
the exafs at 50 k can not be fitted with the above distribution of bond lengths and coordination numbers .
different combinations of coordination numbers were tried and the best fits were obtained for two cases : one with coordination number of 4 for all three pr - o and two pr - mn bond lengths and second one with 6 , 3 and 3 combination for three pr - o bond lengths and 4 + 4 combination for pr - mn bond lengths . the second model was rejected as it gave negative @xmath39 for one of the sm - o bond length when it was applied to sm k exafs .
fourier transforms of sm k - edge exafs of sm@xmath13pr@xmath12sr@xmath2mno@xmath3 and their fitting in k - space are presented in fig .
[ fig : foursm1 ] and fig .
[ fig : sm1qfits ] respectively .
sm - o and sm - mn bond lengths are relatively smaller compared to the corresponding pr - o and pr - mn bond lengths [ table [ smexafs ] ] , due to the smaller ionic radius of sm ion .
however , the changes in sm - o and sm - mn distances in going from 300k to 50k are similar to those in pr - o and pr - mn respectively .
furthermore , it may also be noted from table [ prexafs ] and [ smexafs ] that there is an overall contraction of pr / sm - o bonds and expansion of pr / sm - mn bonds as the system evolves from pmi ( 300k ) to co ( 50k ) state .
.[prexafs ] structural parameters of pr ion in sm@xmath13pr@xmath12sr@xmath2mno@xmath3 .
figures in brackets indicate uncertainty in the last digit . [
cols="<,^,^,^,^,^ " , ] local structure investigations on this material are modeled on orthorhombic structure in space group pnma . in going from pmi to co state
most of the pr / sm - o and pr / sm - mn bond - lengths remain nearly similar .
the changes happen in the coordination numbers which change from 3 + 6 + 3 to 4 + 4 + 4 for pr / sm - o bond lengths and from 2 + 6 to 4 + 4 for pr / sm - mn bond lengths .
the pr / sm - o bond length at about 2.6 at 300k is made up of 4 feff calculated pr / sm - o bond lengths spread from 2.59 to 2.88 . this wide distribution in the bond distances is reflected in the corresponding value of @xmath39 which is more than 0.02 . the changes in pr / sm - o coordination numbers at rt and 50 k [ tables [ prexafs ] and [ smexafs ] ] can be understood in terms of rotation and tilting of mn - o octahedra along b - axis . in charge - ordered state ,
the neighboring mno@xmath6 octahedra rotate about b - axis in opposite directions as shown in fig .
[ prsmunit ] .
this cooperative rotation effect affects mainly the pr / sm - o distances in the @xmath41 and @xmath42 planes . in effect
one of the six bond lengths at 2.6 shortens to 2.44 while another one in the diagonally opposite plane elongates to @xmath43 3.25 . at room temperature ,
oxygen ions in mn - o - mn chain along b - axis are off - axis resulting in mn - o - mn bond angle of @xmath44 174@xmath24 .
this mn - o - mn angle decreases in co state due to tilting of the octahedra and results in shortening of one of the pr / sm - o bonds and lengthening of the other pr / sm - o bond .
the reduced mn - o - mn angle also decreases the double exchange transport in the mn - o - mn chain rendering the sample insulating . in the co
phase the crystal structure is reported to belong to p2@xmath45/n space group@xcite .
the pr / sm - o and pr / sm - mn bond lengths generated from this space group can be grouped into 6 + 3 + 3 and 2 + 6 respectively .
it has been already mentioned that 6 + 3 + 3 distribution of sm - o bond lengths resulted in negative @xmath39 or a much higher r - factor of fitting . however , even with this distribution of oxygens around the rare - earth ion , pr / sm - mn coordination numbers had to be 4 + 4 and not 2 + 6 .
this change in the distribution of mn ions around the rare - earth can only be reconciled if one assumes a shift of the rare - earth or the mn ion along c - axis .
shifting of the rare - earth ion along c - axis decreases two pr / sm - mn bond lengths from 3.36 to 3.28 .
such a shift of cations has been reported in ce - type charge - ordered compounds @xcite . the cooperative rotation of mn - o octahedra observed here indirectly from the changes in pr / sm - o bond lengths also hint towards jt distortions .
jt distortion results in unequal distribution of mn - o bond lenghts in mn@xmath8o@xmath6 octahedra .
the rotation of mno@xmath6 octahedra along the b - axis , moves one of the oxygen ion closer to a particular mn site , simultaneously increasing its distance from another mn ion in the neighboring octahedra .
such type of unequal bond distribution in neighboring mno@xmath46 octahedra results in jt distortions and is known in ce type charge - ordered materials @xcite .
the overlap between o(2p@xmath47 ) and 4f orbitals of rare - earth cations is one of probable causes for inducing charge - ordered state in manganites @xcite . in the present case ,
shifting of re cations increases the overlap between 4f orbitals of rare - earth cation and o(2p@xmath47 ) orbitals .
such an overlap reduces itinerant electron density in the e@xmath15 band .
furthermore , tilting and rotation of the mno@xmath6 octahedra reduce the mn - o - mn angle , thereby decreasing the ferromagentic double exchange transfer in mn - o - mn chains . as a combined result , with 3d@xmath48 orbitals depleted of electron density and rotation of mno@xmath6 octahedra the ferromagnetic double exchange between mn@xmath8-o - mn@xmath9 zener pairs
is drastically reduced . moreover , unequal mn - o distance in o - mn - o chain strengthens the antiferromagnetic superexchange between the t@xmath49 electrons . hence , shifting of re ion and rotation together with tilting of mno@xmath6 octahedra drive the material into ce type afi state at low temperatures .
we have investigated the local structure of mno@xmath6 octahedra in sm@xmath13pr@xmath12sr@xmath2mno@xmath3 sample with the help of pr and sm k - edge exafs . in this case the transition to co ground state is strongly associated with local structural environment of a - site cation .
co state in sm@xmath13pr@xmath12sr@xmath2mno@xmath3 arises from the combined effect of rotation and tilting of mno@xmath6 octahedra and shifting of the re cations .
rotation and tilting of mno@xmath6 octahedra leads to reduced mn - o - mn angle and shifting of re cations increases the overlap between o@xmath50 and 4f orbitals of a - site cations thereby decreasing the carrier density in the e@xmath15 level .
this effect reduces ferromagnetic exchange between zener pairs and strengthens antiferromagnetic superexchange interactions .
trapping of e@xmath15 electrons which can be viewed as the jt polaronic effect that leads to the charge ordering state .
authors [ krp and prs ] would like to gratefully acknowledge jasri for local hospitality and beam time on bl01b1 and travel assistance from department of science and technology , government of india , new delhi .
krp thanks dst for financial support under dst fast track scheme for young scientists ( ps-19 ) .
zener c. 1951 phys . rev.*82 * 403 cheong s w and mostovoy m 2007 nat . mater .
* 6 * , 13 serrao c r , sundereshan a and rao c n r 2007 j. phys . : condens .
mater * 19 * 496217 van den brink j and khomskii d i 2008 j. phys . : condens .
mater ( in print ) , preprint cond - mat/0803.2964 efremov d v , van den brink j and khomskii d i 2004 nat .
mater . * 3 * , 853 damay f , maignan a , martin c , and raveau b 1997 j. appl .
phys.*81 * , 1372 gu k m , tang t , cao q q , chen y q , wang j h and zhang s y 2002 j. phys . :
mater * 14 * , 8853 martin c , maignan a , hervieu m and raveau b 1999 phys .
b * 60 * , 12191 knizek k , jirak z , pollert e , zounova f and vratislav s 1998 j. solid state chem . * 100 * , 292 tomioka y , asamitsu a , moritomo y , kuwahara h and tokura y 1995 phys .
lett . * 74 * , 5108 damay f , martin c , maignan a and raveau b 1997 j. appl . phys . *
82 * , 6181 arulraj a , santhosh p n , gopalan r s , guha a , raychaudhuri a k , kumar n and rao c n r 1998 j.phys .
: condens .
mater * 10 * , 8497 rao c n r , arulraj a , cheetham a k and raveau b 2000 j.phys . :
* 12 * , r83 tokura y , kuwahara h , moritomo y , tomioka y and asamitsu a 1996 phys .
lett . * 76 * , 3184 smolyaninova v n , lofland s e , hill c , budham r c , gonen z s , eichhorn b w and greene r l 2002 j. magn .
248 * , 348 fiebig m , lottermoser th , frohlich d , goltsev a v and pisarev r v 2002 nature 419 , 818 kimura t , goto t , shintani h , ishizaka k , arima t and tokura y 2003 nature 426 , 55 hur n , park s , sharma p a , ahn j s , guha s and cheong s w 2004 nature 429 , 392 rao c n r , arulraj a , santosh p n and cheetham a k 1998 chem . mater . * 10 * , 2714 mizokawa t , khomskii d i and sawatzky g a 2000 phys . rev .
* 024403 meneghini c , cimino r , pascarelli s , mobilio s , raghu c and sarma d d 1997 phys . rev .
b * 56 * , 3520 booth c h , bridges f , snyder g j and geballe t h 1996 phys . rev .
b * 54 * r15606 subias g , garcia j , blasco j and proietti m g 1998 phys . rev .
b * 57 * 748 shibata t , bunker b a and mitchell j f 2003 phys . rev .
b * 68 * , 024103 cao d , bridges f , anerson m , ramirez a p , olapinski m , subramanian m a , booth c h and kwei g h 2001 phys .
b * 64 * , 184409 priolkar k r , kulkarni v d , sarode p r , kumashiro r and emura s 2005 physica scripta * tt105 * , 442 newville m , livins p , yacobi y , rehr j j and stern e a 1993 phys . rev .
b * 47 * , 14126 kurbakov a i , lazuta a v , ryzhov v a , trounov v a , larionov i i , martin c , maignan a and hervieu m 2005 phys . rev .
b * 72 * 184432 radaelli p c , cox d e , marezio m , cheong s w 1996 phys .
b * 55 * , 3015 zabinski s i , rehr j j , ankudinov a , albers r c and eller m j 1995 phys .
b * 52 * , 2996 blasco j , garcia j , de teresa j m , ibarra m r , perez j , algarabel p a , marquina c and ritter c 1997 j. phys . : condens .
matter * 9 * , 10321 | sm@xmath0pr@xmath1sr@xmath2mno@xmath3 exhibits variety of ground states as @xmath4 is varied from 0 to 0.5 . at an intermediate doping of @xmath5
a charge - ordered ce type antiferromagnetic insulating ( afi ) ground state is seen .
the transition to this ground state is from a paramagnetic insulating ( pmi ) phase through a ferromagnetic metallic phase ( fmm ) .
local structures in pmi and afi phases of x = 0.3 sample have been investigated using pr k - edge and sm k - edge extended x - ray absorption fine structure ( exafs ) .
it can be seen that the tilting and rotation of the mno@xmath6 octahedra about the b - axis are responsible for the charge ordered ce - type antiferromagnetic ground state at low temperatures .
in addition a shift in the position of the rare earth ion along the c - axis has to be considered to account for observed distribution of bond distances around the rare earth ion . |
peculiar velocities are deviations from the uniform expansion of the universe . in the standard gravitational - instability paradigm , they are generated by inhomogeneities in the matter distribution .
measurements of peculiar velocities using galaxies rely on distance indicators to subtract the hubble expansion , which limits the volume probed to @xmath3mpc ( see strauss & willick 1995 ; kashlinsky , atrio - barandela & ebeling 2012 for reviews ) . as an alternative ,
kashlinsky & atrio - barandela ( 2000 , hereafter kab ) proposed a method to probe the large - scale velocity field via the dipole induced by anisotropies in the cmb temperature that are caused by the kinematic sunyaev - zeldovich ( sz ; sunyaev & zeldovich 1970 , 1972 ) effect in the direction of clusters of galaxies .
the sz has two components : a thermal one ( tsz ) , caused by the thermal motions of electrons in the potential wells of clusters , and a kinematic one ( ksz ) , caused by the motion of the cluster as a whole ( kinematic sz , ksz ; see e.g. the review by birkinshaw 1999 ) .
once it is imprinted on the cmb , the sz distortion remains fixed .
it does not depend on distance , making it a very useful tool to detect clusters at high redshift ( planck collaboration , planck early results xxvi , 2011 ) .
the ksz effect measures motion directly with respect to the cmb frame and does not require subtraction of the hubble expansion .
the kab method evaluates the dipole from an all - sky cmb temperature map at the locations of known x - ray luminous clusters .
since the main contaminant in foreground clean maps is the primary cmb signal , the data need to be filtered to allow the peculiar velocities of clusters to be measured ( haehnelt & tegmark 1996 ) ; kab proposed to use a wiener - type filter for this purpose . in kashlinsky
et al ( 2008 , hereafter kabke , 2009 ) we showed that , within cluster apertures containing zero cmb monopole , a residual cmb dipole exists which is aligned with the cmb dipole .
the amplitude of this residual dipole scales with cluster x - ray luminosity ( kashlinsky et al 2010 ; kashlinsky , atrio - barandela & ebeling 2011 ) proving that the signal is indeed associated with clusters .
we have interpreted this dipole as the signature of a bulk flow .
if the flow extends to the cosmological horizon , part of the all - sky cmb dipole could be viewed as being of primordial , rather than kinematic , origin . as of this writing ,
no alternative interpretations of the signal , which would have been of considerable interest , have been offered in peer - reviewed literature .
our results have been challenged by other authors .
keisler ( 2009 ) , as well as the contemporaneous study by kashlinsky et al ( 2010 ) , noted that the residuals for each of the maps for the eight wmap das ( differential assemblies ) in the kabke measurement are correlated , which increases the errors quoted by kabke , but by no more than @xmath4 , where @xmath5 is the number of das of the wmap dataset .
keisler then claimed that the detection of a large - scale flow was in fact not statistically significant , his error corrections exceeding the @xmath6 factor .
atrio - barandela et al ( 2010 ) pointed out that keisler ( 2009 ) had failed to subtract the dipole of the filtered maps outside the galactic mask , thereby overestimating his error bars .
in addition , we also proved that the filter devised by kab removes the primary cmb down to the fundamental limit of cosmic variance ( atrio - barandela et al 2010 ) .
it was this highly efficient filtering scheme that allowed a detection of the flow at 3.54@xmath7 confidence level , depending on the cluster sample used .
all data needed to verify the measurement have been made publicly available for examination by the interested reader .
osborne et al ( 2011 , omcp ) studied the effect of different filters on bulk - flow measurements from wmap data . in their appendix b.2 omcp reproduced our results at the @xmath8 confidence level but failed to notice that this in itself invalidates the statistical uncertainties claimed by keisler ( 2009 ) .
the lowered significance ( compared to our 3.54 @xmath7 ) , is due to differences in the catalogs used ( kashlinsky et al 2011 ) and to the relatively small number of simulations ( @xmath9 ) carried out by omcp to compute the errors .
omcp also implemented alternative , `` matched '' filters , with which they found no evidence of bulk flows in wmap data .
those matched filters , however , were designed more to detect radio point sources than to remove the primary cmb anisotropies . as a result
, they were insensitive to bulk motions to the extent that the ksz signal is weaker in the filtered maps than in the original data .
as can be seen in omcp s fig . 13 ,
a sample of @xmath10 clusters moving at velocities of @xmath11 km s@xmath1 is required to produce a detectable signal in the filtered maps . at such extreme velocities ,
the motion of individual clusters should be detectable directly in the original cmb maps without any need for filtering . in another study , mody & hajian ( 2012 , mh ) , using a different wiener filter , concluded that the measured bulk - flow amplitude was consistent with the @xmath12cdm prediction " .
this statement ( made prominently in the mh abstract ) is largely meaningless though , since the mh error bars of 1,4002,600 km s@xmath1 per velocity component render their methodology unable to discriminate between the concordance @xmath12cdm prediction and a bulk flow of amplitude 6001,000 km s@xmath1 , as measured by kabke . in spite of the aforementioned problems with these studies , both the omcp and mh
results have been cited repeatedly as evidence against the existence of a dark flow ( df ) . in this paper , we examine in more detail why the filters used by omcp and mh failed to detect a statistically significant dipole signal .
we stress that our focus is on the dipole signal , _ irrespective of its interpretation_. specifically , we aim to answer the question of why the dipole signal was not detected with the other filtering schemes .
do they reduce the s / n to levels that render the dipole statistically insignificant ?
do they redistribute the dipole signal to lie outside the aperture ranges probed there ? equally important , none of the proposed new filters have been demonstrated to probe the same flow than the kab filter , so the discrepancies between filters could be due to variations of the flow . in this article
we will show that the omcp and mh filters are not as efficient as kab , but that , when correctly implemented , either filter recovers a signal that is in fact compatible with the df identified by us . in sec .
[ sec : filters ] we discuss the properties of filters , their effect on the data , and summarize previous results . in sec .
[ sec : sec3 ] we compute the cmb dipole at the locations of clusters in maps processed with the different filters and discuss their relative merits and their consistency , in sec .
[ sec : summary ] we summarize our results and present our conclusions .
at the positions of clusters , the microwave temperature anisotropies have four main components : the primary cmb , the tsz and ksz signal , and instrument noise . in foreground
clean maps one needs to expect some level of foreground residuals , but these residuals will not correlate with clusters in the sky and outside the galactic plane are negligible compared with the cosmological signal . the ksz anisotropy is given by @xmath13 , where @xmath14 is the projected electron density along the line of sight , @xmath15 the bulk - flow velocity of our cluster sample , @xmath16 the speed of light , and @xmath17 the cmb blackbody temperature .
the bulk flow motion of all clusters in a given sample will generate a cmb dipole at the cluster location that we estimated it to be ( kab ) @xmath18 where the terms on the right - hand side describe the contributions from the ksz , the cosmological cmb , noise , and the tsz , respectively . in this expression , @xmath19 clusters that occupy a solid angle of @xmath20 pixels in the sky .
instrument noise ( the third term ) can be further reduced as @xmath21 by combining @xmath22 differential assemblies operating at different frequency bands .
in addition , the fourth term , reflecting the contribution from the tsz effect , can be isolated and eliminated by using its distinctive frequency dependence . nevertheless , the tsz dipole due to the inhomogeneous distribution of clusters in the sky is bounded above by the tsz monopole . in kashlinsky
et al ( 2008 , 2009 ) we proposed to measure dipoles at zero monopole aperture to limit the tsz dipole contribution even without frequency information . the largest source of errors in eq .
( [ eq : dipole ] ) is the sample - variance uncertainty caused by the small number of pixels at which the dipole is evaluated . to determine peculiar velocities
it is vital to increase the signal - to - noise ratio ( s / n ) for these pixels by reducing the contribution from the primary cmb anisotropy .
however , unlike for the tsz component , frequency information can not be used to discriminate between the cmb and ksz contributions to the overall dipole . to address this problem , kab proposed to use the statistical properties of the primary cmb to design a filter that effectively removes its contribution while preserving the ksz component , thereby increasing the s / n of the proposed measurement . reflecting the spatial isotropy of the cmb temperature anisotropies , filters are spherically symmetric ;
in @xmath23-space , the filter depends on @xmath23 but not on @xmath24 .
if @xmath25 and @xmath26 are the expansions of the temperature map in spherical harmonics and of the filter in legendre polynomials , respectively , then the filtered map is given by @xmath27 where @xmath28 denotes convolution .
in the following sections we will discuss the distribution of a ksz dipole in @xmath23-space , followed by a brief description of the filters under investigation here .
our bulk - flow studies were based on an all - sky cluster sample created by combining the rosat - eso flux limited x - ray catalog ( reflex ) ( bhringer et al .
2004 ) in the southern hemisphere , the extended brightest cluster sample ( ebcs ) ( ebeling et al .
1998 , 2000 ) in the north , and the clusters in the zone of avoidance ( ciza ) ( ebeling et al . 2002
; kocevski et al .
2007 ) sample along the galactic plane .
our full catalog contains @xmath29 entries including clusters and galaxy groups . to illustrate the performance of the different filters , we select a subset of 506 clusters located outside wmap s kq75 galactic and point source mask .
these clusters are located at redshifts @xmath30 and have x - ray luminosities @xmath31\ge 0.5\times 10^{44}erg\,s^{-1}$ ] . for the purposes of this paper
we restrict our analysis to clusters for which the parameters of a @xmath32 profile ( cavaliere & fusco - femiano 1976 ) have been measured by fitting to rosat x - ray data . this sample is smaller than the one used in kashlinsky et al ( 2010 ) and our final statistical significance
will be degraded with respect to our earlier results .
we divided our set into three independent samples , selected according to luminosity .
we define three intervals : @xmath33,[1.,2],[>2])\times 10^{44}$ ] erg s@xmath1 . the distribution of our clusters on the sky is shown in fig .
[ fig : sample ] using a healpix projection of @xmath34 ( gorski et al . ,
all clusters are drawn as discs of radius @xmath35 .
the set of clusters in the final sample occupies less than 1% of the sky . in such a small solid angle , spherical harmonics are not orthogonal . before discussing how filters operate , we need to discuss how the dipole power at @xmath36 , that is evaluated only at the cluster positions , is redistributed among all multipoles .
if @xmath37 are the spherical harmonic transforms of the ksz component of the ksz dipole evaluated at @xmath19 locations in the sky then the power at each multipole @xmath23 is @xmath38 .
we define @xmath39 as the contribution of all multipoles @xmath40 to the total dipole in the map as : @xmath41 to predict @xmath42 and @xmath39 we need the electron - density profile of all clusters .
suitable data to derive these profiles accurately are not yet available . to compute the dipole power
we will assume that all clusters subtend the same solid angle and move coherently in the direction @xmath43 .
for simplicity we assign a dipole anisotropy @xmath44k@xmath45 to all pixels of each cluster , where @xmath46 is the angle between the line of sight of the cluster and the apex of the motion . figs .
[ fig : signal]a and [ fig : signal]b depict the resulting power and the integrated dipole signal @xmath39 , respectively , for identical clusters of angular radius 10 , 15 and 30 . fig .
[ fig : signal]a shows that the lowest multipoles have the largest amplitude , and that power is transferred preferentially to odd multipoles .
[ fig : signal]b shows that less than 10% of the signal remains at @xmath36 ; the bulk of the signal is contained in the range @xmath47 even for unresolved clusters .
the exact distribution of the ksz dipole in @xmath23-space will depend on the ( unknown ) profile of clusters and their extent . since clusters have a radius of 10 - 30 , the effect of the cluster profile will be important for multipoles @xmath48 .
therefore , our conclusion that more than 50% of the ksz dipole signal lies below those scales is not affected by taking a constant profile .
the actual signal will be a combination of the contributions from resolved ( 1530 ) and unresolved ( 10 ) clusters , with clusters of different mass contributing at different relative weights .
the kab filter was specifically designed to remove the primary cmb of the measured cosmological model .
the filter is of the wiener type and was constructed to minimize the contribution from the cosmological signal in the filtered maps in the presence of noise , @xmath49 , i.e. it minimizes @xmath50 ( kashlinsky et al 2009 ) . in @xmath23-space the filter is given by @xmath51 where @xmath52 is the actual realization of the radiation power spectrum in the sky that includes noise , tsz , ksz , foreground residuals , and primary cmb ; @xmath53 is the power spectrum of the @xmath12cdm model that best fits the data , and @xmath54 is the antenna beam for a given da .
since the quadrupole and octupole are aligned with the dipole , the kab filter is set to zero for @xmath55 to avoid any cross - talk from the corresponding large angular scales that could mimic a dipole . in fig .
[ fig : filters]a we represent the kab filter for the w1 da .
the filter has structure at all multipoles and is therefore sensitive to ksz power across the full multipole range .
omcp used matched filters ( mf ) , i.e. , filters that are suited to distinguish point sources from a background signal , provided the profile of the source is known . in their analysis , omcp assumed that all clusters are unresolved and approximated their profiles by the wmap beam function .
in @xmath23-space this means @xmath56 where @xmath57 and @xmath58 are the power spectrum of the tsz effect and of the white and homogeneous instrumental noise , respectively .
@xmath54 is the antenna beam .
in addition , omcp considered clusters to be isothermal , in which case the tsz and ksz profiles are proportional to each other . by measuring the tsz component in the original and in the filtered data
, the fraction of the ksz lost due to filtering could be computed exactly . however , since clusters are _ not _ isothermal ( as demonstrated , e.g. , by atrio - barandela et al .
2008 ) , their electron - density and pressure profiles are in fact different , causing filters that disregard this fact to potentially suppress the ksz signal relative to the tsz contribution . in fig .
[ fig : filters]b we plot the omcp filters for eight wmap das , corresponding to the q ( dot - dashed green line ) , v ( dashed blue ) and w ( solid red ) bands .
these filters are effectively band - pass filters , removing all power below @xmath59 and above @xmath60 .
omcp also considered a modified mf , named unbiased multifrequency matched filter ( uf ) , designed to remove the tsz contribution .
since this second filter was found to be even less sensitive to bulk flows than mf , we will not consider it further .
our approach to reducing any tsz contamination will be the same for all filters , namely to compute dipoles at the zero - monopole aperture .
mh used a wiener filter designed to remove the tsz and instrument noise contributions together with the intrinsic cmb , leaving only the ksz component in the data .
their filter is defined by @xmath61b_\ell^2+n_\ell}\ , , \label{eq : mh}\ ] ] where @xmath62 is the power spectrum of the ksz contribution , and the other terms have the same meaning as in eqs .
( [ eq : kab ] , [ eq : omcp ] ) .
this filter neglects the inhomogeneities in the instrumental noise and assumes that @xmath62 , the quantity to be measured , is known . to construct their filter
, mh need to 1 ) specify the cluster distribution on the sky , 2 ) assume an electron density profile for all clusters , and 3 ) fix the cluster extent , all quantities that have yet to be measured from the data . in fig .
[ fig : filters]c we plot the mh filter for assumed cluster extents of 10 and 15 ; both are convolved with the beam of the w1 da .
the filter is normalized to a ksz dipole in the z - direction with an amplitude of @xmath63k ; convolution with the beam reduces the amplitude to @xmath64k .
the resulting two curves can be thought of as two extreme cases of what the actual mh filter might be .
taking into account that very few clusters are resolved in the wmap w band , we will in the subsequent discussion only consider the mh filter constructed assuming an angular radius of 10 .
filtering not only removes the intrinsic cmb component , it also modifies the ksz signal , affecting the cluster extent and its electron density profile . to evaluate the monopole and dipole at @xmath65 cluster positions on the sky we first need to discuss what would be the radial extent of each cluster in the filtered data . in kashlinsky
et al ( 2008 ) we considered that clusters extended 46 times the radial extent of the region emitting 99% the total x - ray flux . since kashlinsky et al . ( 2010 ) we quote instead the dipole evaluated on discs of @xmath66 aperture radius , the same size for all clusters . in both cases , the monopole was close to zero and the measured dipole did not differ significantly . _ evaluating dipoles at
the zero monopole aperture is crucial to guarantee that the measured dipole has no significant contribution from the tsz effect _ :
since the monopole is dominated by the tsz contribution , the amplitude of the former is an upper limit to any dipole due to the inhomogeneous distribution of clusters on the sky .
the cluster profile , @xmath67 , when convolved with the antenna beam @xmath54 and the filter @xmath68 , becomes : @xmath69 where @xmath70 is a normalization constant . in figs .
[ fig : filters]d f we plot the filter profile in real space of the w - band ( solid red lines )
. this would be the cluster profile in the filtered map if it were a dirac @xmath71-function , i.e. , if the cluster was unresolved . for comparison , the dashed line represents gaussian beam with the resolution of the w band . to assess the impact of the aperture choice , we compute the measured cluster profile from eq .
( [ eq : profile ] ) , averaged over a disc of angular radius @xmath72 , as it was done in the data @xmath73 d\cos\alpha\ , .
\label{eq : filter_average}\ ] ] in fig .
[ fig : filters]g h we plot the above average vs the angular radius @xmath72 of the disc for the kab , omcp , and mh filters , respectively . to avoid overcrowding the plots ,
we only show the q1 , v1 and w1 das . for the kab filter
the signal vanishes around 20 , while it extends to 30 for the omcp filters , and well beyond 40 for the mh filter .
these radial extents correspond to unresolved clusters ; for real sources the signal will spread to apertures of even _
larger _ radii .
empirically , we found that the zero monopole aperture was around 25 for the kab filter , whereas slightly above 20 would correspond to all clusters being unresolved by the antenna beam .
the zero monopole defines a physically well motivated aperture over which to compute the dipole .
_ this point was missed by omcp and mh , _ who never have addressed this issue .
the filters discussed in this section are spherically symmetric .
they can not imprint a dipole at zero monopole exclusively at cluster locations on a filtered map but , if not carefully designed and implemented , they can erase it to below the statistically significant level .
omcp implemented two matched filters , named mf , described in sec .
[ subsec : omcp ] , and uf . they measured the dipole at the position of 736 clusters of galaxies and found no significant ksz signal .
omcp tested the performance of their filters using simulated maps and showed that their filters begin to recover the input velocity only at speeds @xmath74 km s@xmath1 for the mf and @xmath75 km s@xmath1 for the uf . the average mean optical depth of the clusters in their sample is quoted as @xmath76 ( sec . 4.5 of omcp ) , and the resulting temperature anisotropy would be @xmath77 . velocities approaching or exceeding 4,000 km s@xmath1 thus induce anisotropies that would be measurable in the original data and for each individual cluster , without any need for filtering .
the root cause of this problem is that omcp s implementation of their filters did not increase the s / n but strongly reduced it , to the extent of rendering their filters unable to detect a cmb dipole corresponding to a bulk - flow velocity of @xmath78 km s@xmath1 .
the poor performance of their filters is also due to the fact that omcp effectively measured the dipole only at the central cluster pixels because their filters were optimized to reconstruct the source amplitude if the source was centered on that pixel ( mak et al 2011 ) . since clusters in the filtered maps reach a zero monopole aperture at @xmath7930 - 40(see fig .
[ fig : filters]d i ) the implementation by these authors does not use all the information available in the filtered map . also , omcp assumed that the ksz and tsz cluster profiles are identical and unresolved .
however , any filter that efficiently removes the tsz signal will also remove a substantial part of the ksz signal .
this is the reason why the better designed uf requires three times larger velocities than the mf to detect the average motion of @xmath80 clusters .
mh , by contrast , used a different approach . instead of computing the dipole at the locations of clusters , mh fit dipole templates to their filtered maps .
this approach is flawed in two important respects : ( a ) mh require that the total number of clusters , their profiles , and extents be known in advance in order to construct their filter , and ( b ) by matching the same cluster template to the filtered data mh implicitly assume that all clusters have the same profile and extent in the sky as in the filtered cmb data .
hence , changing the profile , number , or extent of clusters changes the filter ( and the filtered map ) and the final results . as both omcp and mh restricted their analyses to apertures smaller than the ones used by kabke
, the dipole measured by us could have remained undetected in the residual cmb and noise of their filtered maps .
to measure the dipole at the locations of clusters in cmb maps filtered following kab , omcp , and mh , our pipeline proceeds as follows : 1 . for each da ,
the data are multiplied by the extended mask kq75 for the wmap7 data release .
then , monopole , dipole , and quadrupole are subtracted from the regions outside the mask .
next , we compute the multipole expansion coefficients @xmath81 .
the power lost due to masking is corrected for by multiplying each multipole by @xmath82 , where @xmath83 is the fraction of the sky outside the mask .
the @xmath81 coefficients are multiplied by the filter @xmath68 before transforming them back into real space to create the filtered map .
the monopole and dipole outside the mask are removed from the filtered maps .
dipoles are computed at the cluster positions using the same aperture of radius @xmath72 for all clusters .
we repeat the measurement for different @xmath72 and , as indicated in sec .
2.5 , we select the dipole measured at the _ zero monopole _ aperture . to translate the measured dipole amplitude from temperature units to velocity units , we need to compute how the mean cluster opacity changes with filtering .
owing to the different functional form of the three filters , they will affect the ksz power differently .
since @xmath84 , to compute the average @xmath85 over our cluster sample we need the electron density profile and of the cluster radial extent of all clusters , information that is not presently available . in the kab filter
, this uncertainty translates into a calibration uncertainty but does not affect the amplitude of the dipole in @xmath86k nor its direction . as discussed , this uncertainty is built into the definition of the omcp and mh filters . in order to facilitate the comparison of the results obtained with the different filtering schemes , we will use the results presented in figs .
[ fig : filters]d f to normalize the filtered maps as described below . the filters defined in sec . 2 and depicted in fig .
[ fig : filters]a - c do not only differ in shape but also in normalization . to normalize a map we need to derive a coefficient @xmath87 that measures how much the dipole is diluted by the filter .
if @xmath88 and @xmath89 are the cluster optical depth in the unfiltered and filtered maps , then the normalization coefficient would be @xmath90 , where the average is taken over the solid angle subtended by the clusters in unfiltered and filtered data .
this parameter is not simple to compute since both profiles and angular extents are unknown .
instead , an approximated normalization coefficient can be computed by taking into account that relaxed clusters are close to isothermal in their central regions ( pratt et al .
2011 ) and electron pressure and electron density are proportional to each other . hence , in those regions _ filtering would dilute the tsz and ksz signal in the same proportion .
_ then , @xmath91 where the average is on a small disc of radius @xmath72 around the cluster center in both the unfiltered and filtered maps . in table
[ table1 ] we list the monopole as measured within an aperture of 5 radius in the original wmap data and in the kab , omcp , and mh filtered maps .
the quoted values correspond to the mean and rms dispersion of the four w das .
we divide our sample according to x - ray luminosity in three independent bins to test the correlation of our measurements with x - ray luminosity as in kashlinsky et al ( 2010 ) . within 5
the measured @xmath87 is rather uncertain since the temperature anisotropy contains cmb residuals and noise together with tsz .
this is specially evident for the less massive clusters ( sample i ) . from table
[ table1 ] , the renormalization coefficient is @xmath92 for the kab filter , @xmath93 for the mh filter whereas the omcp filter does not dilute but boosts the signal ( but not the s / n ) by about a factor of 3 .
since all clusters in our sample have been fit to a @xmath32-model , we can use this profile to compute numerically @xmath87 .
it should be computed separately for each cluster sample and da . nevertheless , _ to avoid introducing an extra variance , we will use the same values of @xmath87 for all cluster configurations and bands .
_ in this way , we do not change the s / n ratio of the measured dipoles when expressed in @xmath86k . for the full sample
the ratios are @xmath94 $ ] for the kab , omcp and mh filters , respectively .
since different bands have different normalization coefficients and different zero monopole aperture , we concentrate our study on the 4 das of the w - band .
although the results averaged over all wmap bands are very similar due to the noise , by using only the w - band , the statistical significance of the measured dipole is slightly decreased . for the purpose of this paper
this is not important since we are only interested in comparing the s / n ratio of the three filters .
.different cluster samples considered in this work and some of their properties .
@xmath65 is the number of clusters in each subsample and x - ray luminosities are given in units of @xmath95erg s@xmath1 . [
cols="<,<,^,^,^,^,^,^,^,^ " , ] to compare the performance of the three filters , we computed dipoles for each filter and for the three cluster samples of table [ table1 ] .
we used the following apertures of radius @xmath96 $ ] arcmin in order to select a suitable zero - monopole aperture for each filter .
as indicated , all data are renormalized by multiplying the filtered kab , omcp and mh maps by the coefficients @xmath97 $ ] , respectively . with this normalization ,
the monopole evaluated on apertures of 5 would have the same amplitude in the three filtered maps .
ideally , this would also be true for the dipole . by using a single coefficient , the
s / n of the dipole measurement in each cluster configuration remains unaltered . in table
[ table1 ] we also list the value of the monopole residual at 30 in the renormalized maps .
the kab zero monopole aperture occurs at @xmath98 radius , for omcp is at @xmath7930 - 40 and for mh is at @xmath9950 ; table [ table1 ] shows that at 30 the mh filter still contains non - zero monopole residuals , in agreement with fig .
[ fig : fig6 ] . in figs .
[ fig : dipole]a - c we plot the three components ( x , y , z ) of the dipoles measured with an aperture of 30versus the tsz monopole at 10 radius . in fig . [
fig : dipole]d we plot the dipole moduli .
the three cluster subsamples of table [ table1 ] are represented with the same symbol .
they can be easily distinguished since the average tsz monopole increases with increasing x - ray luminosity .
notice that each cluster subsample has similar monopoles at 10 for the three filters as expected since our normalization was constructed to produce the same tsz amplitude in the central parts of clusters at 5 .
the errors on the measured dipoles @xmath100 , @xmath101 , were obtained from the dispersion of dipoles measured in random cluster positions in the sky . for each component ,
the error scales approximately as ( atrio - barandela et al 2010 ) @xmath102 where @xmath103 is the rms dispersion of the cmb residual in the filtered map , @xmath104 the dispersion of the noise , @xmath20 the number of pixels occupied by the cluster sample and @xmath22 the number of das used in the analysis . we considered four cluster samples with different numbers of clusters ( 100 , 200 , 400 and 800 clusters ) . for each sample
, we generated 4,000 random cluster distributions .
if clusters sampled the sky homogeneously then the error on the monopole , @xmath105 , and in each dipole component would be related by @xmath106 . for an inhomogeneous cluster distribution
the relation is given by @xmath107 where @xmath108 is the average of the direction cosines of the clusters in the sample ( atrio - barandela et al .
2010 , atrio - barandela 2013 ) .
since the galaxy removes a large fraction of the sky , preferentially in the x direction , errors are larger for @xmath109 than for @xmath110 or @xmath111 .
we verified that this was the case for the three filters , being 10 - 18% and 2 - 3% larger with respect to a homogeneous distribution of clusters for x and y , respectively , and a 15% smaller for z. since the errors are dominated by the residual cmb , they scale as @xmath112 , as dictated by eq .
( [ eq : sigmas ] ) . for each cluster sample ,
the error on each dipole component is scaled by the number of clusters from the error obtained on the simulations .
the errors on the modulus of fig .
[ fig : dipole]d were obtained from the rms dispersion of 10,000 dipoles .
those dipoles were generated by adding to the each measured dipole component a random value drawn from a gaussian distribution with zero mean and dispersion @xmath100 .
the results from the three filters show a reassuring degree of consistency regarding the value of the y - component of the dipole .
this component is always negative for all filters and all cluster configurations , in agreement with the df results , which combine a pronounced negative y - component with a positive z - component of lower amplitude and an x - component that is consistent with zero within the measurement uncertainties .
the kab filter ( blue filled circles ) shows a very clear correlation of the y dipole component with the monopole at 10 ( fig . [
fig : dipole]b ) and so does the overall dipole amplitude ( fig .
[ fig : dipole]d ) .
this correlation is found at 30 , when the tsz monopole is zero .
then , there can not be a tsz dipole contribution due to an inhomogeneous cluster distribution .
it is instructive to revisit fig .
[ fig : fig6]a : cluster by cluster , the average temperature anisotropy on a disc of 10 is uncorrelated with the same average over a disc of 30 .
[ fig : dipole]a also shows a strong correlation for x - component in the mh filter ( green triangles ) .
however , in this case we expect this correlation to be due to the tsz effect . as our error analysis show
, the x - component of the dipole is the worst measured since the galaxy is most predominant in this direction .
then , one has to expect an important tsz dipole due to the inhomogeneous distribution of clusters in the sky in agreement with the fact that for the brightest clusters , sample iii , the residual tsz monopole at 30 is @xmath113k , large in comparison with the value of the other two filters ( see table [ table1 ] ) .
it is also in agreement with the result of fig .
[ fig : fig6]c that shows that the average temperature anisotropy at cluster locations within discs of radius 30 ( tsz monopole ) is strongly correlated with the same average at 10 and , consequently , so it would be the tsz dipole at 30 .
we verified that the measured dipoles are in fact very small at the 50 aperture , the zero monopole aperture for this filter .
with respect to the omcp filter ( open red squares ) the dipole is always small and shows no correlation with the monopole at 10 .
the low dipole amplitudes were to be expected since this filter cuts off all the power at multipoles @xmath114 ( compare fig .
[ fig : signal]b and fig .
[ fig : filters]b ) and therefore is insensitive to half the dipole signal . at this filter zero monopole aperture ,
@xmath115 the results were not very different , showing no significant dipole either . since the three cluster samples of table [ table1 ] are independent the motion of all the clusters can be obtained by averaging the measured dipoles @xmath116 in each sample @xmath117 weighted by their respective errors @xmath118 @xmath119 in table [ table2 ] we give the resulting dipoles for the three filters .
we also give @xmath105 , the rms dispersion of the monopole .
consistently , the error bars on the dipoles are @xmath120 , being larger for the x and smaller for the z components , as we found with the simulated dipoles ( see sec [ sec : dipoles ] ) . in the table we also give the dipole amplitude and direction for the three filters .
the errors were computed from the dispersion of 10,000 dipole moduli and directions as in fig [ fig : dipole]d .
notice the remarkable agreement between the direction of the three filters . the angular separation between the central value of the direction of the kab filter and that of omcp and mh is about 16 .
more remarkable is the close alignment between the direction of the omcp and mh dipoles .
however , the errors on the direction are large and we can not derive a final conclusion of this coincidence since the mh dipoles have probably an important contamination from the tsz monopole , as discussed in sec [ sec : dipoles ]
. the results of table [ table2 ] are crucial to understand the performance of each filter .
of the three filters , the kab filter has the largest s / n . by effectively removing all multipoles with @xmath114 , omcp
filtered maps contain a small fraction of cmb residuals and give the smallest error bars .
unfortunately , the filter is also very effective in removing the dipole signal , degrading the overall s / n .
thus , our results agree with the original omcp analysis since velocities above 4,000km / s were required on samples of @xmath80 clusters for the dipole to be detectable . by comparison ,
the mh filter measures dipoles of larger amplitude but the error bars are also large .
only the kab filter provides statistically significant result . for this cluster sample
our measurement is significant at the 93% confidence level , smaller than in kashlinsky et al ( 2010 ) ( see atrio - barandela et al , 2013 , for a full discussion ) while the results of the omcp and mh have a significance below 60% , reflecting the poor performance of these two filters . to compare properly the results of the different filters ,
dipoles need to be expressed in velocity and not in temperature units . in kashlinsky
et al ( 2009 ) we denoted the calibration coefficient as : @xmath121 , with @xmath122 the dipole in @xmath86k and @xmath123 the _ filtered _ cluster profile averaged solid angle subtended by the cluster population . for the correct profile
, @xmath124 would be independent of the aperture used to measure the dipole . to account for the dilution effect introduced by the filter
our filtered maps have been renormalized by @xmath87 ( see sec 3.1 ) . in the renormalized maps , @xmath125 . using the @xmath32 model data
, we can compute @xmath85 for our cluster sample for different apertures .
we find that for @xmath126 , the coefficient averaged over the four w das is @xmath127k@xmath128 , respectively . since fig .
[ fig : fig5 ] shows that at the 10 aperture the three renormalized filtered maps have similar monopoles at the cluster locations , we can assume that @xmath85 is also similar for the three filters out to that aperture .
then , by taking the conversion factor for 10 we can translate the dipole into a velocity @xmath129 with this normalization , the dipole amplitude in km s@xmath1 is lower than the value of @xmath130 km s@xmath1 found previously ( see kashlinsky et al 2009 , 2011 for a full discussion on our previous method ) . nevertheless , our calibration of the flow is still not exact since the @xmath32 model is not an accurate description of cluster profiles outside the x - ray emitting inner part .
the dipole is measured at 30 while @xmath131 is computed at 10 . if @xmath131 is smaller at 30 , as suggested by the calculation at 15 , then the corresponding velocities will be larger .
for instance , if @xmath132k/@xmath133 $ ] then the velocities will be those of our previous work . even if our calibration of the flow is still uncertain due to lack of proper data , it does not affect the statistical significance of the dipole measured with any of the three filters .
the kab filter does not only measures a dipole that correlates with cluster x - ray luminosity ; it provides the largest signal to noise ratio of the three filters discussed here .
nevertheless , the discrepancies between the three filters could be real if the three filters are probing the flow at different scales .
for instance , since the omcp filter is sensitive primarily to scales corresponding to @xmath134 , it gives all the weight to unresolved clusters which are , on average , farther away .
by contrast , the kab dipole includes contributions from nearby systems , owing to the filter s greater sensitivity to low multipoles . while some degree of discrepancy between filters could simply reflect spatial variations in the properties of the flow ,
the fact is that none of the filters but kab has been shown to be limited by cosmic variance , has provided a statistically significant measurement or has shown a correlation with x - ray luminosity at the zero monopole aperture .
therefore , the results obtained using the omcp or mh filters can not be interpreted as ruling out any possible large scale flow detected by the kab filter .
we can now proceed with the summary of the appropriateness of the three filtering schemes in uncovering the ksz - like dipole signal in the kab method . *
* kab filter*. this filter is the most appropriate of the three as has already been explained in the original kab paper .
it removes the primary cmb down to the fundamental limit imposed by the cosmic variance as demonstrated analytically and numerically in atrio - barandela et al ( 2010 ) .
it has the smallest zero monopole aperture of the three filters and , as fig .
[ fig : fig6 ] shows , at the final - zero monopole - aperture the cmb does not correlate in any meaningful manner with that at the central parts .
this therefore allows to look for the ksz dipole component after the tsz has been made vanishingly small because of the cluster x - ray temperature decrease toward the outer parts ( kashlinsky et al 2009 ) .
the dipole uncovered at cluster positions at such aperture contains a negligible tsz component , which dominates the central parts at @xmath135 .
the filter is sensitive to a ksz dipole spread over the whole @xmath23 space and shows the largest s / n of all the filters anlyzed here . * * omcp filter * is clearly not as efficient as the kab . by cutting power at @xmath114
it reduces the residual cmb that dominates the error bar but also decreases the signal , reducing the overall s / n of the filter .
this point was already demonstrated implicitly in fig .
13 of omcp which shows that no bulk flow can be reliably measured with this filter at velocities @xmath136 km / sec ; we find it strange that , given those numbers , claims are occasionally still made about the alleged efficiency of this filter .
[ fig : fig6 ] and table [ table2 ] demonstrate explicitly that if one uses this filter , particular care must be made to work at the appropriately large apertures of at least @xmath137 in radius .
nevertheless , as the filter is only sensitive to a rather small range of multipoles ( @xmath134 ) , even at these large apertures the signal has been erased as to make the dipole undetectable .
* * mh filter*. this filter is the least useful for the application to the kab method . even at apertures
as large as @xmath137 it still leaves primary cmb and tsz components , as demonstrated by the clear correlation in fig .
[ fig : fig6]c , that confuse any determination of the ksz - like dipole signal .
furthermore , the filter does detect a dipole at the 30 aperture , but due to the large monopole residuals , there exists important contributions due to tsz dipoles .
the measured dipoles have the largest error bars and , like in the omcp filter , the results are not statistically significant . to conclude ,
we have compared the dipoles measured at the locations of x - ray selected galaxy clusters using the filters proposed by omcp and mh with the results obtained with the kab filter .
_ we find that all filters recover dipoles ; the central values of the directions are aligned but the kab dipole is the most efficient since it has the largest statistical significance . _
the discrepancy between these findings of ours and the results published by omcp and mh is due to the latter teams failing to chose the right aperture for their measurement .
in addition , their simulations and calibrations assume the same profile for the electron density and electron pressure , hypothesis that is bound to degrade their filter performances if real clusters are not described by their profiles . in comparison with the other filters , the kab filter makes no assumptions about cluster profiles or extents , and by design removes the intrinsic cmb , the largest contaminant ( see eq .
[ eq : dipole ] ) down to the limits imposed by cosmic variance .
the impact of all uncertainties associated with our limited knowledge of the cluster profile and extent is limited to a single normalization coefficient @xmath131 that does not affect the filter definition .
in addition , the zero - monopole aperture for the kab filter has a radius of @xmath138 , smaller than what is required by the other two filters . finally , the dipole signal recovered with the kab filter is strongly correlated with the central tsz monopole ( or the equivalent x - ray luminosity , kashlinsky et al 2010 ) , a clear indication that the dipole is not produced by some random contribution of the residual cmb .
these three aspects make the kab filter the least affected by possible systematics among the filters suggested to date .
alternative filtering schemes can possibly be designed , but their design must ensure that they increase the s / n of the measurement , not decrease it . finally , filters redistribute the information from real to filtered space by weighing angular scales differently .
therefore , they are probing the velocity field on different scales and the discrepancies on modulus and direction could reflect the spatial variations of the flow . to demonstrate that filters probe the same or different flows would require accurate cluster profiles and it goes beyond the scope of the present work .
recently , and well after this paper had been submitted to peer - review , the planck collaboration carried out the analysis of the filters described here using the recently released planck data ( planck intermediate result xiii , planck collaboration , 2013 ) .
the ksz team analyzed peculiar velocities using the internal linear combination ( ilc ) map , constructed by combining data at different frequencies with varying weights in @xmath23-space in order to reduce the foreground contribution .
the results found using the kab filter were in agreement with ours but the planck collaboration assigned larger statistical uncertainties , reducing their statistical significance .
the flaws of their analysis , that lead to an overestimation of their errors , are discussed in atrio - barandela ( 2013 ) . also , the ilc data is not the best suited data for this analysis . since the weights of different frequencies vary with @xmath23 , the power of the sz signal is scrambled across different multipoles and makes it very difficult to establish how filters that operate on @xmath23-space transfer the cluster signal .
peculiar velocity measurements will be better studied using planck frequency data .
frequency information will be critical to establish the origin , nature , and scale of the dipole measured in the wmap data .
cluster profiles can be measured individually to determine the effect of filtering on each of the clusters ; the frequency coverage would be crucial to remove any tsz dipole component and , specifically , the 217 ghz channel would be important to test for systematics .
fab acknowledges financial support from the spanish ministerio de educacin y ciencia ( grants fis2009 - 07238 , fis2012 - 30926 and csd 2007 - 00050 ) .
he gratefully acknowledges funding provided by nasa grant nnx10aj69 g . | recent measurements of large - scale peculiar velocities from the cumulative kinematic sunyaev - zeldovich effect using wmap data and x - ray selected clusters from rosat have identified a bulk flow of galaxy clusters at @xmath0 km s@xmath1 on scales of @xmath2 gpc , roughly aligned with the all - sky cosmic microwave background ( cmb ) dipole .
the flow is inferred from the detection of a residual dipole generated by cmb fluctuations exclusively in the direction of galaxy clusters , and measured within apertures containing zero monopole .
consistent with this interpretation , the amplitude of the dipole correlates with the x - ray luminosity of the clusters . to enable this measurement ,
the cmb data need to be filtered to remove the primary cmb , thereby increasing the data s signal - to - noise ratio .
filtering can not imprint a signal with the mentioned properties at cluster positions ; however , an inadequately designed filter can greatly suppress such a signal .
we show here that recent studies that failed to detect a large - scale flow with various filters indeed adopted flawed implementations ; when correctly implemented , these alternative filters lead to results that are in fact consistent with the dark flow signal . the discrepancies can be traced to the likely presence of residual dipoles caused by the thermal sunyaev - zeldovich effect , to assumptions about cluster profiles incompatible with the data as well as failure to compute dipoles at the zero monopole aperture .
planck maps , with their large frequency coverage and a 217 ghz channel , will be instrumental to probe bulk flows , to remove spurious dipole signals and to help identify filtering schemes appropriate for this measurement . |
the dynamics of the fluctuating magnetic field @xmath1 , is governed by the induction equation .
the velocity is assumed to be the sum of a gaussian random , delta - correlated in time @xmath2 , and an ambipolar diffusion type component @xmath3 $ ] .
( here @xmath4 , @xmath5 is some response time , and @xmath6 is the fluid density ) . assuming that the magnetic field is also gaussian random , subramanian ( 1997 , 1999 ; s99 ) derived closure equations for the longitudinal correlation function @xmath7 and the correlation function for magnetic helicity density , @xmath8 .
the random @xmath2 has a longitudinal correlation function @xmath9 and a correlation function for the kinetic helicity density , @xmath10 .
defining the operators @xmath11 , and @xmath12 , we then have , @xmath13 where @xmath14 is the correlation function of the current helicity , @xmath15 is the effective induction . also @xmath16 and @xmath17 are functions resembling the usual @xmath0-effect and the total magnetic diffusivity . here
@xmath18 $ ] and @xmath19 and @xmath20 the microscopic diffusion .
note that at large scales @xmath21 , @xmath22 and @xmath23 .
the @xmath0-effect suppression is similar to the @xmath0-suppression formula first found by pouquet et al .
( 1976 ) ; @xmath24 is however enhanced by the growing field energy , as in ambipolar diffusion . the ssd problem , with @xmath25 , @xmath26 has solutions with @xmath27 , and was first solved by kazantsev ( 1968 ) .
a transformation of the form @xmath28 , maps the problem of getting @xmath29 modes , into a bound state problem in time - independent quantum mechanics ( qm ) . for turbulent motions on a scale @xmath30 , with a velocity @xmath31 ,
bound states obtain provided the magnetic reynolds number ( mrn ) @xmath32 , and imply @xmath33 growing at rate @xmath34 .
further , the bound - state eigen - function describes a field , which is strongly concentrated within the diffusive scale , @xmath35 , and curved on the scale @xmath30 . for @xmath36 ,
@xmath20 is simply replaced by an effective , time dependent diffusion @xmath37 .
so as the field ( and @xmath33 ) grows , the effective mrn @xmath38 is driven to the critical value @xmath39 .
the final saturated state is obtained when @xmath40 .
so at saturation , the average energy density @xmath41 . for @xmath42 , @xmath43 is a small fraction @xmath44 , of the equipartition value . if we intepret the saturated field configuration in terms of flux ropes with peak field @xmath45 , thickness @xmath46 , and curved on scale @xmath30 , @xmath47 . using @xmath48 , and @xmath42
, we then have @xmath49 , where , remarkably , the @xmath50 dependence has disappeared .
so the ssd could saturate with the small - scale field of equipartition strength , being concentrated into flux ropes of thickness @xmath51 , and curved on scale @xmath30 . for helical turbulence with @xmath52 ,
new generation terms arise at @xmath53 , due to the @xmath0- effect , in the form @xmath54 and @xmath55 .
these lead to the growth of large - scale correlations on a scale @xmath56 , with a growth rate @xmath57 , as in the large - scale @xmath58- dynamo . a special wavenumber , @xmath59 , is also picked out for any quasi - stationary state . for such states , which obtain when one neglects slow resistive evolution ( see below ) , @xmath60 , implying @xmath61 . and
if we define @xmath62 , the equation @xmath63 , can again be mapped to a qm potential problem , for the zero - energy eigen - state .
however the modified potential now tends to a negative definite constant value of @xmath64 at large @xmath65 and so allows tunneling ( of the bound state ) ( see s99 ) .
in fact for @xmath53 , one has an analytic solution @xmath66 , exactly as one would get if the large scale field , @xmath67 , was random and force - free with @xmath68 .
the closure equations have also been solved numerically by brandenburg and subramanian ( 2000 ) ( bs2k ) , adopting forms for @xmath9 and @xmath10 to match closely with the direct simulation of brandenburg ( 2000 ) ( b2000 ) ( run 5 ) .
one sees an initial exponential growth of the magnetic field , which terminates when its energy becomes comparable to the kinetic energy .
note that our closure equations satisfy the helicity constraint @xmath69 .
the numerical solutions show that , after some time @xmath70 , the current helicity @xmath71 , is driven to a constant value which however is such that @xmath72 remains finite .
a constant @xmath73 implies that the magnetic helicity @xmath74 grows linearly at a rate proportional to @xmath20 . during this phase
the magnetic field correlations can extend to larger and larger scales .
the corresponding magnetic energy spectra , @xmath75 are shown in figure 1 .
the resulting magnetic field is strongly helical ( cf .
section 1.2 ) and the magnetic helicity spectra ( not shown ) satisfy @xmath76 .
the development of a helicity wave travelling towards smaller and smaller @xmath77 , as seen in figure 1 , is in agreement with the closure model of pouquet et al .
we have also checked that to a very good approximation the wavenumber of the peak is given by @xmath78 , as expected from section 1.2 , and it decreases with time because @xmath79 tends to a finite limit and @xmath80 increases .
further , since the large scale field is helical , and since most of the magnetic energy is by now ( after @xmath81 ) in the large scales , the magnetic energy is proportional to @xmath82 , and can therefore only continue to grow at a resistively limited rate .
these results are analogous to the full mhd case ( b2000 ) ; the helicity constraint is independent of the nature of the feedback ! in conclusion , our closure model with ambipolar diffusion type non - linearity provides a useful model , enabling progress to be made in understanding nonlinear dynamos .
one now needs to think of ways , to relax the helicity constraint , ( cf .
blackman and field ( 2000 ) , kleeorin et al .
( 2000 ) ) , so that large - scale magnetic fields can grow fast enough .
blackman , e. g. , & field , g. f. 2000 , , 534 , 984 brandenburg , a. 2000 , apj , ( astro - ph/0006186 ) ( b2000 ) brandenburg , a. and subramanian , k. , , 361 , l33 ( bs2k ) kazantsev , a. p. 1968
jetp , 26 , 1031 .
kleeorin , n. i , moss , d. , rogachevskii , i. , & sokoloff , d. 2000 , , 361 , l5 pouquet , a. , frisch , u. , & lorat , j. 1976 , jfm , 77 , 321 .
subramanian , k. 1997 ( astro - ph/9708216 ) subramanian , k. 1999 , , 83 , 2957 ( s99 ) | using a closure model for the evolution of magnetic correlations , we uncover an interesting plausible saturated state of the small - scale fluctuation dynamo ( ssd ) and a novel anology between quantum mechanical tunneling and the generation of large - scale fields .
large scale fields develop via the @xmath0-effect , but as magnetic helicity can only change on a resistive timescale , the time it takes to organize the field into large scales increases with magnetic reynolds number .
this is very similar to the results which obtain from simulations using full mhd .
# 1_#1 _ # 1_#1 _ = # 1 1.25 in .125 in .25 in |
accretion is one of the most important physical processes in astrophysics .
it is widely accepted that the accreting matter toward the central object is a source of power active galactic nuclei ( agn ) , and galactic x - ray sources ( see for a review e.g. frank et al .
this idea is also well - applicable to interpret many observations of astrophysical phenomena , such as , prototype stellar objects ( chang & choi 2002 ) , symbiotic stars ( lee & park 1999 ) , gamma - ray bursts ( brown et al . 2000 ) .
there are some types of accretion flows that a significant fraction of the generated heat by dissipation processes retains in the fluid rather than being radiated away , have been the subject of considerable attention in recent years ( ogilvie 1999 ; chang 2005 , akizuki & fukue 2006 , khesali & faghei 2009 , faghei 2011 ) .
these advection dominated accretion flows ( adaf ) place an intermediate position between the spherically symmetric accretion flow of non - rotating fluid ( bondi 1952 ) and the cool thin disc of classical accretion disc theory ( e. g. pringle 1981 ) .
these types of accretion flows have been widely applied to explain observations of the galactic black hole candidate ( e. g. narayan et al .
1996 , hameury et al .
1997 ) , the spectral transition of cyg x-1 ( esin 1996 ) and multi - wavelength spectral properties of sgr a@xmath0 ( narayan & yi 1995 ; manmoto et al .
2000 ; narayan et al .
1997 ) .
the x - ray observations of black holes imply that they are capable of accreting gas under a variety of flow configurations .
in particular , observational evidences confirm existence of hot accretion flow , contrasted with the classical cold and thin accretion disc scenario ( shakura & sunyaev 1973 ) . hot accretion flow can be found in the population of supermassive black holes in galactic nuclei and during quiescent of accretion onto stellar - mass black holes in x - ray transients ( e.g. , narayan et al .
1998a ; lasota et al .
1996 ; di matteo et al .
2000 ; esin et al .
1997 , 2001 ; menou et al . 1999
; see narayan et al .
1998b ; melia & falcke 2001 ; narayan 2002 ; narayan & quataert 2005 for reviews ) .
_ chandra _ observations provide tight constraints on the density and temperature of gas at or near the bondi capture radius in sgr a@xmath0 and several other nearby galactic nuclei .
tanaka & menou ( 2006 ) used these constraints ( loewenstein et al .
2001 ; baganoff et al . 2003 ; di matteo et al . 2003 ; ho et al .
2003 ) to calculate the mean free path for the observed gas .
they suggested that accretion in these systems will be proceeded under the weakly collisional condition .
furthermore , they suggested that thermal conduction can be as a possible mechanism by which the sufficient extra heating is provided in hot advection dominated accretion flows .
generally , semi - analytical studies of hot accretion flows with thermal conduction have been related to steady state models ( e. g. tanaka & menou 2006 ; johnson & quataert 2007 ; shadmehri 2008 ; abbassi et al .
2008 , 2010 ; ghanbari et al .
2009 ) , and dynamics of such systems have been studied in simulation models ( e. g. sharma et al .
2008 ; wu et al .
for example , tanaka & menou ( 2006 ) have carried out a related analysis and found the accretion flow can spontaneously produce thermal outflows driven in part by conduction .
their analysis is two - dimensional but self - similar in radius .
their assumption of self - similarity enforces a density profile that varies as @xmath2 , whereas simulations of adafs consistently find density profiles shallower than this ( e.g. , stone et al .
1999 ; igumenshchev & abramowicz 1999 ; stone & pringle 2001 ; hawley & balbus 2002 ; igumenshchev et al .
johnson & quataert ( 2007 ) studied the effects of electron thermal conduction on the properties of hot accretion flows , under the assumption of spherical symmetry . since , electron heat conduction is important for low accretion rate systems , thus their model is applicable for sgr a@xmath0 in the galactic centre .
they show that heat conduction leads to supervirial temperatures , implying that conduction significantly modifies the structure of the accretion flow .
their model similar to tanaka & menou ( 2006 ) was the steady state , but they solved their equations numerically .
as mentioned , semi - analytical studies of hot accretion flows with thermal conduction have been in a steady state .
thus , it will be interesting to study _ dynamics _ of such systems . ogilvie ( 1999 ) by the unsteady self - similar method studied time - dependence of quasi - spherical accretion flow without thermal conduction . the solutions of ogilvie ( 1999 ) provided some insight into the dynamics of quasi - spherical accretion and avoided many of the limits of the steady self - similar solution . in this research , we want to explore how thermal conduction can affect the dynamics of a rotating and accreting viscous gas .
we answer this question by solving ogilvie ( 1999 ) model that is affected by thermal conduction .
this paper is organized as follows . in section 2
, we define the general problem of constructing a model for hot quasi - spherical accretion flow . in section 3
, we use the unsteady self - similar method to solve the integrated equations that govern the dynamical behaviour of the accreting gas , and numerical study of the model is brought in this section , too .
we will present a summary of the model in section 4 .
we start with the approach adopted by ogilvie ( 1999 ) , who studied quasi - spherical accretion flows without thermal conduction .
thus , we derive the basic equations that describe the physics of accretion flow with thermal conduction .
we use the spherical coordinates @xmath3 centred on the accreting object and make the following standard assumptions : 1 .
the gravitational force on a fluid element is characterized by the newtonian potential of a point mass , @xmath4 , with @xmath5 representing the gravitational constant and @xmath6 standing for the mass of the central star .
the written equations in spherical coordinates are considered in the equatorial plane @xmath7 and terms with any @xmath8 and @xmath9 dependence are neglected , hence all quantities will be expressed in terms of spherical radius @xmath10 and time @xmath11 .
3 . for simplicity ,
self - gravity and general relativistic effects have been neglected . under these assumptions ,
the dynamics of accretion flow describes by the following equations : + the continuity equation @xmath12 the radial force equation @xmath13 the azimuthal force equation @xmath14= \frac{1}{r^{2}}\frac{\partial}{\partial r}\left[\nu\rho r^{4}\frac{\partial \omega}{\partial r}\right],\ ] ] the energy equation @xmath15 + \frac{\gamma}{\gamma-1}\frac{p}{r^2}\frac{\partial}{\partial r}\left(r^2 v_r\right)=\\ q_{vis}-q_{rad}+q_{cond}.\end{aligned}\ ] ] here @xmath16 the density , @xmath17 the radial velocity , @xmath18 the angular velocity , @xmath19 $ ] keplerian angular velocity , @xmath20 the gas pressure , @xmath21 is the adiabatic index , @xmath22 the kinematic viscosity coefficient and it is given as in narayan & yi ( 1995a ) by an @xmath1-model @xmath23 the parameter of @xmath1 is assumed to be a constant less than unity .
the terms on the right - hand side of the energy equation , @xmath24 is the heating rate of the gas by the viscous dissipation , @xmath25 represents the energy loss through radiative cooling , and @xmath26 is the transported energy by thermal conduction . for the right - hand side of the energy equation , we can write @xmath27 where @xmath28 is the advective transport of energy .
we employ the advection factor , @xmath29 , that describes the fraction of the dissipation energy which is stored in the accretion flow and advected into the central object rather than being radiated away
. the advection factor of @xmath30 in general depends on the details of the heating and radiative cooling mechanism and will vary with position ( e.g. watari 2006 , 2007 ; sinha et al .
2009 ) . however , we assume a constant @xmath30 for simplicity
. clearly , the case @xmath31 corresponds to the extreme limit of no radiative cooling and in the limit of efficient radiative cooling , we have @xmath32 . in a collisional plasma ,
mean free path for electron energy exchange , @xmath33 , is shorter than temperature scale height , @xmath34 , and thus the heat flux due to thermal conduction can be written as @xmath35 where @xmath36 is the thermal conductivity coefficient .
thermal conductivity in a dense , fully ionized gas is given by the spitzer ( 1962 ) formula , @xmath37 where @xmath38 is the electron temperature ( @xmath39 for a one - temperature plasma ) and @xmath40 is coulomb logarithm that for @xmath41 is @xmath42 the heat is conducted by the electron , and equation ( 8) includes the effect of the self - consistent electric required to maintain the electric current at zero ; this reduces @xmath36 by a factor of about @xmath43 from the value it would otherwise have ( cowie & mckee 1977 , hereafter cm77 ) .
as noted in the introduction , the inner regions of hot accretion flows are , in many cases , collisionless with electron mean free path due to coulomb collision larger than the radius ( e. g. tanaka & menou 2006 ) . when the mean free path of an electron becomes comparable to or larger than the temperature gradient scale @xmath44 , equation ( 7 ) for the heat flux
is no longer valid ; cm77 described this effect as _
saturation_. the maximum heat flux in a plasma can be expressed as @xmath45 , where @xmath46 is a characteristic velocity which one might expect to be the order of the electron thermal velocity ( parker 1963 ) .
assuming maxwellian distribution for heat source , the characteristic velocity can be written as ( williams 1971 ; cm77 ) @xmath47 similar to cm77 , we assume that the heat flux is reduced by the same factor of @xmath43 in the saturated case as in the classical ( collisional ) case so that the saturated heat flux is @xmath48 cm77 showed that the saturated heat flux is significantly less than conjectured by parker ( 1963 ) .
thus , in order to explicitly allow for uncertainty in the estimate of @xmath49 , they introduced a factor of @xmath50 , which was less than unity and rewrote equation ( 11 ) as @xmath51 where @xmath52 is sound speed , which is defined as @xmath53 .
the factor of @xmath50 is called as saturation constant ( cm77 ) .
now , the viscous heating rate and the energy transport by thermal conduction are expressed as @xmath54 @xmath55 by using equations ( 13 ) and ( 14 ) for the advective transport of energy , we can write @xmath56 the mass accretion rate in a qausi - spherical accretion flow can be written as @xmath57 we will use this quantity in the next section , and will investigate effects of saturation constant and viscous parameter on it .
@xmath70 + \frac{\gamma}{\gamma-1}\frac{\pi}{\xi^2}\frac{d}{d\xi}\left(\xi^2v \right)\\ = \alpha f \pi \xi^{7/2}\left(\frac{d\omega}{d\xi}\right)^2 -\frac{5 \phi_s}{\xi^2 } \frac{d}{d \xi } \left(\xi^2 \pi \sqrt{\frac{\pi}{r } } \right).\end{aligned}\ ] ] these equations provide a fourth - order system of non - linear ordinary differential equations that must be solved numerically . an appropriate asymptotic solution as @xmath71 is the form as @xmath72 @xmath73 @xmath74 @xmath75 in which undetermined coefficients of @xmath76 , @xmath77 , @xmath78 , and etc must be specified . by substituting above relations in equations ( 23)-(26 ) and choosing the significant sentences
, we can write @xmath79 @xmath80 @xmath81\approx0.\ ] ] also , the dimensionless mass accretion rate , @xmath82 , under the above asymptotic solution becomes @xmath83 where @xmath84 is the value of @xmath85 at @xmath86 , where @xmath86 is a point near to the centre .
after algebraic manipulations for equations ( 31)-(34 ) , we obtain an algebraic equation for @xmath76 : @xmath87 and the rest of the physical variables are @xmath88 @xmath89 @xmath90 without thermal conduction , @xmath91 , equation ( 35 ) can be solved analytically . since , we want to consider systems with non - zero saturation constant , @xmath92 , we will solve this equation numerically .
= 7.5 cm = 7.5 cm = 7.5 cm = 7.5 cm = 7.5 cm = 7.5 cm = 7.5 cm = 7.5 cm if the value of @xmath86 is guessed , i.e. we take a point very near to the centre , the equations ( 23)-(26 ) by runge - kutta - fehlberg fourth - fifth order method can be integrated from this point outward by the above expansion [ ( 27)-(30 ) ] .
examples of such solutions are presented in figs 1 - 5 .
the delineated quantity of @xmath93 in figs 1 and 2 is the sound speed square in self - similar flow , which is rescaled in the course of time and represents the flow temperature .
the profiles of @xmath93 in fig .
1 show the flow temperature decreased by adding saturation constant , @xmath50 . because , the generated heat by viscous dissipation can be transfered by thermal conduction .
furthermore , this temperature decrease is qualitatively consistent with simulation results of sharma et al .
( 2008 ) and wu et al .
we know the viscous dissipation of the flow increases by adding @xmath1 parameter .
thus , the temperature increased by viscous parameter confirmed by the temperature profiles in fig 2 .
the density profiles show the gas density increased by adding the @xmath50 parameter
. it can be due to temperature fall of fluid .
increasing density by adding saturation constant is another consistency of our results with simulations of wu et al .
also , the density profiles show that it decreases by adding the viscous parameter .
this also could be result of the temperature rising .
the viscous turbulence in this paper is proportional to the gas temperature ( @xmath94 ) .
thus , increase or decrease of temperature will affect dynamics of the accreting gas . as we said
the temperature decreased by saturation constant implies that the viscous turbulence decreases , too .
the decreasing of viscous turbulence reduces the effect of negative viscous torque in angular momentum equation .
thus , we expect the flow rotates faster by adding saturation constant that confirmed by the angular velocity profiles in fig .
since , the efficiency of the angular momentum transport decreases by adding the saturation constant , we expect the decrease of radial infall velocity that the radial velocity profiles in fig .
1 confirm it .
the efficiency of angular momentum transport increases by adding the viscous parameter of @xmath1 .
thus , we expect the flow rotates slower and accretes faster by adding the viscous parameter .
the profiles of radial and angular velocities in fig .
2 confirm them .
= 8.5cm=8.5 cm the behaviour of mass accretion rate as a function of similarity variable @xmath58 for several values of the viscous parameter and saturation constant are plotted in fig .
3 . in the present model
, the mass accretion rate is reduced by radius . while , the mass accretion rate in steady hot accretion flows is a constant ( tanaka & menou 2006 ) .
there are some researches in steady hot accretion flows that have studied power - law function of mass accretion rate ( shadmehri 2008 ; abbassi et al .
however , the mass accretion rate in their models is not dependent on important parameters such as saturation constant and viscous parameter .
the profiles of mass accretion rate in fig . 3 show that it is reduced by adding the saturation constant .
this property is qualitatively consistent with numerical results of johnson & quataert ( 2007 ) .
also , the profiles of mass accretion rate imply that it increases by adding the viscous parameter of @xmath1 .
this property is qualitatively consistent with previous works in accretion flows ( e. g. park 2009 ) .
= 8.5 cm = 8.5 cm here , it will be interesting to investigate the existence of the transonic point in hot accretion flow .
the transonic point occurs in place that the amount of _ mach number _ becomes equal to unity .
the mach number referring to the reference frame is defined as ( gaffet & fukue 1983 ; fukue 1984 ) @xmath95 where @xmath96 is the velocity of the reference frame which is moving outward as time goes by , which the sound speed can be subsequently expressed as @xmath97 and , @xmath98 the sound speed in self - similar flow is rescaled in the course of time .
the mach number introduced so far , represents the _ instantaneous _ and _ local _ mach number of the unsteady self - similar flow . in steady self - similar solution
( e. g. tanaka & menou 2006 ) , the mach number does not vary by radii and is a constant . while , the mach number in unsteady self - similar varies by radii ( see fig .
4 ) . as seen in fig .
4 , there is a transonic point ( @xmath99 ) .
the dependency of transonic point to saturation constant shows that this point moves inward by adding the parameter of @xmath50 . because , thermal conduction transfers the heat to larger radii , so the sound speed / temperature decreases by adding the saturation constant . in other words ,
whatever the radii smaller , the radial velocity relative to the sound speed larger .
also , the mach number profiles show the transonic point decreasing by adding the viscous parameter which can be due to increase of radial velocity along with adding the viscous parameter .
= 8.5 cm = 8.5 cm = 8.5 cm = 8.5 cm as a comparison between steady and unsteady self - similar solution , the physical quantities in unsteady self - similar solutions are divided into their radial dependence in steady self - similar solution then plotted in terms of @xmath58 in fig .
also , the effect of adiabatic index on hot accretion flow is investigated in fig .
the delineated quantities ( @xmath100 , @xmath101 , @xmath102 @xmath102 @xmath102 ) in fig 5 are constant in steady self - similar solutions of hot accretion flows ( tanaka & menou 2006 ; shadmehri 2008 ; abbassi et al . 2008 ; ghanbari et al .
2009 ) . while they vary with position in this research .
5 represents the density profile varies shallower than @xmath103 , that this property is qualitatively consistent with simulation results ( e.g. , stone et al .
1999 ; igumenshchev & abramowicz 1999 ; stone & pringle 2001 ; hawley & balbus 2002 ; igumenshchev et al .
2003 ) . the radial dependency of the temperature and radial velocity in unsteady self - similar solution show that they vary deeper than the steady self - similar solution . also ,
the study of the angular velocity show that it varies shallower than @xmath103 .
thus , the physical quantities in the present model avoid the limits of the steady self - similar solution .
the physical quantities profiles in fig . 5 show that their radial dependency in the unsteady self - similar limited to the steady self - similar solution by adding adiabatic index @xmath21 .
in hot accretion flows , the collision timescale between ions and electrons is longer than the inflow timescale . thus , the inflow plasma is collisionless , and the transfer of energy by thermal conduction can be dynamically important .
the low collisional rate of the gas is confirmed by direct observation , particularly in the case of the galactic centre ( quataert 2004 ; tanaka & menou 2006 ) and in the intracluster medium of galaxy clusters ( sarazin 1986 ) . here
, we have investigated how thermal conduction affects dynamics of hot quasi - spherical accretion flows .
we adopted the presented solutions by ogilvie ( 1999 ) and tanaka & menou ( 2006 ) .
thus , we assumed that angular momentum transport is due to viscous turbulence and the @xmath1-prescription is used for the kinematic coefficient of viscosity .
we also assumed the flow does not have a good cooling efficiency and so a fraction of energy accretes along with matter on to the central object .
the effect of thermal conduction is studied by a saturation form of it introduced by cowie & mckee ( 1977 ) . to solve the equations that govern the dynamical behaviour of hot accretion flow
, we have used unsteady self - similar solution .
the effect of saturation constant and the viscous parameter on the present model is investigated .
the solutions show that with the increase of conductivity , the equatorial density becomes denser and the temperature becomes lower .
these results are qualitatively consistent with simulation results of wu et al .
furthermore , the solutions show that by adding the saturation constant , the angular velocity becomes larger and the radial velocity decreases .
the mass accretion rate is reduced by adding the saturation constant that is qualitatively consistent with the result of johnson & quataert ( 2007 ) .
the solutions imply that the viscous parameter has opposite effects in comparison to saturation constant on physical quantities of the system .
also , the study of physical quantities of the present model in comparison to steady self - similar solution show that our results deviate from steady self - similar solution and do not have its limits . here , we studied dynamical behaviour of hot accretion flow in one - dimensional approach ignored by latitudinal dependence of physical quantities .
although , some authors have shown that latitudinal dependence of physical quantities is important for structure and dynamics of hot accretion flow ( tanaka & menou 2006 ; ghanbari et al . 2009 ; wu et al . 2010 ) .
thus , latitudinal behaviour of the present model can be investigated in other studies .
i would like to thank the anonymous referee for very useful comments that helped me to improve the initial version of the paper . | the purpose of this paper is to explore the dynamical behaviour of hot accretion flow with thermal conduction .
the importance of thermal conduction on hot accretion flow is confirmed by observations of the hot gas that surrounds sgr a@xmath0 and a few other nearby galactic nuclei . in this research ,
the effect of thermal conduction is studied by a saturated form of it , as is appropriate for weakly collisional systems .
the angular momentum transport is assumed to be a result of viscous turbulence and the @xmath1-prescription is used for the kinematic coefficient of viscosity .
the equations of accretion flow are solved in a simplified one - dimensional model that neglects the latitudinal dependence of the flow . to solve the integrated equations that govern the dynamical behaviour of the accretion flow
, we have used an unsteady self - similar solution .
the solution provides some insights into the dynamics of quasi - spherical accretion flow and avoids from limits of the steady self - similar solution . in comparison to accretion flows without thermal conduction ,
the disc generally becomes cooler and denser .
these properties are qualitatively consistent with performed simulations in hot accretion flows . moreover , the angular velocity increases with the magnitude of conduction , while the radial infall velocity decreases .
the mass accretion rate onto the central object is reduced in the presence of thermal conduction .
we found that the viscosity and thermal conduction have the opposite effects on the physical variables .
furthermore , the flow represents a transonic point that moves inward with the magnitude of conduction or viscosity .
[ firstpage ] accretion , accretion discs , conduction hydrodynamics . |
we report a low temperature measurement technique and magnetization data of a quantum molecular spin , by implementing an on - chip squid technique .
this technique enables the squid magnetometery in high magnetic fields , up to 7 tesla .
the main challenges and the calibration process are detailed .
the measurement protocol is used to observe quantum tunneling jumps of the @xmath0 molecular magnet , mn@xmath1-tbuac .
the effect of transverse field on the tunneling splitting for this molecular system is addressed as well .
_ keywords _ : squid magnetometery , on - chip devices , quantum tunneling , high magnetic field recent investigations on molecular magnets @xcite or less complex systems containing diluted spins @xcite have been motivated by their quantum mechanical properties and for future implementation in quantum computing @xcite .
such studies require a large number of identical and independent magnetic cores ( often with a large spin ) , to provide a natural amplification of the fundamental quantum mechanical phenomena of a single molecule to measurable signals .
remarkable quantum phenomena , such as the quantum tunneling of magnetization @xcite and the berry phase quantum interference @xcite , have been observed so far .
sensitive magnetization measurements in sweeping magnetic fields have been performed by means of superconducting quantum interference devices ( squids ) @xcite , hall probe magnetometers @xcite or electron paramagnetic resonance ( epr ) @xcite spectroscopy .
such techniques are highly desirable to gather more information about effects such as the anisotropy - induced tunneling gap@xcite , spin states entanglements , phonon bottleneck effects @xcite and others .
squid devices are well established as highly sensitive magnetic flux detectors for low temperature applications .
a commercial squid fabricated with nb - alo@xmath2-nb josephson junctions is usually magnetically shielded and coupled to the sample through a pick - up coil @xcite .
this arrangement keeps the magnetic field from disturbing the squid .
however , it also leads to a non - optimal coupling factor and excludes it from measuring samples of submicron size .
a better coupling factor can be achieved by reducing the squid loop size and coupling to it directly , in the presence of a magnetic field .
this represent an important advantage over hall probes , since josephson junctions can have widths of the order of the nanometer and have the sample on top or inside of it for optimal coupling .
it is estimated that such squid implementation can detect , without extensive averaging , only few bohr magnetons @xcite .
the planar tunneling junctions are not suitable because of their sensitivity to large magnetic field and relatively large dimensions .
another type of josephson junctions ( nano - bridges ) @xcite led to a promising design in which the squid is less sensitive to a field aligned along its plane , since the critical field value of a superconducting thin film can be much higher than that of the bulk in such a case @xcite . in this report , we present an improved on - chip squid measurement technique , in which a field up to 7 t can be aligned with high precision in the plane of a nb squid .
thus , we have extended the observation of fundamental quantum phenomena by means of on - chip squid measurements into a higher magnetic field range , _
i.e. _ to a larger family of molecular magnets .
the technique allows real time ( on - the - fly " ) squid magnetization measurements on samples placed in the vicinity of the squid loop , to optimize the magnetic coupling . both the sample and the squid detector
are placed in the center of a 3-d vector magnet and thermally attached to the mixing chamber of a dilution refrigerator to ensure an efficient cooling .
the on - chip squid is made of a nb film of only 5.5 nm thickness ensuring a superconducting state even for 7 t of in - plane magnetic field .
the squid loop has a 2@xmath32 @xmath4 area and contains two parallel nano - bridge josephson junctions .
therefore , the i - v characteristic of a squid is similar to a single josephson junction with flux - dependent critical current and it can be explained with the resistively and capacitively shunted junction ( rcsj ) model @xcite . in the rcsj model ,
the phase of a squid is described by the motion equation of a ball trapped in a washboard potential .
an increase in the squid current tilts the washboard potential , and the phase ball has a probability to escape ( and roll down along the tilted washboard ) due to both thermal activation and quantum tunneling @xcite .
the rolling of the phase ball generates a voltage across the squid and switches it from the superconducting state into the normal state .
therefore the switching probability @xmath5 increases with the current amplitude .
we define the switching current @xmath6 as the current corresponding to @xmath7% @xcite . in our experiment ,
current pulses are generated by a pulsed voltage source and an in - series resistor of 15 k@xmath8 .
the @xmath6 current shows a typical periodic modulation as a function of the trapped magnetic flux in the squid loop ( ( a ) ) , with a period of @xmath9 ( @xmath10 is planck s constant and @xmath11 the electron charge ) .
the modulation amplitude is 5% , much smaller than typical values for squids with tunneling junctions but it can be used in fields up to 7 t. modulation ( shown as the height of voltage pulses applied on squid in series with a 15 k@xmath8 resistor , such that @xmath7% ) as a function of the field generated by the feedback coil ( expressed in ma of coil current ) .
( b ) same modulation as in ( a ) , obtained by fixing the applied pulse height , such that @xmath7% for the designated working point , and ramping the coil current .
the steep slopes leads to squid s high sensitivity . ]
operating the squid in pulse mode , minimizes the heating generated by its switching into the normal state .
the stochastic switching process is monitored by sending a series of current pulses of length @xmath12s to the squid and monitoring the voltage across it .
when the squid switches into the normal state ( to a resistance of @xmath134 k@xmath8 ) , voltage pulses are generated , amplified , and counted at the readout port .
statistically , the ratio of the counted output voltage pulses by the number of applied current pulses gives @xmath5 for that current value .
this switching probability is modulated by the magnetic flux , as shown in ( b ) .
the counting of switching events is expedited by operating the pulse mode at a high repetition frequency ( here , 50 khz ) .
the mechanism used to perform an real time measurement will now be described . as shown in ( b ) , we define the middle point in the modulation curve as the `` working point '' . during the measurement , a current pulse with a height corresponding to the working point is sent to the squid .
it is important to note that the variation of the probability near the working point is very sharp , which is the key point behind the high flux sensitivity of a squid . to lock the squid at its working point and perform a feedback measurement , we place a small superconducting coil above the squid ( see below , insert ) to actively compensate for flux changes caused by the magnetic sample .
the feedback coil current is proportionally related to the change in the magnetic moment of the sample in an on - the - fly measurement .
although the principle of the measurement is relatively simple , its practical realization is quite challenging .
the squid s thin film remains superconducting only when the high field is perfectly parallel to the plane .
even the smallest deviation from a precise alignment , can generate unwanted normal fields leading to vortices in the thin nb film and altering the measurements .
large deviation simply switches the squid into the normal state . also , the critical current of any superconducting material decreases with the increase of magnetic field , and so is the pulse height corresponding to the working point .
the feedback mechanism only can be operated with a proper setting of the working point and therefore , the height of the current pulses corresponding to @xmath7% has to be carefully calibrated , as explained below .
the precise alignment of the magnetic field in the squid plane is done by combining fields generated by a 3-axis magnet @xmath14 .
initially , the squid plane is roughly aligned along @xmath15 .
hence , we only need to scan @xmath16 to fine - tune the direction of the field in the squid plane .
the squid can only sustain a small normal field ( @xmath1315 - 20 mt ) , beyond which , the squid switches and stops working . in the calibration process , we scan @xmath16 ( chosen roughly normal to the squid plane ) for a series of given z fields , and record the field region in which the squid works ( see , triangles ) . the @xmath16 value in the middle of that region ( shown with squares in ) combined with that given @xmath15 creates a total field in the squid plane . and @xmath17 . the required @xmath16 ( squares ) is at the middle of the region ( limited by triangles ) where the squid is not switched into the normal state by the magnetic field .
the linear fit leads to the misalignment angle between the holder and the coil system .
( insert ) maxima and minima ( triangles ) of the @xmath6 modulation as a function of @xmath15 ( as in the main plot ) .
the working point ( squares ) is chosen at half - distance and a parabolic fit calibrates the required decrease in the excitation pulse to keep the squid ( in series with a 15 k@xmath8 resistor ) at @xmath1850% . ] during the field alignment described above , the maximum and minimum values of @xmath5 modulation are recorded as a function of the magnetic field .
these values are decreasing with field increase in a parabolic manner ( see insert ) , as a consequence of the critical current decrease , mentioned above .
the working point is calibrated at the mid - distance between the two parabolas , so that the switching probability is kept constant at 50% during squid operation .
we demonstrate our first magnetization tunneling measurements with this high - field squid detection technique , on the mn@xmath1-tbuac crystal , a molecular magnet requiring a high field ( 5.5 t ) to reach magnetization saturation .
a trace , starting from high negative fields ( negative saturation ) and up to @xmath195.5 t , is shown in .
( thick line ) .
tunneling steps are clearly visible ( see @xcite for a similar hall - probe based measurement ) .
also , we were able to construct a constant transverse field normal to the crystal s magnetic easy axis , while keeping the total field perfectly aligned in the squid plane during the sweep of the field in the longitudinal direction .
the magnetization curves of the sample show differences in the size of the tunneling steps ( , thin line ) .
this indicates that the transverse magnetic field modifies the size of the tunneling gaps of this molecular magnet , similarly to a berry - phase process ( as in @xcite ) .
-tbuac single crystal .
the geometry is sketched in the insert : a single crystal is placed near an on - chip squid and under the feedback coil .
the sample is first saturated at -5.5 t , and a real time measurement follows up to + 5.5 t , here at a field rate of 1 mt / s .
the two curves shows the differences in the quantum tunneling jumps - and therefore in the tunneling splittings- for two values of the transverse field , @xmath20=0 t ( thick line ) and 0.2 t ( thin line ) respectively . ]
we have implemented the described feedback mechanism , simultaneously to two identical , neighboring squids . in this arrangement ,
the two squids sense similar external background field . if the studied sample is placed between the two squids , the magnetic flux lines generated by it are penetrating the squid loops in opposite directions .
therefore , a simple subtraction of two measurements can eliminate the uniform background and double the magnetic signal .
individual runs on each ( empty ) squid show a certain wavy background signal .
following their subtraction , which can be done in real time during a field sweep , the background is significantly flattened , suggesting that such a differential readout can further improve sensitive squid - based magnetic detection . in conclusion , our on - chip squid detection has extended the possibility of performing magnetization studies on quantum spins into the high field regime ( up to 7 t ) . this versatile technique will allow the exploration of the quantum properties of more types of magnetic molecules . during its operation
, the sweeping magnetic field has to be perfectly aligned in the squid plane and the excitation current pulses have to be properly calibrated with field . the magnetic detection is using the feedback information from a superconducting coil fixing the working point of the squid .
we have also succesfully tested a differential two - squid method which significantly reduces unwanted background signals and thus improves the squid capability .
we acknowledge support from the nsf cooperative agreement grant no .
dmr-0654118 , the nsf - career grant no .
dmr-0645408 and the alfred p. sloan foundation .
ww thanks the financial support from the anr - pnano projects molnanospin anr-08-nano-002 and the erc advanced grant molnanospin 226558 . |
|
in the last two decades there has been an increasing interest in non - hermitian(nh ) quantum mechanics ( qm ) in many different fields of physics . in
many studied cases in atomic , molecular and nuclear physics the hamiltonian is nh due to the use of different type of analytical continuation transformations .
one such method is complex scaling ( cs ) where @xmath1 @xcite@xcite@xcite@xcite .
another common method is the introduction of complex absorbing local energy independent potentials ( caps ) , also known as optical potentials , into the hamiltonian@xcite@xcite such that @xmath2 .
the complex eigenvalues and eigenfunctions of the nh hamiltonians , in these studies , were calculated using square - integrable methods that were originally developed for the calculation of bound states ( see references to reviews mentioned above ) . the need to replace the scalar product which is used in conventional qm by a generalized inner product
is well understood and has been discussed quite extensively in the literature .
see for example the different approaches for generalized inner products in nh - qm in ref .
@xcite and ref.@xcite and references therein .
however , for the sake of clarity in section [ c - product ] we provide a brief account of the generalized inner product ( so called c - product@xcite ) which has been used until now for calculating time independent observables in nh - qm@xcite .
it is quite straightforward to apply the same generalized definition of the inner product to nh time periodic hamiltonians ( see for example ref.@xcite ) , which describe the interaction of atoms or molecules with oscillating electric field .
the application of the nh formalism to laser driven system has been very useful in explaining physical phenomena . for example
, remarkable agreement has been achieved between the results obtained from theoretical nh - qm calculations and the experimental results of the probability of generating odd high - order harmonics ( i.e. , generation of high energy photons resulting from the absorbtion of many low energy photons)@xcite .
however , one important problem remains open . how to calculate the entire high - order harmonic generation spectra and not only the probabilities to obtain even or odd harmonics ?
this question is related to a more general question . how to calculate time dependent expectation values in nh - qm ? in nh - qm
the propagated right wavepacket ( wp ) is the solution of the time dependent schrdinger equation ( tdse ) where @xmath3 is a nh hamiltonian .
@xmath4 extensive wp propagation calculations were carried out for complex potential energy surfaces by solving the nh - tdse ( see for example refs.@xcite ) . however , in order to calculate time dependent expectation values one needs to carry out wp propagation calculations from @xmath5 to @xmath6 .
the problem is that in nh - qm it is impossible to propagate the initial wp from t=0 to @xmath6 .
we briefly review this known problem in section [ asymmetry ] and relate it to the concept of time asymmetry in qm which is a subject of numerous studies in the literature ( see refs.@xcite and references therein ) .
+ very recently a possible solution to the time asymmetry problem in nh - qm was proposed by moiseyev and lein@xcite .
the solution requires modification of the currently used inner product . as we will show in section [ f - product ] this modification of the well known generalized inner product ( briefly reviewed in section [ c - product ] ) which we will term `` f - product '' ( fp )
is applicable for cases where the hamiltonian is either a time - independent or time dependent nh operator .
naturally this modification of the generalized inner product reduces to the conventional scalar product when the hamiltonian is hermitian .
moiseyev and lein @xcite have shown that using the modified generalized definition of the inner product not only solves the time asymmetry problem in nh - qm , but also has a very important physical implications .
they have shown that the analytical expression for the high - order harmonic generation spectra ( hhgs ) which has been derived within the framework of the fp modified nh time dependent formalism , clearly indicates why only odd high - order harmonics were observed in experiments where high - intensity laser pulses have been used . moreover , the analysis of the expressions they obtained for the hhgs indicates under what conditions also even high - order harmonics would be observed .
simulation calculations within the framework of the conventional ( hermitian ) qm based on their finding supported their results .
the possibility of generating also the even - order harmonics as well as the odd ones should help in generating ultra short high intensity laser pulses .
however , so far the fp generalized inner product as first proposed in ref@xcite has never been tested for numerical nh calculations .
the purpose of this work is to calculate time - dependent observable quantities for a well known test case study problem within the framework of the fp formalism and compare it with the results obtained from the conventional ( i.e. , hermitian ) qm . as we will show here such a comparison clearly shows that time dependent observables should be calculated within the framework of the fp formalism and also give some physical interpretation of this formalism .
non - hermitian hamiltonians have been used to describe a variety of physical phenomena .
the methods by which the hamiltonian becomes nh are also diverse .
such methods include for example the addition of complex absorbing potentials to the hamiltonian or the scaling of the coordinate by a complex factor .
the use of nh hamiltonians results in complex eigenvalues , @xmath7 where often the real part , @xmath8 ) , is associated with the energy while the complex part is related to the decay rate , @xmath9 of a metastable state by : @xmath10 such that the lifetime of the metastable state is given by @xmath11 .
when using nh hamiltonians one needs to address the aspects that differ from conventional qm . as a result of the nh nature of the hamiltonian ,
there is a need to define a generalized inner product .
the question of the definition of the inner product when the hamiltonian is nh is a crucial one , especially in time dependent calculations .
let us consider a general case where the hamiltonian , @xmath12 can be either hermitian or nh hamiltonian . since the concept of a generalized inner product
is well defined in linear algebra for general not necessarily hermitian matrices@xcite we begin our brief review by the representation of the hamiltonian by a matrix @xmath13 .
the hamiltonian matrix elements are defined as @xmath14 where @xmath15 are orthonormal square integrable basis set functions in the hilbert space .
we assume here that we can truncate the number of the basis functions and represent the hamiltonian by a finite general complex matrix
. the matrix can be as large as one wishes .
the matrix @xmath13 has right and left eigenvectors .
let us denote the right ( column ) eigenvectors by @xmath16 and the left ( row ) eigenvectors by @xmath17 .
that is , @xmath18 and @xmath19 by taking the transpose of eq.[left ] one gets , @xmath20 the matrix @xmath13 and its transpose @xmath21 support the same spectrum .
we define the generalized inner product , termed c - product @xcite , such that for non degenerate states where @xmath22 , @xmath23 from eq.[left2 ] it is obvious that when @xmath13 is hermitian matrix ( i.e. , @xmath24 ) then @xmath25 and the c - product @xmath26 is equal to the known scalar product @xmath27 . if for example @xmath13 is a symmetric matrix ( note any matrix can be transformed to a symmetric form and therefore a complex symmetric matrix is a general representation of a nh matrix ) then , @xmath28 do the eigenvectors of a non - hermitian matrix @xmath13 form a complete set ?
a complete set implies that the number of linearly independent eigenvectors is equal to the dimension of the matrix .
it may happen that due to a coalescence of two eigenvectors ( coalescence of more than two eigenvectors is very unlikely phenomena@xcite@xcite ) the number of the linearly independent eigenvectors is smaller than the dimension of the matrix and the spectrum is incomplete .
this coalescence of the eigenvalues happens at @xmath29 ( where `` b '' stands for a branch point in the complex energy plane@xcite)for @xmath30 the specific eigenvector denoted by @xmath29 is `` self - orthogonal '' in the sense that , @xmath31 .
however , due to round off numerical errors it is quite impossible to get the incomplete spectrum and always in the numerical calculations the eigenvectors are normalizable such that
. latexmath:[$({\vec \phi}_{\alpha } functions , @xmath15 , which are used to construct the hamiltonian matrix in eq.[ham])is increased , the function @xmath33 given by : @xmath34_i \varphi_i\ ] ] gives a better description of the @xmath35 eigenfunction of the hamiltonian .
it is out of the scope of this paper to describe the complex ( i.e. , nh ) variational theorem that has been proved for complex symmetric hamiltonian matrices ( see ref.@xcite and references therein ) .
however , on the basis of the complex variational principle the exact eigenfunction , @xmath36 can be expressed as , @xmath37 and similarly , @xmath38_i \varphi_i^*\ ] ] where , @xmath39 and @xmath40 here @xmath41 is the hermitian conjugate of @xmath42 such that , @xmath43^*\ ] ] the basis functions are square integrable functions as described above .
now , the generalized inner product ( so called c - product ) for the left and right eigenfunction , @xmath44 , will be defined by : @xmath45 note that the c - product , @xmath46 , may be complex and may have negative real or imaginary parts , therefore the c - product is not a metric scalar product @xcite . for a physical interpretation of the complex density probability @xmath47
see ref.@xcite .
time asymmetry in physics is a concept closely related to irreversibility which has been a subject of many theoretical studies .
the conventional hermitian qm is time - symmetric in the sense that it is described by an equation symmetrical with respect to time and by time - symmetric boundary conditions@xcite .
the process of the decay of a metastable state is an irreversible one in the sense that it can only be described from a certain time @xmath5 to time @xmath48 .
authors such as bohm @xcite , prigogine @xcite , sudarshan et .
@xcite have constructed formalisms which incorporate the irreversible nature of the resonance phenomena into qm usually by relying on the introduction of rigged hilbert spaces .
this leads to semi group evolution which distinguished between `` prepared '' stated and `` measured '' states @xcite .
however , it has been argued by nicolaides @xcite that time asymmetry results from complex energy distribution without the need to introduce rigged hilbert spaces .
+ indeed , in the spirit of ref.@xcite , we will show that in nh - qm , within the spectral representation of the nh hamiltonian , time asymmetry will pose a problem in the propagation of wp s .
the need to carry out wp calculations within the framework of nh - qm is a crucial point in the study of systems where the dynamics are not controlled by a single resonance state .
when one tries to propagate a wp in nh - qm the evolution of each of the stationary eigenfunctions in eq.[phi_r ] is governed by the nh - tdse and the stationary solutions of eq.[nh - tdse ] are given by : @xmath49 when one attempts to propagate the corresponding stationary solutions of the transposed nh hamiltonian @xmath50 ( see eq.[phi_l ] ) , at time @xmath51 the left eigenfunctions will be given by : @xmath52 since @xmath53 the left eigenfunctions diverge exponentially as @xmath54 .
in hermitian qm @xmath55 and @xmath56 is real , thus , @xmath57 does not diverge . since the only difference from the evolution of @xmath58 is in the @xmath59 sign in the exponent it is similar in a way to propagating the initial state to a negative time .
since in nh - qm the energy is complex , the wavefunction can not be propagated backward to @xmath60 but only forward to @xmath61 .
this time asymmetry problem dissolves when studying cases where only one metastable state dominates since when calculating any observable quantity the time dependent phases of the right and left eigenfunctions , @xmath62 , cancel out .
when one studies the evolution of a wp , the right function obeys the nh - tdse in eq.[nh - tdse ] and can be constructed as a linear combination of the eigenfunctions of the nh hamiltonian given in eq.[psi_r ] . the left function will be a linear combination of the eigenfunctions given in eq.[psi_l ] .
now , when calculating any observable quantity , which does not commute with @xmath3 , there will be cross terms between different eigenfunctions with time dependent phase of the form , @xmath63 .
these terms will diverge at long times as @xmath64 and prevent the calculation of observable quantities of a wp in nh - qm .
+ this is , in short , the time asymmetry problem in nh - qm . due to this problem
the calculation of observables in nh - qm is feasible only when one long lived metastable state controls the dynamics of the system and @xmath65 is time independent .
a definition of an observable for a general superposition of resonance states is required .
this subject will be discussed in sections [ f - product ] and [ 1dmodel ] .
as stated earlier the time asymmetry problem prevents the propagation of wp s backward in time in nh - qm .
the existing definition of the left eigenfunctions of the nh hamiltonian is based on the c - product as defined in section [ c - product ] by eq.[psi_l ] , which diverges exponentially in time due to the complex part of the energy . the one - state probability density
@xmath66 is complex and time independent .
however , as mentioned before the imaginary part of the complex energy is associated with the decay of a metastable state therefore one would expect that the probability to find the particle in such a state should decay in time with a rate of decay @xmath9 .
+ in order to impose this decay behavior of a metstable state we apply the finite - range f - product ( fp ) formalism which was proposed in ref.@xcite but was never tested by carrying out wp propagation calculations within the framework of nh - qm . for the sake of clarity let us first explain the idea behind the f - product formalism . as we have seen in section [ asymmetry ] , when the left function is _ stationary _ solution of eq .
[ phi_l ] it is defined by eq.[psi_l ] . although @xmath67 diverges exponentially as time goes to infinity , the norm is preserved and @xmath68 and @xmath69 which project the system into subspaces of discrete and continuum states respectively .
the resulting hamiltonian for the bound part of the system is given by : @xmath70 where , @xmath71 is the hamiltonian for the `` system '' and @xmath72 is the green operator for the particle in the continuous or dissociative part of the spectrum which describes the `` surrounding '' to which the particles decay .
the effective hamiltonian as defined in eq .
[ feshham ] is a nh operator .
it is clear that here the effective hamiltonian is nh due to splitting of space into an interaction region where our system is located and a `` surrounding '' which absorbs the emitted particles ( sub - systems ) .
however , we wish to associate the decay phenomena with particles which `` escape '' from the interaction region to its `` surrounding '' even in cases where the nh formalism takes the entire space into account .
how to split the entire space in the most general case into an interaction region and its `` surrounding '' ?
we will return to this question in the next section . here
we argue that when the dynamics is controlled by a single isolated metastable state within the nh - qm formalism then we may use the finite box quantization approach where the interaction region can be defined as a box as large as one wishes . in such a case
the integrals are calculated not over the entire space , i.e. , from @xmath73 to @xmath61 , but from @xmath74 to @xmath75 where the size of the box , @xmath76 , is as large as one wishes , hence the name - finite - range given to this new formalism .
since we want to describe a decay phenomena we now define the evolution of the left eigenfunctions as , @xmath77 now the probability do detect the particle in a decaying state will decrease exponentially in time according to : @xmath78 this definition also prevents the divergence in time of the left functions and enables the propagation of a wp in time . within the fp formalism
a wp will be now defined by a superposition of the eigenfunctions of the nh hamiltonian .
@xmath79 where @xmath80 when the hamiltonian is nh due to the application of one of the complex scaling ( cs ) similarity transformations , then the initial state @xmath81 is given by @xmath82 where @xmath83 is one of the cs transformations@xcite .
on the other hand when cap is added to the hamiltonian and the initial state is localized in the interaction region it remains unscaled .
an important point in the new definition of the inner product is that now the right - wp satisfies the nh - tdse as defined in eq.[nh - tdse ] , whereas the left - wp does _ not _ satisfy a corresponding wave equation .
this is a very important point in our search for solution of the time asymmetric problem in nh - qm .
the fp definition of the inner product generalizes only the time dependence of the c - product definition . therefore , when the functions are time independent the f - product reduces to the c - product .
based on the eq .
[ fp - psi ] the decay of a wp will now be given by : @xmath84 note that although there are no cross terms between different @xmath85 s there are still interference effects since @xmath86 have complex values .
nevertheless the absolute value of this expression decays continuously as time passes .
the goal of this paper is to check the fp formalism for the first time for a simple test model problem and compare the results with those obtained from conventional qm calculations .
such a comparison is made in section [ 1dmodel ] .
we would like to implement the new formalism on a simple one dimensional time independent hamiltonian , which will serve as a test model : @xmath87 this model hamiltonian has been often used as a test case for new theories and computational methods(see for example @xcite ) .
the potential in eq.[jolanta ] consists of a potential between two barriers ( see fig.[jola - cs ] ) . in the example
presented below the hamiltonian becomes nh upon complex scaling of the coordinate by a complex phase , @xmath88 , such that @xmath89 and is given by : @xmath90 the model potential shown in fig.[jola - cs ] supports a bound state and two metastable resonances below the top of the barrier as well as several other resonances over the barrier .
the resonances are localized states that decay in time and can not be described by a single eigenfunction of the hermitian hamiltonian in eq.[jolanta ] , however in nh - qm they are represented by a single square integrable eigenfunction of the nh hamiltonian in eq.[jolanta - cs ] as can be seen in fig.[jola - cs ] .
our objective is to show that the fp formalism will yield similar results to those obtained by conventional hermitian qm , without the need to divide our space to `` system '' and `` surrounding '' . for this purpose lets define now the interaction region as the area between the top off the two barriers in fig.[jola - cs ] and return to this point later on .
+ in order to observe the decay of a wp with time we will place a gaussian wp in the center of the potential in eq.[jolanta ] of the form : @xmath91 where @xmath92 is the width of the wp and @xmath93 its initial momentum . upon cs the right initial state wp becomes : @xmath94 while the left initial wp is given by : @xmath95 note that at time @xmath5 the definitions of the c - product and the f - product are identical and @xmath96 .
the norm of the wp , @xmath97 , within the fp formalism , as defined in eq.[fp - psi ] , and is given by : @xmath98 assuming the decay of the wp is a first order process the effective decay rate will be given by : @xmath99 when the wp populates several resonance states the effective decay rate will be time dependent , but when only one resonance state is populated the anticipated constant effective decay rate will be obtained , @xmath100 .
in contrast , the decay will not be observed in hermitian qm where the norm is conserved .
therefore we choose a new quantity for comparison with the norm in eq.[normfp ] , which we will label @xmath101 and is defined by the part of the wp which is localized inside the interaction region .
@xmath102 where @xmath103 is the result obtained from conventional propagation calculations , when by conventional we mean the solution of the hermitian tdse .
this enables us to define an effective decay rate even in hermitian qm based on eq.[normhqm ] .
@xmath104 the question we address ourselves with now is how to define the interaction region ? or similarly what is @xmath76 in eq.[normhqm ] ?
is the initial assumption that we can define the interaction region as the area between the top of the two barriers valid ? here , we are not using the feshbach formalism and therefore we look for a simple universal definition .
we propose here to define the interaction region ( the parameter `` a '' in our 1d case or the vector @xmath105 in multidimensional case ) based on the nh calculations .
more precisely we wish to define the interaction region as the region where the resonances are localized .
the continuum states in the same energy range have a very small ( almost vanishing ) amplitude in this region as can be seen in fig.[jola - cs ] .
this definition is obviously not an exact one and its important to note that using the fp formalism we avoid the need to define the interaction region and thus @xmath76 is only relevant when trying to relate the results to conventional qm . the results which will be presented below strongly support our conjecture .
+ returning to the problem of the decay of the norm of a wp , we first place a gaussian wp as shown in eq.[gaussian ] with @xmath106 and @xmath107 .
it is evident on fig.[norm1 ] that on the long time scale there is a remarkable correspondence between the results of eq.[normfp ] and eq.[normhqm ] . on short times
there is deviation which results from the fact that the decay of the resonances starts at time @xmath5 whereas for an hermitian wp it will take time to reach the boundary of the interaction region .
the idea is simple . using the nh f - product formalism the resonance states decay at any time including the extremely short time regime . in conventional propagation calculations the initial wp oscillates in between the barriers and a significant part of the tunnelling takes place when the oscillating wp hits the inner classical turning points of the potential barriers .
therefore the deviation between the results obtained from nh - qm and conventional qm calculations is during the time it takes for the initial wp to reach the inner classical turning points . to further illustrate this if we study a narrow wp with @xmath108 , which is fully localized between the barriers inside the potential in fig .
[ jola - cs ] it is obvious from the above argument that in hermitian qm it will take the wp some time to escape out of the barriers while in nh - qm decay starts immediately at @xmath5 ( see fig[norm2 ] ) .
when one studies the behavior of different wp s , the best choice for @xmath76 in eq.[normhqm ] varies and depends on the wp but it is always in the vicinity of the top of the barriers in fig.[jola - cs ] .
this can be understood again by realizing that the continuum wavefunctions under the top of the barrier are localized outside of the potential while the resonance states are localized inside the potential ( even over the barrier ) , therefore , in nh - qm we have a clear distinction between the `` system '' and its `` surrounding '' .
+ in previous nuclear physics studies of the decay of narrow isolated resonances which are associated with the nh effective hamiltonian given in eq.[feshham ] the decay law has been obtained by the conventional scalar product @xcite .
namely , following this approach the effective decay rate is given by : @xmath109 where @xmath110 is the solution of eq.[nh - tdse ] which can be expanded in the basis set of the eigenfunctions of eq.[phi_r ] , @xmath111 ( with corresponding eigenvalues @xmath56 ) , such that @xmath112 .
using this approach one gets that , @xmath113\left
< \phi^r_{\alpha ' } | \phi^r_\alpha \right > \quad,\label{norm_np}\ ] ] where the notation @xmath114 stands for the formalism used in nuclear physics in ref.@xcite . when the resonances are not isolated there is a deviation from the effective decay rate due to interference between the different states
when trying to apply this approach to our simple test model , one fails to get converged results and moreover the behavior of @xmath115 based on eq.[norm_np ] , as can be seen in fig.[norm1 ] , is quite irregular as it oscillates instead of constantly decaying .
this shows that the formalism depicted in eq.[norm_np ] can only be used in specific cases whereas the fp formalism as portrayed in eq .
[ normfp ] should be used in general .
+ let us now test the fp formalism for calculating time dependent expectations values for nh hamiltonians . if one wishes to measure an observable quantity @xmath116 which will be defined in nh - qm , within the fp formalism , by : @xmath117 where in principle @xmath118 can get complex values . in such a case the phase of @xmath118 should be a measurable quantity ( and therefore should be @xmath88 independent although both @xmath119 and @xmath120 vary with @xmath88 ( the rotational angle associated with the cs transformation ) . in our case ( i.e.
, the hamiltonian is nh due to the cs similarity transformation . )
@xmath121 is the scaled operator for the desired observable .
as we have seen in fig s [ norm1],[norm2 ] the nh formalism describes the part of the wp which remains in the interaction region .
thus , we expect to find correspondence to a quantity similar to that defined in eq.[normhqm ] .
we will label @xmath122 as an observable quantity in the interaction region calculated by the conventional ( hermitian ) quantum - mechanics approach .
@xmath123 where @xmath124 is the solution obtained by solving the conventional tdse . returning to our one - dimensional test model ,
the results for the observable @xmath125 of the average position for a wp placed at @xmath126 with width , @xmath106 , and initial momentum @xmath127 based on eq.s [ fpobs],[irobs ] are given in fig.[avex ] .
once again there is very good correspondence on the long time scale while on the short time scale there are deviations .
this suggests that the nh formalism which describes the resonance states applies on times longer than some initial rearrangement time in which the resonance states are populated .
+ since the scaling parameter is not associated with any physical quantity , time dependent observables should get real values in spite of the analytical continuation of the hamiltonian . indeed , the results presented in fig.[xp - phase ] show very clearly that at any given time the mean position , @xmath128,(@xmath129 in eq.[fpobs ] ) , and mean momentum @xmath130,(@xmath131 in eq.[fpobs ] ) are real quantities even though they have been obtained by nh calculation based on eq.[fpobs ] .
the phase of both the average position and momentum is zero with abrupt shifts in @xmath132 every time an observable vanishes ( and thus ca nt be measured at that given time ) .
the jumps in @xmath132 of the phase of @xmath118 ( @xmath128 or @xmath130 ) result from the fact that the observable changes its sign from positive to negative or vise - versa and therefore would appear also in standard qm calculations .
we have shown that using the recently introduced fp formalism it is possible to overcome the time - asymmetry problem in nh - qm and to propagate a wp which populates many resonance states .
this formalism describes the dynamics without the need to separate the entire space into `` system '' and `` surrounding '' which is necessary when using hermitian qm .
the nh description of the system is valid on the timescales which are long enough to regard only the localized part of the wp after the scattered part of the wp has left the interaction region .
it has also been shown that observables such as position and momentum in the f - product formalism obtain real values despite the nh nature of the hamiltonian .
this reasserts the validity of this formalism which makes it applicable to systems where particles are temporarily trapped in the interaction region .
the application of these novel approach to many electron systems is under current study .
a. bohm , phys .
a , * 60 * , 861 ( 1999 ) ; a. bohm and n.l .
harshman , lecture notes in physics , * 504 * , 181 ( 1998 ) ; a. bohm , adv .
* 122 * , 301(2002 ) ; a. bohm et .
al . , fort .
der phys . , * 51 * , 551(2003 ) . c. a. nicolaides , lecture notes in physics , * 622 * , 357 ( 2003 ) ; c. a. nicolaides and d. r. beck , int .
j. quantum chem .
, * 14*,457 ( 1978 ) ; c. a. nicolaides , int . j. quantum chem .
, * 89 * , 94 ( 2002 ) and references therein .
n. moiseyev , p.r .
certain , and f. weinhold , mol .
phys . , * 36 * , 1613 ( 1978 ) ; n. lipkin , n. moiseyev , and e. brndas , phys . rev . a , * 40*,549 ( 1989);n .
moiseyev , mol .
phys.,*47*,585 ( 1982 ) ; h. j. korsch , h. laurent , r. mhlenkamp , mol .
phys . , * 43 * , 1441 ( 1981 ) ; m. rittby , n. elander , e. brndas , phys . rev .
a , * 24 * , 1636 ( 1981 ) . | the solutions of the time independent schrdinger equation for non - hermitian ( nh ) hamiltonians have been extensively studied and calculated in many different fields of physics by using @xmath0 methods that originally have been developed for the calculations of bound states .
the existing non - hermitian formalism breaks down when dealing with wavepackets(wp ) .
an open question is how time dependent expectation values can be calculated when the hamiltonian is nh ? using the f - product formalism , which was recently proposed , [ j. phys .
chem .
, * 107 * , 7181 ( 2003 ) ] we calculate the time dependent expectation values of different observable quantities for a simple well known study test case model hamiltonian .
we carry out a comparison between these results with those obtained from conventional(i.e . ,
hermitian ) quantum mechanics ( qm ) calculations .
the remarkable agreement between these results emphasizes the fact that in the nh - qm , unlike standard qm , there is no need to split the entire space into two regions ; i.e. , the interaction region and its surrounding .
our results open a door for new type of wp propagation calculations within the nh - qm formalism that until now were impossible .
in particular our work is relevant to the many different fields in physics and chemistry where complex absorbing potentials are introduced in order to reduce the propagation calculations into a restricted region in space where the artificial reflections from the edge of the numerical grid / box are avoided .
department of chemistry and minerva center for nonlinear physics of complex systems , technion
israel institute of technology , haifa 32000 , israel . |
the relativistic contribution to the rate of precession of perihelion of mercury is calculated accurately using general relativity @xcite .
however , the problem is commonly discussed in undergraduate and graduate classical mechanics textbooks , without introduction of an entirely new , metric theory of gravity .
one approach @xcite is to define a lagrangian that is consistent with both newtonian gravity and the momentum - velocity relation of special relativity .
the resulting equation of motion is solved perturbatively , and an approximate rate of precession of perihelion of mercury is extracted .
this approach is satisfying in that all steps are proved , and a brief introduction to special relativity is included . on the other hand , one must be content with an approximate rate of precession that is about one - sixth the correct value .
another approach @xcite is that of a mathematical exercise and history lesson .
a modification to newtonian gravity is given , without proof , resulting in an equation of motion that is the same as that derived from general relativity .
the equation of motion is then solved using appropriate approximations , and the correct rate of precession of perihelion of mercury is extracted .
both approaches provide an opportunity for students of classical mechanics to understand that relativity is responsible for a small contribution to perihelic precession and to calculate that contribution .
we present a review of the approach using only special relativity , followed by an alternative solution of the equation of motion derived from lagrange s equations
. an approximate rate of perihelic precession is derived that agrees with established calculations .
this effect arises as one of several small corrections to kepler s orbits , including reduced radius of circular orbit and increased eccentricity .
the method of solution makes use of coordinate transformations and the correspondence principle , rather than the standard perturbative techniques , and is approachable by undergraduate physics majors . a relativistic particle of mass @xmath0 orbiting a central mass @xmath1 is commonly described by the lagrangian @xcite @xmath2 where : @xmath3 ; @xmath4 ; and @xmath5 .
( @xmath6 is newton s universal gravitational constant , and @xmath7 is the speed of light in vacuum . )
the equations of motion follow from lagrange s equations , @xmath8 for each of @xmath9 , where @xmath10 .
the results are : @xmath11 = 0 , \label{eq_angmom}\ ] ] which implies that @xmath12 ; and @xmath13 the first of these [ eq . ( [ eq_angmom ] ) ] is the relativistic analogue to the newtonian equation for the conservation of angular momentum per unit mass , and is used to eliminate @xmath14 in eq .
( [ eq_eom ] ) , @xmath15 time is eliminated by successive applications of the chain rule , together with the conserved angular momentum @xcite : @xmath16 and , therefore , @xmath17 substituting eqs .
( [ eq_part1 ] ) and ( [ eq_part2 ] ) into the equation of motion eq .
( [ eq_eom ] ) results in @xmath18 we anticipate a solution of eq .
( [ eq_eom2 ] ) that is near keplerian and introduce the radius of a circular orbit for a nonrelativistic particle with the same angular momentum , @xmath19 .
the result is @xmath20 where @xmath21 is a velocity - dependent correction to newtonian orbits due to special relativity .
the conic - sections of newtonian mechanics @xcite are recovered by setting @xmath22 @xmath23 : @xmath24 which implies that @xmath25 where @xmath26 is the eccentricity .
the planets of our solar system are described by near - circular orbits @xmath27 and require only small relativistic corrections @xmath28 .
mercury has the largest eccentricity @xmath29 , and the next largest is that of mars @xmath30 .
therefore , @xmath31 [ defined after eq .
( [ eq_sr ] ) ] is taken to be a small relativistic correction to near - circular orbits of newtonian mechanics ( keplerian orbits ) . neglecting the radial component of the velocity , which is small compared to the tangential component , and expanding @xmath32 to first order in @xmath33 results in @xmath34 ( see sec .
[ sec_discussion ] and app .
[ app_series ] for a thorough discussion of this approximation . ) once again using the angular momentum to eliminate @xmath14 , this may be expressed as @xmath35 , or @xmath36 the equation of motion eq .
( [ eq_sr ] ) is now expressed approximately as @xmath37 where @xmath38 .
the conic - sections of newtonian mechanics , eq . ( [ eq_newton0 ] ) and eq . ( [ eq_newton ] ) , are now recovered by setting @xmath39 @xmath23 .
the solution of eq .
( [ eq_sr2 ] ) for @xmath40 approximately describes keplerian orbits with small corrections due to special relativity . if @xmath41 is taken to be a small relativistic correction to keplerian orbits , it is convenient to make the change of variable @xmath42 .
the last term on the right - hand - side of eq .
( [ eq_sr2 ] ) is then approximated as @xmath43 , resulting in a linear differential equation for @xmath44 : @xmath45 the additional change of variable @xmath46 results in the familiar form : @xmath47 where @xmath48 .
the solution is similar to that of eq .
( [ eq_newton0 ] ) : @xmath49 where @xmath50 is an arbitrary constant of integration . in terms of the original coordinates ,
( [ eq_sr4 ] ) becomes @xmath51 where @xmath52 according to the correspondence principle , kepler s orbits , eq .
( [ eq_newton ] ) , must be recovered in the limit @xmath53 @xmath23 , so that @xmath54 is the eccentricity of newtonian mechanics .
to first order in @xmath41 , eqs .
( [ eq_s_coeff_r0])([eq_s_coeff_phi0 ] ) become @xmath55 so that relativistic orbits in this limit are described concisely by @xmath56 this approximate orbit equation has the same form as that derived from general relativity @xcite , @xmath57 and clearly displays three characteristics : precession of perihelion ; reduced radius of circular orbit ; and increased eccentricity .
the approximate orbit equation eq .
( [ eq_class_rel ] ) predicts a shift in the perihelion through an angle @xmath58 per revolution .
this prediction is identical to that derived using the standard approach @xcite to incorporating special relativity into the kepler problem , and is compared to observations assuming that the relativistic and keplerian angular momenta are approximately equal . for a keplerian orbit @xcite @xmath59 , where @xmath60m@xmath61kg@xmath62s@xmath63 , @xmath64 kg is the mass of the sun , and @xmath65 and @xmath26 are the semi - major axis and eccentricity of the orbit , respectively .
therefore , the relativistic correction defined after eq .
( [ eq_sr2 ] ) , @xmath66 is largest for planets closest to the sun and for planets with very eccentric orbits . for mercury @xcite @xmath67 m and @xmath68 , so that @xmath69 . ( the speed of light is taken to be @xmath70 . ) according to eq .
( [ eq_s_def_precess ] ) , mercury precesses through an angle @xmath71 per revolution .
this angle is very small and is usually expressed cumulatively in arc seconds per century .
the orbital period of mercury is 0.24085 terrestrial years , so that @xmath72 the general relativistic ( gr ) treatment results in a prediction of 43.0arcsec / century @xcite , and agrees with the observed precession of perihelia of the inner planets @xcite .
historically , this contribution to the precession of perihelion of mercury s orbit precisely accounted for the observed discrepancy , serving as the first triumph of the general theory of relativity @xcite .
the present approach , using only special relativity , accounts for about one - sixth of the observed discrepancy , eq .
( [ eq_prec_num ] ) .
precession due to special relativity is illustrated in fig .
[ fig1 ] . the approximate relativistic orbit equation eq .
( [ eq_class_rel ] ) [ or eq .
( [ eq_s_eoo ] ) together with eqs .
( [ eq_s_coeff_r])([eq_s_coeff_phi ] ) ] predicts that a relativistic orbit in this limit has a reduced radius of circular orbit @xmath73 .
this characteristic is not discussed in the standard approach to incorporating special relativity into the kepler problem , but is consistent with the gr description .
an effective potential naturally arises in the gr treatment of the central - mass problem @xcite , @xmath74 that reduces to the newtonian effective potential in the limit @xmath75 . in the keplerian limit
, the gr angular momentum per unit mass @xmath76 is also taken to be approximately equal to that for a keplerian orbit @xcite .
minimizing @xmath77 with respect to @xmath78 results in the radius of a stable circular orbit , @xmath79 so that the radius of circular orbit is predicted to be reduced : @xmath80 ( there is also an unstable circular orbit .
see fig .
[ fig2 ] . )
this reduction in radius of circular orbit is six times that predicted by the present treatment using only special relativity , for which @xmath81 [ see eq .
( [ eq_s_coeff_r ] ) . ]
most discussions of the gr effective potential eq .
( [ eq_schwarz_eff_pot ] ) emphasize relativistic capture , rather than reduced radius of circular orbit .
the @xmath82 term in eq .
( [ eq_schwarz_eff_pot ] ) contributes negatively to the effective potential , resulting in a finite centrifugal barrier and affecting orbits very near the central mass ( large - velocity orbits ) .
( see fig .
[ fig2 ] . )
this purely gr effect is not expected to be described by the approximate orbit equation eq .
( [ eq_class_rel ] ) , which is derived using only special relativity and assumes orbits very far from the central mass ( small - velocity orbits ) .
an additional characteristic of relativistic orbits is that of increased eccentricity .
equation ( [ eq_class_rel ] ) predicts that a relativistic orbit will have increased eccentricity , when compared to a keplerian orbit with the same angular momentum : @xmath83 .
[ see eq .
[ eq_s_coeff_e ] .
] this characteristic of relativistic orbits also is not discussed in the standard approach to incorporating special relativity into the kepler problem , but is consistent with the gr description . the gr orbit equation in this keplerian limit
( [ eq_gen_rel ] ) predicts an increase in eccentricity @xmath84 , which is six times that predicted by the present treatment using only special relativity .
the approximate orbit equation in eq .
( [ eq_class_rel ] ) provides small corrections to kepler s orbits due to special relativity .
[ compare eqs .
( [ eq_class_rel ] ) and ( [ eq_newton ] ) . ] a systematic verification may be carried out by substituting eq .
( [ eq_class_rel ] ) into eq .
( [ eq_sr2 ] ) , keeping terms of orders @xmath26 , @xmath41 , and @xmath85 only .
the domain of validity is expressed by subjecting the solution eq .
( [ eq_class_rel ] ) to the condition @xmath86 for the smallest value of @xmath78 . evaluating the orbit equation eq .
( [ eq_class_rel ] ) at the perihelion @xmath87 results in @xmath88 the substitution of this result into eq .
( [ eq_dov ] ) results in the domain of validity : @xmath89 therefore , the relativistic eccentricity @xmath90 , and eq .
( [ eq_class_rel ] ) is limited to describing relativistic corrections to near - circular ( keplerian ) orbits .
also , the relativistic correction @xmath91 , and thus the orbit equation eq .
( [ eq_class_rel ] ) is valid only for small relativistic corrections . in sec .
[ sec_keplerianlimit ] the relativistic correction to keplerian orbits @xmath31 [ defined after eq .
( [ eq_sr ] ) ] is approximated by : neglecting the radial component of the velocity , @xmath92 ; and keeping terms only up to first order in the expansion @xmath93 . neglecting the radial component of the velocity in the relativistic correction @xmath31 is consistent with the assumption of near - circular , approximately keplerian orbits , and is complementary to the assumption @xmath94 preceding eq .
( [ eq_linearized ] ) .
it is emphasized that the radial component of the velocity is neglected only in the relativistic correction @xmath31 ; it is not neglected in the derivation of the relativistic equation of motion eq .
( [ eq_sr ] ) . that there is no explicit appearance of @xmath95 in the relativistic equation of motion eq .
( [ eq_sr ] ) , other than in the definition of @xmath32 , is due to a fortunate cancellation after eq .
( [ eq_part2 ] ) .
furthermore , the approximate orbit equation eq .
( [ eq_class_rel ] ) has the same form as that [ eq .
( [ eq_gen_rel ] ) ] arising from the gr description , in which the radial component of the velocity is not explicitly neglected .
a first - order series approximation for @xmath32 is consistent with the keplerian limit .
interestingly , however , the problem is soluble without truncating the series .
( see appendix [ app_series ] . )
this more elaborate derivation yields diminishing returns due to the complementary assumption @xmath94 , which implicitly constrains the velocity to be much smaller than the speed of light .
the resulting approximate orbit equation [ eq . ( [ eq_ssr ] ) ] , @xmath96 is almost identical to that derived using a much simpler approach in sec .
[ sec_keplerianlimit ] [ eq .
( [ eq_class_rel ] ) ] .
although this alternative orbit equation predicts a relativistic correction to eccentricity that is nearly one - half that of the gr result , it lacks the symmetry of the gr orbit equation eq .
( [ eq_gen_rel ] ) , and it does not provide significant further qualitative understanding of relativistic corrections to keplerian orbits .
the present approach to incorporating special relativity into the kepler problem results in an approximate orbit equation [ eq . ( [ eq_class_rel ] ) ] that has the same form as that derived from general relativity in this limit [ eq . ( [ eq_gen_rel ] ) ] and is easily compared to that describing kepler s orbits [ eq .
( [ eq_newton0 ] ) ] .
this orbit equation clearly describes three corrections to a keplerian orbit due to special relativity : precession of perihelion ; reduced radius of circular orbit ; and increased eccentricity .
the predicted rate of precession of perihelion of mercury is identical to established calculations using only special relativity .
each of these corrections is exactly one - sixth of the corresponding correction described by general relativity in the keplerian limit .
this derivation of an approximate orbit equation is complementary to existing calculations of the rate of precession of perihelion of mercury using only special relativity .
the central - mass problem is described by a lagrangian that is consistent with both newtonian gravity and the momentum - velocity relation of special relativity .
however , coordinate transformations and the correspondence principle are used to solve the equations of motion resulting from lagrange s equations , rather than the standard perturbative methods .
the resulting closed - form , approximate orbit equation exhibits several characteristics of relativistic orbits at once , but is limited to describing small relativistic corrections to approximately newtonian , near - circular orbits .
this orbit equation , derived using only special relativity , provides a qualitative description of corrections to keplerian orbits due to general relativity .
exact solutions to the special relativistic kepler problem require a thorough understanding of special relativistic mechanics @xcite and are , therefore , inaccessible to most undergraduate physics majors .
the present approach and method of solution is understandable to nonspecialists , including undergraduate physics majors whom have not had a course dedicated to relativity . for near - circular orbits
the radial component of the velocity is small compared to the tangential component and is neglected in the relativistic correction @xmath97 .
[ see eq .
( [ eq_sr ] ) . ] an exact infinite - series representation , @xmath98 is used , rather than the approximate first - order truncated series used in sec .
[ sec_keplerianlimit ] .
( the remainder of this derivation parallels that of sec .
[ sec_keplerianlimit ] . ) conservation of angular momentum is used to eliminate @xmath14 , resulting in @xmath99 } - 1 , \label{eq_series_lambda}\ ] ] in terms of which the equation of motion eq .
( [ eq_sr ] ) is now expressed approximately as @xmath100 the conic - sections of newtonian mechanics , eqs .
( [ eq_newton0 ] ) and ( [ eq_newton ] ) , are recovered by setting @xmath75 . for near - circular ,
approximately keplerian orbits it is convenient to make the change of variable @xmath101 , so that @xmath102 . in terms of this new variable , the relativistic correction eq .
( [ eq_series_lambda ] ) is @xmath103 where @xmath104 , and the following series is identified : @xmath105 thus , the equation of motion eq .
( [ eq_sr5 ] ) is linearized : @xmath106 the additional change of variable @xmath107 results in the familiar form : @xmath108 where @xmath109 .
the solution is similar to that of eq .
( [ eq_newton0 ] ) : @xmath110 where @xmath50 is an arbitrary constant of integration . in terms of the original coordinates ,
( [ eq_sr6 ] ) becomes @xmath111 where ( including first - order approximations ) @xmath112 according to the correspondence principle , kepler s orbits [ eq .
( [ eq_newton ] ) with @xmath113 must be recovered in the limit @xmath53 @xmath23 , so that @xmath54 is the eccentricity of newtonian mechanics .
therefore , relativistic orbits in this limit are described concisely by @xmath114 this alternative orbit equation differs from that [ eq . ( [ eq_class_rel ] ) ] derived using the much simpler approach in sec .
[ sec_keplerianlimit ] only in the relativistic correction to eccentricity .
see the discussion at the end of sec .
[ sec_discussion ] .
a. einstein , `` explanation of the perihelion motion of mercury from the general theory of relativity , '' in _ the collected papers of albert einstein _ , translated by a. engel ( princeton university press , princeton , 1997 ) , vol .
this article is the english translation of ref . .
k. schwarzschild , `` ber das gravitationsfeld eines massenpunktes nach der einsteinschen theorie , '' sitzungsber .
, phys .- math .
* 1916 * , 189196 ( 1916 ) .
reprinted in translation as `` on the gravitational field of a mass point according to einstein s theory , '' arxiv : physics/9905030v1 [ physics.hist-ph ] .
j. droste , `` het veld van een enkel centrum in einstein s theorie der zwaartekracht , en de beweging van een stoffelijk punt in dat veld , '' versl .
* 25 * , 163180 ( 1916 - 1917 ) .
reprinted in translation as `` the field of a single centre in einstein s theory of gravitation , and the motion of a particle in that field , '' proc .
. wetensch .
* 19 * ( 1 ) , 197215 ( 1917 ) , link:<adsabs.harvard.edu / abs/1917knab ... 19 .. 197d>[<adsabs.harvard.edu / abs/1917knab ... 19 .. 197d > ] .
a. m. nobili and i. w. roxburgh , `` simulation of general relativistic corrections in long term numerical integrations of planetary orbits , '' in _ relativity in celestial mechanics and astrometry : high precision dynamical thoeries and observational verifications _ , edited by j. kovalevsky and v. a. brumberg ( iau , dordrecht , 1986 ) , pp .
105110 , link:<adsabs.harvard.edu / abs/1986iaus .. 114 .. 105n>[<adsabs.harvard.edu / abs/1986iaus .. 114 .. 105n > ] .
n. ashby , `` planetary perturbation equations based on relativistic keplerian motion , '' in _ relativity in celestial mechanics and astrometry : high precision dynamical thoeries and observational verifications _ , edited by j. kovalevsky and v. a. brumberg ( iau , dordrecht , 1986 ) , pp .
4152 , link:<adsabs.harvard.edu / abs/1986iaus .. 114 ... 41a>[<adsabs.harvard.edu / abs/1986iaus .. 114 ... 41a > ] .
d. brouwer and g. m. clemence , `` orbits and masses of planets and satellites , '' in _ the solar system : planets and satellites _ , edited by g. p. kuiper and b. m. middlehurst ( university of chicago press , chicago , 1961 ) , vol .
iii , pp .
3194 . relativistic orbit in a keplerian limit ( solid line ) , as described by eq .
( [ eq_class_rel ] ) , compared to a corresponding keplerian orbit ( dashed line ) [ eq .
( [ eq_newton ] ) ] .
the precession of perihelion is one orbital characteristic due to special relativity and is illustrated here for @xmath115 .
this characteristic is exaggerated by both the choice of eccentricity @xmath116 and relativistic correction parameter @xmath117 for purposes of illustration .
precession is present for smaller ( non - zero ) reasonably chosen values of @xmath26 and @xmath41 as well .
( the same value of @xmath26 is chosen for both curves . ) ] effective potential commonly defined in the newtonian limit to general relativity ( solid line ) [ eq .
( [ eq_schwarz_eff_pot ] ) ] , compared to that derived from newtonian mechanics ( dashed line ) .
the vertical dotted lines identify the radii of circular orbits , @xmath118 and @xmath119 , as calculated using general relativity and newtonian mechanics , respectively .
general relativity predicts [ eq . ( [ eq_s_rc_eff_pot ] ) ] a smaller radius of the circular orbit than that predicted by newtonian mechanics .
the value @xmath120 is chosen for purposes of illustration . ] | beginning with a lagrangian that is consistent with both newtonian gravity and the momentum - velocity relation of special relativity , an approximate relativistic orbit equation is derived that describes relativistic corrections to keplerian orbits .
specifically , corrections to a keplerian orbit due to special relativity include : precession of perihelion , reduced radius of circular orbit , and increased eccentricity .
the prediction for the rate of precession of perihelion of mercury is in agreement with existing calculations using only special relativity , and is one sixth that derived from general relativity .
all three of these corrections are qualitatively correct , though suppressed when compared to the more accurate general - relativistic corrections in this limit .
the resulting orbit equation has the same form as that derived from general relativity and is easily compared to that describing kepler s orbits .
this treatment of the relativistic central - mass problem is complementary to other solutions to the relativistic kepler problem , and is approachable by undergraduate physics majors whom have not had a course dedicated to relativity . |
the detailed investigation of the higgs sector is anticipated as one of the central themes of the accelerator particle physics program after the higgs boson will have been observed , most likely at the lhc .
the study of the higgs profile will have to determine whether the observed particle is indeed responsible for generating the mass of gauge bosons , quarks and charged leptons . a significant _ corpus _ of studies ,
outlining strategies to test the higgs mechanism for gauge bosons and quarks with accuracies down to a few percent , has already been assembled @xcite these rely on the combination of data from the lhc and an @xmath0 linear collider operating at centre - of - mass energies @xmath2 = 0.25 tev - 1.0 tev . in this paper
we discuss the feasibility of a test of the higgs mechanism in the lepton sector , by verifying the scaling @xmath5 , through a precise determination of the @xmath6 coupling in the @xmath7decay , in 3 tev @xmath0 collisions . at the lhc
, the significance of an observation of the @xmath7 decay is estimated to be only @xmath8 3 @xmath9 for 115 gev @xmath10 130 gev , even when combining the cms and atlas data on both the gluon and weak boson fusion production channels for 300 fb@xmath3 of integrated luminosity @xcite .
it further decreases for larger values of @xmath11 .
@xmath0 collisions below 1.0 tev should provide with determination of @xmath12 to a relative statistical accuracy of @xmath8 0.035 for a 120 gev higgs boson but , at best , with only a signal and not with a precision measurement in the @xmath13 yukawa coupling @xcite .
the @xmath14 fusion process , @xmath15 , offers a large higgs cross section , provided high enough @xmath2 energies can be attained , to make this measurement possible . in this paper the detection of the @xmath16 decay in high energy @xmath0 collisions
is studied using full geant-4-based simulation and reconstruction .
the paper is organised as follows . in section 2
we discuss the simulation of the @xmath15 , @xmath16 process and of the relevant physics and machine - induced background .
section 3 describes the detector simulation and section 4 has the results .
higgs boson production in @xmath0 collisions proceeds through two main processes : the so - called higgstrahlung , @xmath17 , and the @xmath14 fusion , @xmath15 reactions . while the associated production of the higgs boson with a @xmath18 has a number of advantages , including the possibility to tag the presence of the higgs boson independent on its decay modes , @xmath14 fusion reaches the largest cross sections , due to its @xmath19 rise , if large enough centre - of - mass energies can be obtained .
this makes the @xmath14 fusion process of special interest for the study of rare higgs decays , such as @xmath20 at large higgs boson masses , and @xmath16 .
but as the energy increases , the @xmath1 production cross section is more and more peaked along the beam axis . above @xmath8
3 tev , the further increase in production cross section is effectively lost , due to the limited detector acceptance in the very forward region .
the production cross section at 3 tev and 5 tev is respectively 2.3 and 3.0 times larger compared to that at 1 tev .
however , restricting the acceptance to higgs bosons produced at @xmath21 0.99 ( 0.92 ) , the gains in cross section become 2.1 ( 1.7 ) and 2.4 ( 1.9 ) , respectively .
the possibility of achieving @xmath0 collisions at multi - tev energies , introduced by the clic concept @xcite , offers a real opportunity for observing this decay and measuring @xmath13 with enough accuracy to perform a significant test of the scaling of the yukawa couplings in the lepton sector over a broad range of higgs boson masses @xcite .
multi - tev @xmath0 collisions become important not only for exploring the highest energy frontier but also for precision studies of suppressed higgs decays .
signal events @xmath22 , @xmath16 are generated with pythia 6.205 @xcite , including beamsstrahlung effects , at several higgs boson mass values from 120 gev up to 160 gev .
cross sections are computed using comphep 4.4.0 @xcite and the higgs decay branching fractions with hdecay @xcite . at @xmath2 = 3 tev ,
the effective @xmath15 production cross section , accounting for initial state radiation , is 0.48 pb to 0.45 pb and the decay branching fraction , br(@xmath16 ) = 2.56@xmath2310@xmath24 to 6.47@xmath2310@xmath25 , for @xmath11 in the range from 120 gev to 150 gev .
the irreducible @xmath26 background is generated with comphep 4.4.0 interfaced with pythia 6.205 .
the total cross section of events with an invariant mass of the @xmath27 system , @xmath28 , in the range 100 gev @xmath29 400 gev is 5.31 fb .
we assume to operate clic at centre of mass energy of 3 tev for a total integrated luminosity of 5 ab@xmath3 , which corresponds to @xmath8 eight years ( 1 yr = 10@xmath30 s ) of operation at its nominal luminosity of 6.5@xmath2310@xmath31 @xmath32 s@xmath3 .
the clic anticipated parameters are given in @xcite , for this analysis we use the luminosity spectrum as in @xcite .
this provides with @xmath8 2.3@xmath2310@xmath33 @xmath15 events , 620 to 150 of which are followed by the @xmath16 decay , depending on the value of the higgs bosons mass in the range from 120 gev to 150 gev .
the machine induced background , which affects this analysis most , is due to parasitic @xmath4 events occurring within the same time - stamp as a physics event .
the anticipated average number of such events for the clic parameters is 3.2 bx@xmath3 , or 161 events within a 25 ns time - stamp .
@xmath34 collisions are simulated using the guineapig program @xcite and the cross sections parametrised according to @xcite .
the hadronic final states are simulated using pythia and overlayed to a signal event .
the average visible energy deposited in the detector by @xmath4 events is @xmath8 90 gev / bx .
charged particles are reconstructed in a multi - layered main tracker , made of high resolution silicon strip detectors , and a precision vertex tracker in a 5 t solenoidal field .
discrete si tracking , following the example of the cms experiment at lhc , emerged as an appealing solution for clic , where local hit occupancy and two - track separation may limit the applicability of gaseous trackers @xcite .
a large magnetic field is beneficial for ensuring an excellent momentum resolution for high @xmath35 particle tracks , as well as for confining low momentum charged particles from machine - induced backgrounds to small radii .
this analysis is based on an implementation of the sid detector @xcite , developed as a concept for the international linear collider ( ilc ) project , for the reconstruction of 3 tev @xmath0 events .
the sid concept has a five - layered si main tracker which ensures a momentum resolution @xmath36 5@xmath2310@xmath25 and polar angle coverage down to 8.3@xmath37 .
the following assumptions on the detector response are made : the detection efficiency is 0.98 per layer , the point resolution 7 @xmath38 m , which includes alignment effects , and the detector is capable of time - stamping hits with 25 ns resolution , corresponding to 50 bunch crossings ( bx ) .
while the sid is designed for the reconstruction of events at centre - of - mass energies below 1 tev , its geometry is well suited for the study of simple events , as those of our signal , up to multi - tev energies .
full event simulation is performed using the sid model implemented in the mokka 06 - 03 program @xcite based on geant-4 @xcite .
simulation results are saved in the lcio format @xcite and used as input to the event reconstruction .
this is performed using a set of dedicated processors implemented in the marlin reconstruction and analysis framework @xcite .
the reconstruction starts from hits generated in the sensitive layers along the track .
the track pattern recognition is performed from the outermost main tracker hits and extends inwards to the vertex tracker . after the first stage of pattern recognition hit bundles
are fitted using a simple helix .
a second stage pattern recognition is performed using hits left unassociated after the first stage .
reconstructed particle tracks are requested to have more than 4 hits and associated hits on at least 70 % of the number of sensitive surfaces traversed , depending on their polar angle , @xmath39 , @xmath40 fit probability in excess of 0.01 , @xmath41 0.98 , and impact parameter significance below 3.5 times the extrapolation resolution , in both coordinates .
these cuts remove poorly measured particle tracks .
the momentum resolution of the reconstructed muon tracks is @xmath42 = ( 3.46 @xmath43 0.10)@xmath2310@xmath25 , in agreement with the sid specifications .
events with two , oppositely charged muon candidate tracks with momentum exceeding 10 gev and below 750 gev and event energy in charged particles 100 gev @xmath44 1.5 tev are selected .
the invariant mass distribution of the selected di - muon pairs is shown in figure [ fig : mass ] .
the distribution is fitted with a gaussian curve to describe the signal and an exponential background .
the mass resolution @xmath45 for signal events is 0.0034 , which corresponds to 0.40 gev for @xmath11=120 gev .
the signal region is defined by a @xmath43 2 @xmath46 interval around the fitted mass peak position and a binned @xmath40 fit is performed to extract the number of signal and background events .
results are summarised in table [ tab : results ] .
@lcccc @xmath11 ( gev ) & nb
. signal evts . &
. bkg .
s/@xmath47 & @xmath48 + 120 & 229.6 & 161.1 & 18.1 & 0.086 + 130 & 153.1 & 88.1 & 16.3 & 0.101 + 140 & 103.2 & 64.3 & 12.9 & 0.125 + 150 & 68.1 & 58.1 & 9.5 & 0.160 + 155 & 68.1 & 58.0 & 5.2 & 0.253 + 160 & 12.1 & 33.0 & 2.1 & + we obtain @xmath49 = 0.086 and 0.160 for @xmath11 = 120 gev and 150 gev , respectively .
these results correspond to a determination of @xmath6 with a relative accuracy of 0.04 to 0.08 .
a significant signal for the decay can be obtained up to @xmath11 = 155 gev , when its branching fraction is only 4@xmath2310@xmath25 , due to the rapid turn on of the @xmath14 decay channel of the higgs boson .
the analysis is repeated by overlaying the @xmath4 background on signal events , for @xmath11 = 120 gev/@xmath50 .
no degradation of the reconstruction efficiency and resolution is observed for the amount of hadronic background corresponding to 50 bx ( see figure [ fig : comp ] ) .
the determination of the muon yukawa coupling @xmath6 in @xmath0 collisions at centre - of - mass energies of 3 tev at clic has been studied through the @xmath15 , @xmath16 process using full simulation and event reconstruction .
the process is observable for higgs mass values up to 155 gev and the @xmath6 yukawa coupling can be determined with a relative statistical accuracy of 0.04 to 0.08 for masses from 120 gev to 150 gev with an integrated luminosity of 5 ab@xmath3 .
the superposition of machine - induced @xmath4 background corresponding to 50 bunch crossings overlayed on a single @xmath51 event does not affect these results .
99 m. battaglia and k. desch , in _ physics and experiments with future linear @xmath0 colliders _ ,
a. para and h.e .
fisk ( editors ) , aip conference proceedings 578 , 2001 , 163 [ arxiv : hep - ph/0101165 ]
. t. l. barklow , in _ proc . of the international conference on linear colliders _ h. videau and j .- c .
brient ( editors ) , paris 2005 , vol . 1 , 127 [ arxiv : hep - ph/0411221 ] . t. han and b. mcelrath , phys . lett .
b * 528 * ( 2002 ) 81 [ arxiv : hep - ph/0201023 ] .
r. w. assmann _ et al .
_ , cern-2000 - 008 . m. battaglia and a. de roeck , in _ proc .
of the aps / dpf / dpb summer study on the future of particle physics ( snowmass 2001 ) _ ed .
n. graf , _ in the proceedings of aps / dpf / dpb summer study on the future of particle physics ( snowmass 2001 ) , snowmass , colorado , 30 jun - 21 jul 2001 , pp e3066 _ [ arxiv : hep - ph/0111307 ] . t. sjostrand , p. eden , c. friberg , l. lonnblad , g. miu , s. mrenna and e. norrbin , comput .
phys . commun .
* 135 * ( 2001 ) 238 [ arxiv : hep - ph/0010017 ] .
e. boos _ et al .
_ [ comphep collaboration ] , nucl .
instrum .
a * 534 * ( 2004 ) 250 [ arxiv : hep - ph/0403113 ] .
a. djouadi , j. kalinowski and m. spira , comput .
commun .
* 108 * ( 1998 )
56 [ arxiv : hep - ph/9704448 ] . h. braun _ et al .
_ , clic note 627 ( 2006 ) .
s. agostinelli _ et al .
_ , nucl .
instrum .
a * 506 * ( 2003 ) , 250 .
f. gaede , t. behnke , n. graf and t. johnson , _ in the proceedings of 2003 conference for computing in high - energy and nuclear physics ( chep 03 ) , la jolla , california , 24 - 28 mar 2003 , pp tukt001 _ [ arxiv : physics/0306114 ] . | multi - tev @xmath0 collisions provide with a large enough sample of @xmath1 bosons to enable measurements of its suppressed decays .
results of a detailed study of the determination of the muon yukawa coupling at @xmath2 = 3 tev , based on full detector simulation and event reconstruction , are presented .
the muon yukawa coupling can be determined with a relative accuracy of 0.04 to 0.08 for higgs bosons masses from 120 gev to 150 gev , with an integrated luminosity of 5 ab@xmath3 .
the result is not affected by overlapping @xmath4 background . |
strongly correlated two - dimensional quantum coulomb systems are the subject of intensive discussions . in particular , it is known that the competition between electrostatic and kinetic energy in an electron gas may be the reason of an unusual phase diagram of a 2d system of electrons .
the liquid state of such system is stable when the kinetic energy dominates while the electrostatically favored `` wigner '' triangular crystal is stable in the opposite case .
if there is a strong competition between these two kinds of energy , different situations are possible .
the question under discussion is the existence of the intermediate anisotropic liquid phase ( hexatic ) under melting of crystal into isotropic liquid .
moreover , the physical mechanism of melting can be influenced by the interaction with substrate and defects .
currently all mentioned phenomena can be extensively investigated by a consistent first - principle numerical simulation , and in this brief paper we present the most interesting results of our numerical experiments by the direct pimc method .
let us consider a two - component neutral e - h plasma .
thermodynamic properties of such plasma are defined by the partition function @xmath0 , which for the case of @xmath1 electrons and @xmath2 holes ( @xmath3 ) , is given by @xmath4 , with @xmath5 , where @xmath6 , @xmath7 , @xmath8 , @xmath9 denotes the space and spin electron coordinates , while @xmath10 denotes the space and spin hole coordinates .
these expressions are valid for 1d , 2d and 3d cases .
the pair distribution functions for a binary mixture of quantum electrons and holes can be written in the form @xmath11 , where @xmath12 and @xmath13 label the particle species , i.e. @xmath14 ( electrons ) , @xmath15 ( holes ) and @xmath16 denote the two - dimensional vectors of coordinates , @xmath17 and @xmath18 .
the exact density matrix @xmath19 of a quantum system for low temperature and high density is in general not known but can be constructed using a path integral representation @xcite .
thus we take into account interaction , exchange ( through permutation operators ) and spin effects both for electrons and holes .
this procedure gives the expression for the density matrix as a multiple integral which is well suited for efficient numerical evaluation by monte carlo techniques ,
e.g. @xcite . we mention here that for all obtained results , the maximum statistical and systematic errors is not more than @xmath20 .
we are interested in strong coulomb correlation effects such as bound state ( excitons , bi - excitons , many particle clusters ) , their transformation by the surrounding plasma and their eventual breakup at high densities ( mott effect ) . beyond the mott density
, we expect the possibility of the hole crystallization if the hole mass is sufficiently large @xcite .
below , the density of the two - component plasma is characterized by the brueckner parameter @xmath21 defined as the ratio of the mean distance between particles @xmath22^{1/2}$ ] and the 3d exciton bohr radius @xmath23 , where @xmath24 and @xmath25 are the electron and hole 2d densities .
the dimensionless temperature will be presented as a ratio of the temperature and the 3d electron - hole binding energy ( rydberg ) , which includes the reduced effective mass and dielectric constant .
( black solid line ) , @xmath26 ( red dashed line ) , @xmath27 ( blue dot - dashed line ) at temperature @xmath28 , the hole to electron mass ratio @xmath29 and densities related to the brueckner parameter @xmath30 and @xmath31 .
right column : red and magenta clouds holes , yellow and cyan clouds
electrons with the opposite spin directions.,title="fig : " ] + ( black solid line ) , @xmath26 ( red dashed line ) , @xmath27 ( blue dot - dashed line ) at temperature @xmath28 , the hole to electron mass ratio @xmath29 and densities related to the brueckner parameter @xmath30 and @xmath31 .
right column : red and magenta clouds holes , yellow and cyan clouds
electrons with the opposite spin directions.,title="fig : " ] we analyze some spacial distribution functions and related spin - resolved typical `` snapshots '' of the e - h state in the simulation box for different particle densities , temperatures and hole - to - electron mass ratios @xmath32 . according to the path integral representation of the density matrix ,
each electron and hole is represented by several tens of points ( `` beads '' ) .
the spatial distribution of the beads of each quantum particle is proportional to its spatial probability distribution .
shows that , the typical size of the cloud of beads for electrons is several times larger than the one for the heavy holes . at low temperature and low ( @xmath30 ) and middle ( @xmath31 ) densities
practically all holes are closely covered by electron beads . from a physical point of view
this means that electrons and holes form bound states , i.e. excitons , bi - excitons and many particle clusters .
the existence of the bound states is also supported by the behavior of the pair distribution functions , exhibiting pronounced maxima at the distances of about half and a bit more of the bohr radius .
we note that raising the temperature at a fixed density leads to a temperature - induced ionization of the bound states . as a result we found a substantial number of free electrons and holes . from it follows that the growth of density results in the increase of the number of particles in clusters .
the structural analysis of large many - particle clusters show the hexagonal ordering of heavy holes inside liquid - like clusters ( lower right panel ) . in this panel
there are only two clusters : one is in the center of the monte carlo cell , while the second one is divided in four parts ( in each corner ) due to the periodic boundary conditions of the monte carlo cell . here besides the inner normal hexagonal structure the holes due to the strong coulomb repulsion form the filament - like structures of the clouds of beads at the bounds of clusters .
the filament - like structures are fully 2d topological effect as in 3d case the holes in analogous many particles clusters have normal liquid - like ordering @xcite . if the particle density is high enough ( @xmath33 ) the electron wavelength becomes larger than the mean inter - particle distance @xmath34 and even larger than the size of the monte carlo cell which is seen by the large extension of the clouds of electron beads . for @xmath35 clusters
become unstable because two electrons bounded to neighboring holes start to overlap allowing for electron tunneling from one cluster to the other ( mott effect ) .
since the hole wave - length is significantly smaller than the electron wavelength , it may still be smaller than @xmath34 and the structure of the hole beads resembles a liquid - like or crystal - like state .
if the hole mass exceeds a critical value , the holes may even form a crystal - like structure @xcite ( lower right panel in ) . here the holes form a crystal - like structure , while the electron density demonstrates the bloch oscillations . at this very high density
the type of the hole crystal is influenced by the boundary conditions of the monte carlo cell ( finite size effect ) .
the detailed analysis of the type of the crystal - like structure is possible for a much more bigger number of particles in the monte carlo cell . , the hole - to - electron mass ratio @xmath29 and densities related to the brueckner parameter @xmath36 and @xmath33 .
notations are the same as in .,title="fig : " ] , the hole - to - electron mass ratio @xmath29 and densities related to the brueckner parameter @xmath36 and @xmath33 .
notations are the same as in .,title="fig : " ]
in this paper we have presented a computer simulation analysis of strong coulomb correlations in dense two - dimensional two - component quantum plasmas at low temperatures . in particular , the formation and dissociation of bound states , such as excitons , bi - excitons and many particle clusters , is analyzed and the density - temperature regions of their occurrence is identified .
we have shown that , above the mott point , two - component plasmas with large mass anisotropy show interesting coulomb correlation phenomena : with increasing density holes can undergo a phase transition to a coulomb hexatic - like liquid and to a wigner crystal which are embedded into a degenerate electron gas .
the crystal - like formation in a two - component plasma is possible for large enough hole - to - electron mass ratios @xcite .
v. filinov acknowledges the hospitality of the institut fr theoretische physik und astrophysik of the christian - albrechts - universitt zu kiel and institut fr physik , ernst - moritz - arndt - universitt zu greifswald . | we investigate structures of 2d quantum electron
hole ( e - h ) plasmas by the direct path integral monte carlo method ( pimc ) in a wide range of temperature , density and hole - to - electron mass ratio . our simulation includes a region of appearance and decay of the bound states ( excitons and biexcitons ) , the mott transition from the neutral e - h plasma to metallic like clusters , formation from clusters the hexatic - like liquid and formation of the crystal - like lattice . |
observational studies of galaxy formation and evolution have progressed at a breathtaking pace over the past couple of years .
data from the refurbished hubble space telescope , the keck and other large telescopes are now providing quantitative information on essential properties of the galaxy population number densities , luminosities , colours , morphologies and star formation rates over a large span of cosmic time .
these data are beginning to sketch out an empirical picture of galaxy formation and evolution from redshift @xmath7 to the present .
evolution has now been established and quantified in : _
( i ) _ the neutral hydrogen and metal content of the universe since @xmath7 ( lanzetta , wolfe & turnshek 1995 , storrie - lombardi 1996 , wolfe 1995 , lu 1996 ) ; _ ( ii ) _ the galaxy luminosity function since @xmath8 ( lilly 1995 ; ellis 1996 ) ; _ ( iii ) _ the morphology of field and cluster galaxies since @xmath9 ( e.g. abraham 1996 , dressler 1994 , smail 1997 ) .
the most recent addition to this remarkable list of observational advances is the discovery of a large population of _ actively star - forming galaxies at @xmath0 _
, identified by their redshifted lyman continuum breaks ( steidel 1996a , hereafter s96 ) .
it is this population that we are primarily concerned with in this paper .
one of the earliest windows on the physical processes at play in galaxy formation is provided by studies of primordial gas clouds detected in absorption against background quasars .
the comoving density of neutral hydrogen present in damped lyman - alpha clouds peaks at @xmath0 when it was comparable to the mass in baryons seen in galactic disks today ( storrie - lombardi 1996 ) .
the decline in the abundance of neutral hydrogen clouds seems to be accompanied by a gradual build - up of their metal content ( lu 1996 ) .
a population of bright galaxies is certainly well established by @xmath10 ( lilly 1995 , ellis 1996 , kauffmann , charlot & white 1996 ) . in the cfrs survey of lilly ,
evolution is manifest in a systematic variation of the shape of the luminosity function of blue galaxies and a brightening of their characteristic luminosity with increasing lookback time .
the luminosity function of red galaxies , on the other hand , seems to have changed little over this redshift interval , although the fraction of galaxies that have the colours of passively evolving ellipticals appears to fall to one third of its present day value by @xmath10 ( kauffmann 1996 ) . on the whole
, galaxies appear to be smaller and increasingly irregular at higher redshift ( driver 1995 , glazebrook 1995a , abraham 1996 , smail 1996 , odewahn 1996 , pascarelle 1996 , lowenthal 1997 . ) for example , the class of `` irregular / merger '' galaxies which are relatively rare at bright magnitudes makes up about a third to a half of all galaxies with @xmath11 .
the median redshift at this apparent magnitude ( the faintest at which automated morphological classification is possible on high resolution hst images ) is @xmath9 .
similarly , the fraction of spirals in rich clusters at @xmath12 is higher than in present day clusters ( dressler 1994 ) .
all these studies leave little doubt that the galaxy population has evolved significantly since @xmath10 .
the discovery of a large population of lyman - break galaxies at @xmath13 provides the first opportunity for statistical studies of evolutionary processes in galaxies beyond @xmath10 .
steidel and collaborators ( steidel & hamilton 1992 , 1993 ; steidel , pettini & hamilton 1995 ; s96 ) searched for high redshift galaxies by selecting objects in a colour - colour plane constructed from images in customised @xmath14 , @xmath15 and @xmath16 band filters .
for galaxies in the redshift range @xmath17 , the lyman limit discontinuity passes through the @xmath14 filter .
opacity due to intervening neutral hydrogen increases the strength of the lyman discontinuity regardless of the shape of the intrinsic spectral energy distribution of the galaxy ( madau 1995 ) .
thus , a galaxy in this redshift range will be faint in the @xmath14 band ( thus becoming a `` uv - dropout '' ) and so will have a very red @xmath18 colour , whilst possibly having a blue @xmath19 colour if it is undergoing significant star formation .
a similar strategy has been successfully implemented in the hubble deep field ( hdf ) by steidel ( 1996b ) and madau ( 1996 ) .
the hst @xmath20 filter has a shorter median wavelength than the @xmath14 filter of the ground - based observations , and so colour selected objects in the hdf span a wider range of redshifts from @xmath21 .
follow - up spectroscopy of the uv drop - out candidates by s96 on the keck telecope confirmed that these galaxies lie mostly in the expected redshift range , @xmath22 .
their spectra resemble those of nearby starburst galaxies . from the apparent @xmath16-band magnitude , a dust - free model for the spectral energy distribution in the uv , and an assumption about the initial stellar mass function ( imf ) , s96 inferred star formation rates in these galaxies in the range @xmath23 for a critical density universe , where we have expressed hubble s constant as @xmath24 .
similarly low star formation rates have been inferred by lowenthal ( 1997 ) for 11 galaxies in the hdf at @xmath25 . from the width of saturated interstellar absorption lines ,
s96 inferred tentative one - dimensional velocity dispersions in the range @xmath26 .
they concluded that the lyman - break galaxies they discovered could be the progenitors of the spheroidal components of present day galaxies .
the current state of empirical knowledge on galaxy formation has been nicely summarized by madau ( 1996 ) and madau ( 1996 ) in the form of a `` cosmic star formation history . '' combining a variety of surveys ( including the cfrs and the lyman - break galaxy surveys ) , they derived metal production and star formation rates as a function of redshift , from @xmath27 to @xmath28 .
observed star formation rates over this redshift range are typically a few solar masses per year for individual galaxies .
the integrated star formation rate never differs by more than an order of magnitude over this entire redshift range , although a peak of activity seems to have occured at @xmath29 .
the total amount of metals produced by the observed populations is comparable to the amount of metals seen in massive galaxies today , suggesting that the bulk of the cosmic star formation has now been identified . in this paper
we employ the semi - analytic model developed in a series of earlier papers ( cole 1994 ; heyl 1995 ; baugh 1996a , 1996b ) , to investigate the significance of the lyman - break galaxy population within the context of hierarchical clustering theories of galaxy formation .
we consider the circumstances under which such a population may form and we focus on the connection between these high redshift objects and galaxies seen in various evolutionary stages at lower redshifts .
we use the available data to test in detail our earlier theoretical predictions for the way in which galaxies are built up from small fluctuations in a universe dominated by cold dark matter ( cdm ) . specifically , we test the prediction of white & frenk ( 1991 ) , lacey ( 1993 ) and cole ( 1994 ) that the bulk of the stars present in galaxies today formed relatively recently , with a median redshift of star formation of only @xmath30 .
the results of these comparisons are very encouraging and suggest a general picture of galaxy formation and evolution which is consistent with the expectations from a broad class of hierarchical clustering cosmologies . a more basic attempt to investigate whether the abundance of s96 lyman - break galaxies is consistent with cold dark matter models was recently carried out by mo & fukugita ( 1996 ) .
they used the press & schechter ( 1974 ) formalism to calculate the number density of halos with velocity dispersion in excess of @xmath31 , making simple assumptions about the time required for a galaxy to form in each halo .
they concluded that a range of low - density cobe - normalized cdm models are compatible with the data .
our semi - analytic galaxy formation scheme is briefly reviewed in section [ s : method1 ] , where we discuss our procedure for generating mock catalogues of high redshift galaxies .
the colour selection criteria of s96 are reviewed in section [ s : col ] and the abundance of galaxies that meet these constraints in a variety of cosmological models is given in section [ s : abun ] .
the expected properties of high redshift galaxies masses , star formation rates , sizes , clustering , etc are presented in section [ s : prop ] .
our models predict the entire evolutionary history of galaxies and so , in section [ s : fate ] , we illustrate the eventual fate of a few high redshift examples and examine the statistical properties of the descendants of the lyman - break objects . in section [
s : epoch ] , we recast the cole ( 1994 ) predictions for the cosmic star formation history in a manner that is directly comparable to the madau ( 1996 ) data , and we also compare our predicted evolution in the neutral gas fraction with observations
. finally , we present our conclusions in section [ s : conc ] .
semi - analytic modelling is a novel technique for calculating _
ab initio _ the evolutionary properties of galaxies in cosmological models in which structure forms hierarchically .
the growth of dark matter halos by accretion and mergers is followed statistically , while physically motivated rules are used to describe the cooling of gas in these halos , the transformation of cold gas into stars , and the effects of feedback from massive stars on the dynamics of the gas .
the spectrophotometric properties of the stars that form are calculated from a spectral synthesis model .
the basic physical concepts , mathematical formalisms , and first applications of semi - analytic modelling are presented in white & rees ( 1978 ) , cole ( 1991 ) , lacey & silk ( 1991 ) , white & frenk ( 1991 ) , kauffmann , white & guiderdoni ( 1993 ) , and cole _
the technique has now been successfully applied to a variety of problems in galaxy formation ( lacey 1993 ; kauffmann , guiderdoni & white 1994 ; heyl 1995 ; kauffmann 1995 , 1996a , b and baugh 1996a , b , frenk 1996 , 1997 ) .
the basic rules that govern the physical processes in the galaxy formation scheme adopted in this paper are presented in detail in cole _
( 1994 ) . in brief , when a dark matter halo collapses , its associated gas is assumed to be shock - heated and to settle into quasistatic equilibrium at the virial temperature of the halo .
this gas cools radiatively over the lifetime of the halo and cold gas is turned into stars at a rate proportional to the instantaneous cold gas mass .
feedback from supernovae and stellar winds returns some of the cold gas to the hot phase , strongly inhibiting star formation in low circular velocity halos . during a merger ,
the dark matter halos coalesce , but the galaxies within them can survive longer , eventually merging on a timescale related to the dynamical friction time . for this analysis , we have upgraded the cole ( 1994 ) model in various ways .
the main modification is the replacement of the `` block model '' ( cole & kaiser 1988 ) as the description of the merger history of dark matter halos . in the new scheme we use a monte carlo method based on the analytical expression for the halo progenitor mass function derived from the extended press - schechter theory " ( bond 1991 ; bower 1991 ; see also lacey & cole 1993 ) to generate binary merger trees .
each tree describes a possible merger history for a halo of specified final mass . at each branch in the tree
a halo splits into two progenitors , but unlike in the `` block model , '' the mass ratio of the two progenitors can take any value .
this technique enables the merger process to be followed with high time resolution , as timesteps are not imposed on the tree but rather are controlled directly by the frequency of mergers .
it is similar in spirit to the method used by kauffmann ( 1993 ) , but has several advantages , including that it does not require the storage of large tables of progenitor distributions .
the new merger scheme is fully described in lacey & cole ( in preparation ) .
a further modification to the scheme is that the singular isothermal sphere model adopted by cole as a description of the dark matter halo density profile has been replaced by the analytical form proposed by navarro , frenk & white ( 1996 ) on the basis of high resolution n - body simulations . in the original cole model all the gas that could cool over the entire life of the halo
was assumed to be available to form stars from the beginning of the halo lifetime .
we now estimate the supply of cold gas available to form stars by calculating the cooling rate at a series of discrete timesteps in which successive shells of gas can cool . in our scheme ,
a cosmological model is specified by an assumption about the nature of the dark matter together with values for the cosmological parameters : the mean cosmic density ( @xmath32 ) , the cosmological constant ( @xmath33 ) , hubble s constant ( @xmath34 ) , the rms mass fluctuations in spheres of radius @xmath35 ( @xmath36 ) , and the mean baryon density in units of the critical density ( @xmath37 ) .
our galaxy formation prescription requires specifying 6 physical parameters : ( i ) a star formation timescale ( @xmath38 ) , ( ii ) a feedback parameter , " ( iii ) the shape of the initial mass function ( imf ) of stars , ( iv ) an overall luminosity normalisation given by the ratio of the total mass in stars , including brown dwarfs , to the mass in luminous stars , ( v ) a merger timescale for galaxies , and ( vi ) the threshold mass for a galaxy merger to turn a disk into a spheroid ( see cole 1994 and baugh 1996b for further details . )
the general strategy that we have adopted in this and previous papers ( baugh 1996ab , frenk 1996 , 1997 ) , is to fix the first 5 basic parameters of the model to obtain the best possible match to the local b - band and k - band galaxy luminosity functions and the sixth parameter to reproduce the local relative abundances of ellipticals , s0 s and spirals .
it turns out that these requirements severely restrict the allowed range of parameter values , except for the imf which in any of the commonly used forms ( ie .
salpeter ( 1955 ) , miller - scalo ( 1979 ) or scalo ( 1986 ) ) has little effect on the predicted _ local _ luminosity function . the evolution of the characteristic luminosity , @xmath39 , however , is sensitive to the choice of imf , which therefore affects predictions for the counts of faint galaxies ( see cole 1994 ) . for the most part
, we have adopted the same values of the parameters as used in the fiducial model of cole et al .
, allowing ourselves the freedom to use different forms for the imf .
we again assume that feedback is a strong function of the halo circular velocity .
the two exceptions are that we have slightly reduced the star formation timescale , @xmath38 , from 2 to 1.5 gyr and we have doubled the ratio of the galaxy merger timescale to the dynamical time in the halo .
the former change leads to an abundance of lyman - break galaxies in better agreement with the data ( c.f .
3.2 ) while the latter change compensates for minor differences introduced by our new monte - carlo scheme for the halo merger trees . with this choice of parameters
, our new model produces luminosity functions that are very similar to those published by cole we have updated the original bruzual - charlot stellar population synthesis model with their new version ( also for solar metallicity only ; bruzual & charlot 1993 , charlot , worthey & bressan 1996 . )
fixing the model parameters by reference to a small subset of the data produces a fully specified model which can then be tested against other data , particularly high redshift data .
thus specified , our model has predictive power and we have presented a number of specific predictions in earlier papers .
two of these are particularly relevant to the present discussion .
the first concerns the redshift distribution of a survey of faint galaxies limited to magnitude @xmath40 ( see figure 20 of cole 1994 ) .
data have now been obtained by glazebrook ( 1995b ) and by cowie ( 1996 ) .
our model predictions are in good agreement with these data , as may be seen in fig . 1 of frenk ( 1997 ) ( see also fig .
15 of white & frenk 1991 and kauffmann , guiderdoni & white 1994 ) .
the second prediction to which we will return in section 5 of this paper , is the cosmic star formation history , presented in figure 21 of cole ( 1994 ) and in fig .
14 below .
= 18.truecm = 21.truecm colour - colour diagrams for galaxies in the standard cdm model .
galaxies brighter than @xmath41 are plotted ; the area of the mock field sampled is @xmath42 square arcminutes .
( a ) @xmath18 colours computed from the true @xmath14 magnitude for all galaxies .
the open circles show galaxies with redshifts @xmath43 and the open triangles galaxies with redshifts in the range @xmath44 .
model results are output at specific redshifts and this is reflected in the discrete regions populated in the diagram .
the redshifts of selected outputs are indicated .
( b ) @xmath18 colours computed by setting the @xmath14 magnitude to a detection limit of @xmath45 whenever @xmath14 is fainter than the detection limit .
galaxies in this class are denoted by filled symbols . as in panel (
a ) , the triangles denote galaxies with redshifts in the range @xmath44 .
[ fig : ug_gr ] we begin by constructing merger trees for a grid of halo masses , specified at some redshift @xmath46 .
typically , we generate between 5 and 20 different realizations for each mass , depending upon the press - schechter abundance .
the galaxy formation rules are applied along the branches of each halo tree , starting at the highest redshift of interest and propagating through to @xmath47 .
volume - limited samples or redshift catalogues are generated from the model output by weighting the galaxies in a halo tree of a given mass by its predicted press - schechter abundance at @xmath48 .
mock catalogues consisting of galaxies selected according to any colour - magnitude criteria can be readily generated from the model output . to construct mock catalogues with the selection criteria of s96 ,
we calculate broad band colours for the model stellar populations using the same set of customised filters employed by s96 .
( the filter functions were kindly provided by c. steidel . ) the dominant effect that determines whether a galaxy is a uv dropout is absorption of the galactic uv light by intervening cold gas .
we calculate the effect of absorption on the spectral energy distributions ( seds ) of high redshift galaxies using the procedure developed by madau ( 1995 ) . in this way , we are able to select model galaxies according to exactly the same colour criteria as applied to the observational data by s96 .
for most of this study we have generated trees starting at @xmath49 .
a high starting redshift is desirable in order to minimize inaccuracies at the high mass end of the mass distribution of progenitors introduced when the monte - carlo scheme is applied over a large range of expansion factor ( lacey & cole , in preparation ) .
we have checked , however , that our results are insensitive to the exact choice of @xmath46 . to obtain the expected total number of galaxies in the apparent magnitude range observed by steidel et al . ,
@xmath50 , we also generated mock galaxy samples from a grid of halo masses laid down at @xmath51 .
at @xmath50 , the median redshift is @xmath52 .
in this section we investigate the properties of the lyman - break galaxies that form in our models and , wherever possible , we compare these to the properties of the real lyman - break galaxies discovered by s96 . specifically , we consider the number density , masses , star formation rates and sizes of these galaxies and we present predictions for their clustering properties . to fully specify a model we need to adopt values for the cosmological parameters , @xmath32 , @xmath33 , @xmath53 and @xmath54 , and also a value for the baryon fraction , @xmath37 , and an imf .
there are considerable uncertainties in these choices .
we carry out calculations in three different cosmological models : the standard cdm model ( @xmath1 , @xmath55 , @xmath56 ) ; a flat , low - density cdm model ( @xmath57 , @xmath58 , @xmath59 , @xmath60 ) and an open cdm model ( @xmath61 , @xmath59 , @xmath62 ) .
these parameters are typical of those favoured by large - scale structure constraints . for example , all of our models produce approximately the correct abundance of galaxy clusters at the present day and , under standard assumptions , the low - density models also match the 4-year cobe microwave background anisotropy data ( bennet 1996 ; liddle 1996 ; white , efstathiou & frenk 1993 ; eke , cole & frenk 1996 ; viana & liddle 1996 ; cole et al . 1997 . ) for our standard @xmath1 cosmology , we vary the normalisation of the primordial fluctuation spectrum , considering both the above value , @xmath56 , adopted in the fiducial model of cole ( 1994 ) , and the lower value , @xmath63 , preferred by eke , cole & frenk ( 1996 ) for consistency with the observed cluster x - ray temperature function .
we consider models with two different baryon fractions , @xmath64 and @xmath65 .
the first agrees with the estimate by copi , schramm & turner ( 1996 ) from big bang nucleosynthesis and the second is consistent with the claim of tytler , fan & burles ( 1996 ) of a low primordial deuterium abundance in gas clouds at high redshift .
we consider three possibilities for the imf , all of which are consistent with solar neighbourhood data , given the uncertainties in its past star formation history : miller - scalo ( 1979 ) , scalo ( 1986 ) and salpeter ( 1955 ) ( see figure 4 in cole 1994 for the specific parametrizations used ) .
the parameters of these models ( and of variants considered below ) are summarized in table 1 .
we first consider the broad - band colours of our model galaxies and test the assumption that high redshift galaxies can be efficiently identified from their location in the @xmath18 _ vs _ @xmath19 colour - colour plot constructed by s96 .
the analysis of steidel , hamilton & pettini ( 1995 ) suggests that galaxies with redshifts in the range @xmath44 should lie within the trapezium bounded by the dashed lines in fig .
1 . at these redshifts
, the lyman break passes through the observer s frame @xmath14 band . our predicted colour - colour diagram , for the case of standard cdm ,
is shown in fig .
the data plotted correspond to an area of @xmath42 square arcminutes , equal to the area of the q0347 - 3819 field observed by steidel , pettini & hamilton ( 1995 ) .
( for clarity the points have been given small random displacements in the @xmath66 and @xmath67 directions .
the localizations of the points in bands reflect the discrete set of output redshifts . ) the open triangles in fig .
1a indicate galaxies that have redshifts in the range @xmath44 . as anticipated by steidel ( 1995 ) , all galaxies with @xmath44 lie inside the trapezium in fig .
1a . however , our models show some contamination by galaxies with redshifts in the range @xmath68 that congregate near the bottom - left corner of the trapezium .
this contamination is , in fact , consistent with the evolutionary tracks shown in figure 2 of steidel ( 1995 ) .
arbitrarily deep lyman breaks can not be measured in practice because the observations are subject to a detection limit in the @xmath14 band . for the q0347 - 3819 field ,
the @xmath69 detection limit is @xmath70 .
1b illustrates the effect of imposing such a detection limit on the appearance of the colour - colour plot . whenever a galaxy has a true @xmath14 magnitude fainter than the field limit , we plot it at the assumed @xmath14 limit using a filled symbol . a significant fraction of high redshift galaxies now lie below the trapezium region , confirming the remark by steidel ( 1995 ) that their selection of candidates is likely to be an underestimate of the true abundance .
comparison of fig .
1b with figures 4 , 7 , 10 and 13 of steidel ( 1995 ) shows that our models tend to produce too few galaxies with @xmath71 and @xmath72 less than unity .
galaxies in this part of the observational diagram are likely to be predominantly faint foreground dwarf irregulars
. the discrepancy may be due in part to our use of solar metallicity stellar populations .
this is also apparent in the comparison between our models and the local luminosity function determined by lilly ( 1995 ) from the cfrs survey ( see figure 16 of baugh , cole & frenk 1996b . ) beyond @xmath73 , however , the luminosity functions of our models for both red and blue galaxies agree reasonably well with the cfrs data .
= 8.6truecm = 8.6truecm predicted redshift distributions for galaxies brighter than apparent magnitude @xmath74 selected in various ways .
panel ( a ) shows results for our standard cdm model a. the solid line shows the distribution for galaxies with redshift in the range @xmath44 . the dotted line shows the distribution for galaxies that meet the colour criteria of steidel ( 1995 ) .
the dashed line refers to galaxies that meet the steidel ( 1995 ) colour selection , after the detection limit in @xmath14 for a typical observed field has been applied - model galaxies fainter than this in @xmath14 are assigned this limiting magnitude . in panel
( b ) the dashed histogram is repeated from panel ( a ) and the solid histogram shows the corresponding redshift distribution of galaxies brighter than @xmath75 in the low-@xmath76 model g brighter than @xmath77 that satisfy the steidel colour selection criteria , taking into account the typical field limit in @xmath14 . these and subsequent histograms are normalized so that they enclose unit area .
[ fig : dndz ] = 16.truecm = 16.truecm predicted galaxy number counts . in panel
( a ) , the higher amplitude pair of lines shows the total counts in model a ( solid line ) and model g ( dotted line ) .
the lower amplitude pair of lines shows the counts of galaxies in each of these models that satisfy the steidel colour selection criteria alone , without any constraint on their @xmath14 magnitudes . in panel ( b ) the total counts in model a are again shown by the solid line .
the dotted line now shows the counts of colour selected galaxies ; the short dashed line shows the counts of galaxies with redshifts in the range @xmath44 ; and the long - dashed line shows the counts of galaxies with redshifts in the range @xmath78 .
[ fig : counts ] [ cols="^,^,^,^,^,^,^ , < " , ] the @xmath27 data plotted in fig .
15 were derived from the h@xmath79 luminosity function of gallego ( 1995 ) .
the high - redshift data in the figure were derived from the data presented by madau ( 1996 ) for galaxies in the hubble deep field identified photometrically as lyman - break galaxies .
the @xmath80-band dropouts are estimated to have @xmath81 and @xmath82 , while the @xmath83-band dropouts are estimated to have @xmath84 and @xmath85 .
madau gives sfrs derived from broad - band magnitudes at wavelengths close to @xmath86 in the rest frame .
we first convert these back to the corresponding values of @xmath87 using madau s own conversion factor , as given in madau ( 1996 ) .
we then convert the @xmath86 and h@xmath79 luminosities into total sfrs using the values tabulated in table 5 and the appropriate value of @xmath88 for each model .
in addition , for models with @xmath89 an approximate scaling has been applied to madau s data points to account for the differences in the comoving volume element and luminosity distance in the different cosmologies .
the curves plotted on fig .
15 display our model predictions .
the upper panel shows results for our standard @xmath1 model a and the lower panel for our flat low-@xmath76 model g , both of which have the miller - scalo imf .
the middle panel shows model c which has @xmath1 and the salpeter rather than the miller - scalo imf .
the theoretical curves are exactly the same for this case as for model a , the only difference being the scalings applied to the observational data points . in all cases the behaviour of the models is qualitatively the same as the observed data .
the sfrs are larger at the intermediate redshift @xmath90 and at @xmath91 they drop back to values similar to those at @xmath27 . the model that best reproduces the observed data is model a. there are , however , considerable uncertainties in this comparison .
for example , the mean redshift of the @xmath80 dropouts in our mock hst catalogue is @xmath92 , smaller than the central value , @xmath93 , assumed for the real data .
this approximation alone would lead to an overestimate of the inferred star formation rate by up to @xmath94% . for the @xmath83 dropouts this effect is smaller , around @xmath95 .
more importantly , these comparisons are sensitive to the details of the adopted imf , as may be seen by the way in which the data points shift in panels @xmath96 and @xmath97 .
the cosmic star formation history may be conveniently summarised by considering the variation with redshift of the total sfr per comoving volume , as in the theoretical predictions of fig .
this quantity is obtained by integrating the differential distributions in fig .
15 over galaxies of all sfrs . for the theoretical models ,
this is straightforward , but for the observations , it involves extrapolating the distribution of sfrs to ranges not directly observed . comparing theoretical and
observed " total sfrs is thus considerably more uncertain than comparing differential distributions . with this caveat ,
we compare in fig .
16 our theoretical predictions with several observational estimates including those based on the data of gallego ( 1995 ) and madau ( 1996 ) already described , and also the @xmath98-band data of lilly ( 1996 ) for @xmath99 . in all cases ,
we have used the observers estimates of the total luminosity density , which are based on fitting a schechter function to the luminosity function and extrapolating it to all luminosities .
the lilly data refer to the luminosity density at @xmath100 .
this is less directly related to the instantaneous star formation rate than the h@xmath79 or @xmath86 luminosities , because it is dominated by somewhat older stars .
the @xmath101 rest - frame luminosity is sampled by the observed b - band flux only at @xmath102 , so the estimate at @xmath103 requires a modest extrapolation from longer wavelengths .
this introduces an additional uncertainty .
the constants used to convert @xmath104 to total sfr are also listed in table 5 .
the upper limit plotted at @xmath105 is based on the number of @xmath106 dropouts in the hdf , which are candidates to be lyman - break galaxies at @xmath107 ( madau , private communication ) .
after the original version of this paper was submitted , many more data points have been added to the star formation history diagram and we reproduce here a selection of them . at low redshift , treyer ( 1997 )
have estimated the star formation density from @xmath108 ( rest - frame ) fluxes ; sawicki ( 1997 ) have used photometric redshifts of galaxies in the hdf to infer the star formation density from @xmath109 fluxes ; connolly ( 1997 ) have used both optical and ground based near - infrared imaging of the hdf to infer star formation rates from @xmath101 fluxes .
( the use of infrared data is particularly important for the accuracy of photometric redshifts at @xmath110 . )
apart from the sawicki points at @xmath111 , the level of agreement amongst these different determinations is remarkable , although it is suggestive that both the sawicki and connolly points at @xmath112 lie above those from the cfrs survey , in better agreement with our model predictions .
( the cfrs survey , however , has more galaxies and therefore smaller error bars at these redshifts . )
overall , the agreement between theoretical predictions and data in fig .
16 is impressive .
it must be borne in mind that these are genuine theoretical predictions that predate the observational data .
the theoretical curve in the upper panel of fig .
16 is simply the time derivative of the integrated curve for the fiducial cdm model plotted in figure 21 of cole ( 1994 ) .
note , however , that the location of the observational data points depends on the assumed imf ; cole used a scalo imf whereas in 3.2 , we found a miller - scalo or salpeter imf to be preferable . for a given luminous star formation rate , @xmath113 ,
a miller - scalo imf gives @xmath114 times the h@xmath79 flux , @xmath115 times the @xmath86 flux and @xmath116 times the @xmath101 flux compared to a scalo imf . comparing panels ( a ) and ( b ) of fig .
16 , we see that the main effect of changing the imf from miller - scalo to salpeter is to move the @xmath27 data point based on the h@xmath79 luminosity .
both models a and g show the same qualitative trend as the data , with a broad peak in the total sfr at @xmath117 . for @xmath118 ,
the model sfrs fall off somewhat more slowly than the data .
the completeness of the observational data at these high redshifts , however , is difficult to establish .
a further source of uncertainty is the possible presence of dust in the star - forming galaxies .
even a modest amount of dust would cause attenuation of the ultraviolet flux , leading to a potentially severe underestimate of the star formation rate .
tentative detections of the cosmic infrared background by puget ( 1996 ) and guiderdoni ( 1997a ) and upper limits on it ( kashlinsky , mather & odenwald 1996 ) provide only weak constraints on the amount of dust present in high redshift galaxies ( madau , pozzetti & dickinson , 1997 ; guiderdoni 1997b ) .
monolithic collapse models ( e.g. eggen , lynden - bell & sandage 1962 ) in which a significant fraction of the total star formation in the universe takes place at high redshifts enshrouded in dust ( e.g. meurer 1997 ) , appear to overpredict the total mass of heavy elements in place at early times ( madau 1997 ) , as inferred from observations of damped lyman-@xmath79 systems ( pettini , smith , king & hunstead 1997 ) .
estimates of the factor by which star formation rates deduced from uv flux should be revised to account for the presence of dust span a range of values .
the results are sensitive both to the form of the extinction law adopted and to the assumed age of the primeval galaxy which determines how intrinsically blue it is .
primeval galaxies in our models are not , in general , ultraluminous starbursts since they never experience exceptional star formation rates . in this case , dickinson ( in preparation ) and pettini ( 1997 ) argue that the likely correction is around a factor of @xmath119 at @xmath120 for star formation rates inferred from @xmath86 fluxes .
these corrections are a factor of @xmath121 smaller than those advocated by meurer ( 1997 ) . in fig .
16 , we illustrate the effect of the pettini correction at high redshift by multiplying the points of madau ( 1996 ) by a factor of 3 ( open squares ) .
the correction appropriate to the lower redshift points is also uncertain and we do not attempt to illustrate it in fig .
it is likely to be smaller than at high redshift since the star formation rates are derived from longer wavelength data .
a related observational constraint on the evolution of the galaxy population comes from observations of neutral hydrogen at high redshift using quasar absorption lines .
17 compares the evolution of the cold gas fraction in our models with estimates by storrie - lombardi ( 1996 ) , derived from the statistics of damped lyman - alpha absorption lines .
whereas kauffmann s ( 1996b ) semi - analytic models agree quite well with these data , our own models agree only in the qualitative sense that the comoving cold gas density has a broad peak at a redshift @xmath122@xmath123 .
our models predict consistently more cold gas than is inferred from the observations . on the other hand
, the observational results may underestimate the total cold gas density in galaxies because ( i ) they only include atomic hydrogen at column densities @xmath124 , and do not include ionized or molecular gas at all ; and ( ii ) dust obscuration may cause some absorption systems to be missed . regarding ( i ) ,
all the gas in our model galaxies at @xmath125 is counted as `` cold '' ; the correction for ionized gas and for low column - density hi ( @xmath126 ) is probably not large , but the correction for molecular hydrogen might be significant . regarding ( ii ) , the chemical evolution models of pei & fall ( 1995 ) suggest that because of dust obscuration , the true neutral hydrogen density is 2 - 3 times higher than the `` directly measured '' value , moving the observational points in fig .
17 much closer to the theoretical curves . the dotted - line set of errorbars in fig .
17(a ) show plausible corrections for these effects , using the output from one of the models of pei & fall ( 1995 ) , following figure 2 of storrie - lombardi ( 1996 ) .
= 9.truecm = 9.truecm the total comoving density of cold gas , @xmath127 , in units of the present critical density .
the curve in the upper panel shows the evolution of @xmath127 with redshift in model a. the dependence of @xmath128 with redshift in models b and c which differ from a only in the choice of imf , are identical to that in a. the curve in the lower panel shows results for model g. the data points are observational estimates , based on damped lyman - alpha absorption lines from storrie - lombardi ( 1996 ) .
we have applied an approximate scaling to their @xmath129 estimates to derive the corresponding values for the flat low-@xmath76 model g. the data point at @xmath27 is based on the @xmath130 luminosity function of nearby galaxies derived from @xmath131 observations .
the dotted - line errorbars in the upper panel show the corrections to the data suggested by storrie - lombardi to account for the effects of incompleteness and of dust obscuration , using a model from pei & fall ( 1995 ) .
we have retained the same fractional errors on the corrected data points .
[ fig : cgas ] = 9.truecm = 9.truecm the ratio of the number density @xmath132 of ionizing photons produced per comoving volume to the number density @xmath133 of hydrogen atoms . at the redshift at which @xmath134 , just enough ionizing photons have been produced to ionize every hydrogen atom exactly once .
the solid line is for model a , the dotted for model g and the dashed for model c. [ fig : nphnb ] an interesting issue is whether massive stars in young galaxies can , on their own , produce enough ionizing photons ( @xmath135 ) to re - ionize the igm by @xmath6 .
( additional ionizing photons are produced by quasars , and possibly by pregalactic stars formed at high redshift via molecular hydrogen cooling , e.g. tegmark 1997 ) . a very simple criterion for establishing
if re - ionization is possible by redshift @xmath136 is based on the ratio @xmath137 , where @xmath132 is the total number of ionizing photons produced per comoving volume up to redshift @xmath136 , and @xmath133 is the total comoving density of hydrogen atoms .
when @xmath134 , just enough ionizing photons are produced to ionize each hydrogen atom exactly once .
this criterion neglects absorption of photons within the emitting galaxies , recombination of hydrogen in the igm , and depletion of the igm by collapse of gas into dark halos .
in particular , the fraction @xmath138 of ionizing photons that escape from a galaxy may be quite low : dove & shull ( 1994 ) estimate @xmath139 for our own galaxy , while leitherer ( 1995 ) estimate @xmath140 from far - uv observations of nearby starburst galaxies .
the photon number density , @xmath132 , is also sensitive to the form of the imf above @xmath141 .
18 shows the dependence of @xmath137 on redshift for models a , c and g.
we have used a semi - analytic model of galaxy formation to interpret recent data on galaxy formation and evolution within the context of hierarchical clustering theories , focussing primarily on the properties of the recently discovered population of lyman - break galaxies at @xmath142 .
our modelling technique allows us to identify the role that this population plays within the general scheme of galaxy formation , and to relate these observations to other data at lower redshifts .
our models are quite general , but they inevitably require a number of assumptions and simplifications most of which , in fact , reflect our poor understanding of the processes of star formation and feedback . within these limitations
we attempt to represent the relevant physics using scaling laws that involve the smallest possible number of free parameters .
our modelling strategy is based on fixing these free parameters by requiring that the models should match a small subset of the local data , particularly the field b - band galaxy luminosity function .
thus specified , the models possess predictive power and can be tested against high redshift data . the specific cold dark matter models that we have considered all share the feature that they reproduce the observed abundance of present day rich clusters of galaxies .
this fixes the amplitude of primordial density fluctuations which determines the epoch at which structures on any mass scale form .
current large - scale structure data allow a range of values for the cosmological parameters @xmath32 and @xmath33 . by way of illustration ,
we have explored in detail a critical density model and two low - density models , one open and the other flat .
a second common feature of the models we have considered is that they all agree at some level with most data on the evolutionary properties of galaxies at relatively modest redshifts , @xmath143 , such as the evolution of the luminosity function ( baugh 1996a ) , and the counts of galaxies as a function of magnitude , morphological type , and redshift ( cole 1994 , baugh 1996b , frenk 1996 ) .
several interesting predictions of the models have been corroborated by subsequent data , including , for example , the redshift distribution of @xmath40 counts ( frenk 1997 ) .
an important prediction of semi - analytic models which also predated the acquisition of the relevant data is the cosmic star formation history , first discussed by white & frenk ( 1991 ) and calculated by cole ( 1994 ) and heyl ( 1995 ) for the specific models discussed here ( c.f .
we showed in section 5 that madau s ( 1996 ) recent data agree well with these model predictions , although uncertainties remain in this comparison because of the unknown effects of dust obscuration and possible incompleteness in the observational samples at high redshifts .
in particular , the possibility that a significant fraction of the total star formation may have been missed in recent surveys if it occured in dust enshrouded starbursts at very high redshift has been the subject of some recent debate .
such a population is not predicted in our models , although a certain amount of dust obscuration at high redshift can be accommodated and may even be required : our models , in fact , predict star formation rates which are somewhat higher than those inferred from the data uncorrected for dust obscuration ( c.f .
16 ) . the star formation rate in our models peaks around @xmath144 ,
but it never varies by more than an order of magnitude over the entire range @xmath145 .
in fact , the star formation rate at @xmath28 is almost identical to the star formation rate at the present day
. nevertheless , half of all the stars present today only formed since @xmath146 .
the significance of the lyman - break galaxies in the context of these models lies in the fact that at the epoch when these galaxies are observed , the process of star formation is just beginning in earnest .
thus the lyman - break galaxies are the first massive objects to sustain appreciable star formation and , in this sense , they signal the onset of galaxy formation .
the _ distribution _ of star formation rate at different redshifts provides even stronger constraints on models than the evolution of the integrated star formation rate .
our models match existing data quite well , and further observational measurements of this fundamental quantity are very important . in both models and observations ,
the quiescent star formation rates are relatively low even in the largest protogalaxies , with the majority of galaxies never forming stars at rates exceeding a few solar masses per year .
the exception to this are galaxies that undergo a burst of star formation as a result of experiencing a major merger .
however , predictions for the strength , duration and frequency of such bursts will require more detailed modelling that we have attempted so far . in our models ,
the lyman - break galaxies form from rare peaks in the density field at high redshift . as a result
, we predict that their spatial distribution at @xmath0 should be strongly biased relative to the underlying mass , with a typical bias parameter , @xmath147 , and a comoving clustering length , @xmath4 .
generically , we expect the lyman - break galaxies to be rotating disks , and a simple model for the origin of their angular momentum predicts typical half - light radii of @xmath148 .
the lyman - break galaxies seen at @xmath0 evolve into the present day ordinary ellipticals and spirals that make up the bright end of the luminosity function .
their stars will typically be concentrated in the central regions of their descendants .
these descendants are to be found preferentially in groups and clusters , reflecting their biased origin and strong clustering at high redshift .
the appearance of the first protogalaxies at @xmath149 and the late conversion of most of the gas into stars fit in well with the observation that the neutral hydrogen content of the universe , as determined from measurements of damped lyman - alpha clouds , peaks at around @xmath150 and declines thereafter ( storrie - lombardi 1996 ) .
the neutral hydrogen density in our models also exhibits this overall behaviour and agrees reasonably well with the data once corrections for incompleteness and a small amount of dust obscuration are included .
related semi - analytic models by kauffmann ( 1996b ) agree even better with these data .
our models are consistent with madau s ( 1996 ) view that , with the discovery of lyman - break galaxies , the bulk of the star formation ( and the attendant metal production ) in our universe has , in effect , been identified .
only a small fraction of the star formation activity remains to be detected during the `` dark ages '' prior to @xmath91 .
characterising such activity is , of course , of great importance for testing the general view that galaxies formed by hierarchical clustering . according to our models , the small percentage of stars that formed prior to the lyman - break galaxy epoch produce enough radiation to make at least a significant contribution to the uv flux required photoionize the intergalactic medium by @xmath28 .
although a late epoch of galaxy formation in the standard cdm model was predicted long ago ( frenk 1985 , white & frenk 1991 ) , a surprising result of our analysis is that this is also true of the now popular low - density variants of this model . indeed , the star formation histories of the @xmath1 model and the flat @xmath57 model are remarkably similar .
this is largely coincidental : the detailed star formation histories depend not only on the shape of the power spectrum , but also on its normalisation and on the way in which star formation and feedback are implemented in our galaxy formation models .
the main conclusion of our analysis is that , regardless of the exact values of the cosmological parameters , cdm models that approximately reproduce the abundance of lyman - break galaxies , require massive galaxy formation to begin around @xmath149 . in summary
, we have argued in this paper that the main ingredients of a consistent picture of galaxy formation may now be in place .
the key observation that has unlocked this paradigm is the discovery of a large population of star - forming galaxies at @xmath0 which signal the onset of the epoch of galaxy formation that extends well into the present day . at this time , data and theoretical modelling paint only a broad brush picture of how galaxy formation may have occurred .
fortunately , if this emerging picture is correct , the details should also be accessible to current observational and modelling capabilities .
we thank charles steidel for supplying the transmission of the filters used in his observations and for valuable discussions .
we would like to thank the referee , piero madau , and simon white for their careful reading of the manuscript and for helpful comments which helped improve the final version of this paper .
we also acknowledge discussions with richard ellis , michael fall , max pettini and martin rees .
this research was supported by the european commission through the tmr network on `` the formation and evolution of galaxies '' and in part by a pparc rolling grant .
smc acknowledges a pparc advanced fellowship and csf acknowledges a pparc senior fellowship .
cgl was supported by the danish national research foundation through its establishment of the theoretical astrophysics center .
abraham , r.g . , tanvir , n.r . , santiago , b.x . , ellis , r.s . ,
glazebrook , k. , van den bergh , s. , 1996 279 , l47 baron , e. , white , s.d.m . , 1987 , 322 , 585 baugh , c.m . , efstathiou , g. , 1993 , 265 , 145 baugh , c.m . ,
cole , s. , frenk , c.s .
, 1996a , 282 , l27 baugh , c.m . , cole , s. , frenk , c.s . , 1996b , 283 , 1361 bennet , c.l .
1996 , 464 l1 bond , j.r . ,
cole , s. , efstathiou , g. , kaiser , n. , 1991 , 379 , 440 bower , r.g .
, 1991 , 248 , 332 brainerd , t.g . , smail , i. , mould , j. 1995 , 275 , 781 bruzual , g. , charlot , s. , 1993 , 405 , 538 charlot , s. , worthey , g. , bressan , a. , 1996 , 457 , 625 cole , s. , 1991 , 367 , 45 cole , s. , kaiser , n. , 1988 , 233 , 637 cole , s. , aragn - salamanca , a. , frenk , c.s . , navarro , j.f . , zepf , s.e . , 1994 , 271
, 781 cole , s. , weinberg , d.h .
, frenk , c.s . ,
ratra , b. , 1997 , 289 , 37 connolly , a.j . , szalay , a.s . ,
dickinson , m. , subbarao , m.u . , brunner , r.j .
1997 , 486 , l11 copi , c.j . ,
schramm , d.n . ,
turner , m.s .
, 1996 , nucl .
b. , s51b , 66 .
cowie , l.l . ,
songaila , a. , hu , e.m .
, cohen , j.g . , 1996 , 112 , 839 dove , j.b , shull , j.m . , 1994 , 430 , 222 dressler , a. , oemler , a. , sparks , w.b . ,
lucas , r.a .
, 1994 , 435 , l23 driver , s.p . ,
windhorst , r.a . ,
griffiths , r.e .
, 1995 , 453 , 48 eggen , o.j . ,
lynden - bell , d. , sandage , a.r .
, 1962 , 136 , 748 eke , v.r . , cole , s. , frenk , c.s .
, 1996 , 282 , 263 ellis , r.s . , colless , m. , broadhurst , t. , heyl , j. , glazebrook , k. , 1996 , mnras , 280 , 235 frenk , c.s . ,
baugh , c.m . ,
cole , s. , 1996 , iau symposia 171 , 247 frenk , c.s . ,
baugh , c.m . , cole , s. , lacey , c.g .
, 1997 to appear in dark matter and visible matter in galaxies , eds .
m. persic & p. salucci .
gallego , j. , zamorano , j. , aragn - salamanca , a. , rego , m. , 1995 , 455 , l1 giavalisco , m. , steidel , c.c . ,
macchetto , f.d . , 1996 , 470 , 189 glazebrook , k. , ellis , r. , santiago , b. , griffiths , r. , 1995a , 275 , l19 .
glazebrook , k. , ellis , r. , colless , m. , broadhurst , t. , allington - smith , j. , tanvir , n. , 1995b , 273 , 157 guiderdoni , b. , hivon , e. bouchet , f. r. , maffei , b. 1997a , in press .
guiderdoni , b. , bouchet , f. r. , puget , j. l. , lagache , g , & hivon , e. 1997b , nature , in press .
heyl , j.s . ,
cole , s. , frenk , c.s . ,
navarro , j.f .
, 1995 , 274 , 755 kashlinsky , a. , mather , j.c . ,
odenwald , s. , 1996 , 437 l9 .
kauffmann , g. , 1995 , 274 , 161 kauffmann , g. , 1996a , 281 , 475 kauffmann , g. , 1996b , 281 , 487 kauffmann , g. , white , s.d.m .
, guiderdoni , b. , 1993 , 264 , 201 kauffmann , g. , guiderdoni , b. , white , s.d.m .
, 1994 , 267 , 981 kauffmann , g. , charlot , s. , white , s.d.m . , 1996 283 , l117 lacey , c.g . , cole , s. 1993 262 , 627 lacey , c.g . , guiderdoni , b. , rocca - volmerange , b. , silk , j. , 1993 , 402 , 15 lacey , c.g . , silk , j. , 1991 381,14 lanzetta , k.m . , wolfe , a.m. , turnshek d.a . , 1995 , 440 , 435 leitherer , c. , ferguson , h.c . ,
heckman , t.m . ,
lowenthal , j.d .
, 1995 , 454 , l19 liddle , a. r. , lyth , d. h. , viana , p. t. p. , white , m. , 1996 , 282 ,
281 lilly , s.j . , tresse , l. , hammer , f. , crampton , d. , lefevre , o , 1995 , 455 , 108 lilly , s.j . , lefevre , o , hammer , f. , crampton , d. , 1996 , 460 , l1 limber , d. n. , 1954 , 119 , 655 loveday , j. , peterson , b.a . , efstathiou , g. , maddox , s.j . , 1992 , 390 , 338 lowenthal , j. d. , koo , d. c. , guzman , r. , gallego , j. , phillips , a. c. , faber , s. m. , vogt , n. p. , illingworth , g. d. , gronwall , c. , 1997 , 481 , 673 lu , l.m . ,
sargent , w.l.w . ,
womble , d.s . ,
takadahidai , m. , 1996 , 472 , 509 marzke , r.o . ,
huchra , j.p . ,
geller , m.j .
, 1994 , 428 , 43 madau , p. , 1995 , 441 , 18 madau , p. , 1996
, preprint astro - ph/9612157 madau , p. , ferguson , h.c , dickinson , m. , giavalisco , m. , steidel , c.c . ,
fruchter , a. , 1996 , 283 , 1388 madau , p. , pozzetti , l. , dickinson , m. , 1997 , submitted to astro - ph/9708220 meurer , g.r . ,
heckman , t.m .
, lehnert , m.d . ,
leitherer , c. , lowenthal , j. 1997 , , 114 , 54 miller , g.e . ,
scalo , j.m .
, 1979 , 41 , 513 mo , h.j . , fukugita , m. , 1996 , 467 , l9 mo , h.j . ,
jing , y.p . , white , s.d.m .
, 1996 , 282 , 1096 mo , h.j . ,
white , s.d.m .
, 1996 , 282 , 347 navarro , j.f . ,
frenk , c.s . ,
white , s.d.m .
, 1996 , 462 , 563 odewahn , s.c .
, windhorst , r.a . , driver , s.p . ,
keel , w.c . , 1996 , 472 , l13 pascarelle , s.m . ,
windhorst , r.a . ,
keel , w.c . ,
odewahn , s.c .
1996 , nature , 383 , 45 peacock , j.a . ,
dodds , s.j .
, 1996 , 280 , l19 peebles , p.j.e . , 1980 , large scale structure in the universe princeton .
pei , y.c .
, fall , s.m , 1995 , 154 , 69 pettini , m. , smith , l.j . , king , d.l . , hunstead , r.w . , 1997 , 486 , 665 pettini , m. , steidel , c.c .
, adelberger , k.l . ,
kellogg , m. , dickinson , m. , giavalisco , m. , 1997 , to appear in ` origins ' , ed .
shull , c.e .
woodward , and h. thronson , ( asp conference series ) ( astro - ph/9708117 ) press , w.h . ,
schechter , p.l .
, 1974 , 187 , 425 puget , j.l . , abergel , a. , bernard , j.p . ,
boulanger , f. , burton , w.b . , desert , f.x . ,
hartmann , d. , 1996 , astron .
astrop . , 308 , l5 sawicki , m.j . , lin , h. , yee , h.k.c .
1997 , 113 , 1 salpeter , e.e . , 1955 , 121 , 61 scalo , j.m . , 1986 , fundamentals of cosmic physics , 11 , 1 smail , i. , dressler , a. , kneib , j.p . ,
ellis , r.s . , couch , w.j . ,
sharples , r.m .
, oemler , a. , 1996 , , 469 , 508 .
smail , i. , dressler , a. , couch , w.j . ,
ellis , r.s .
, oemler , a. , butcher , h. , sharples , r.m . , 1997 , 110 , 213 steidel , c.c . ,
hamilton , d. , 1992 , 104 , 941 steidel , c.c . , hamilton , d. , 1993 , 105 , 2017 steidel , c.c . , pettini , m. , hamilton , d. , 1995 , 110 , 2519 steidel , c.c . ,
giavalisco , m. , pettini , m. , dickinson , m. , adelberger , k.l .
, 1996a , 462 , l17 ( s96 ) .
steidel , c.c . ,
giavalisco , m. , dickinson , m. , adelberger , k.l . , 1996b , 112 , 352 storrie - lombardi , l.j . ,
mcmahon , r.g . ,
irwin , m.j .
, 1996 , 283 , l79 tegmark , m. , silk , j. , rees , m.j . ,
blanchard , a. , abel , t. , palla , f. , 1997 , 474,1 treyer , m.a .
, ellis , r.s . , milliard , b. , donas , j. , 1997 , to appear in the ultraviolet universe at low and high redshift : probing the progress of galaxy evolution " , aip press tytler , d. , fan , x.m . ,
burles , s. , 1996 , nature , 381 , 207 viana , p.t.p . ,
liddle , a.r .
, 1996 , 281 , 323 white , s.d.m . ,
efstathiou , g. , frenk , c.s .
, 1993 , 262 , 1023 white , s.d.m . ,
frenk , c.s . ,
1991 , 379 , 52 white , s.d.m . ,
rees , m.j .
, 1978 , 183 , 341 .
williams , r.e . ,
1996 , 112 , 1335 .
wolfe a.m. , lanzetta k.m .
, foltz c.b . ,
chaffee , f.h .
, 1995 , 454 , 698 | we use a semi - analytic model of galaxy formation in hierarchical clustering theories to interpret recent data on galaxy formation and evolution , focussing primarily on the recently discovered population of lyman - break galaxies at @xmath0 . for a variety of cold dark matter ( cdm )
cosmologies we construct mock galaxy catalogues subject to identical selection criteria to those applied to the real data .
we find that the expected number of lyman - break galaxies is very sensitive to the assumed stellar initial mass function and to the normalization of the primordial power spectrum . for reasonable choices of these and other model parameters , it is possible to reproduce the observed abundance of lyman - break galaxies in cdm models with @xmath1 and with @xmath2 .
the characteristic masses , circular velocities and star - formation rates of the model lyman - break galaxies depend somewhat on the values of the cosmological parameters but are broadly in agreement with available data .
these galaxies generally form from rare peaks at high redshift and , as a result , their spatial distribution is strongly biased , with a typical bias parameter , @xmath3 , and a comoving correlation length , @xmath4 .
the typical sizes of these galaxies , @xmath5 , are substantially smaller than those of present day bright galaxies . in combination with data at lower redshifts ,
the lyman - break galaxies can be used to trace the cosmic star formation history .
we compare theoretical predictions for this history with a compilation of recent data .
the observational data match the theoretical predictions reasonably well , both for the distribution of star formation rates at various redshifts and for the integrated star formation rate as a function of redshift .
most galaxies ( in our models and in the data ) never experience star formation rates in excess of a few solar masses per year .
our models predict that even at @xmath6 , the integrated star formation rate is similar to that measured locally , although less than 1% of all the stars have formed prior to this redshift .
the weak dependence of the predicted star formation histories on cosmological parameters allows us to propose a fairly general interpretation of the significance of the lyman - break galaxies as the first galaxy - sized objects that experience significant amounts of star formation .
these galaxies mark the onset of the epoch of galaxy formation that continues into the present day .
the basic ingredients of a consistent picture of galaxy formation may well be now in place .
galaxies : evolution - galaxies : formation - galaxies : fundamental parameters |
the kek - to - kamioka ( k2k ) long baseline neutrino experiment @xcite is designed to explore , with laboratory neutrinos , the mechanism of @xmath7 disappearance indicated by the super - kamiokande ( sk ) @xcite , macro @xcite , and soudan 2 @xcite atmospheric neutrino experiments .
such mechanism appears to be dominated by @xmath0 oscillations @xcite with mass - mixing parameters @xmath8 close to @xmath9 .
the pathlength @xmath10 and typical energies ( @xmath11 ) of k2k @xmath7 s are well suited to study @xmath7 disappearance effects for such parameter values @xcite . the recent data released by the k2k experiment @xcite ( 44 observed neutrino events _ vs _ 63.9 expected ) are already inconsistent with the no - oscillation hypothesis at about 97% c.l.@xcite , and suggest disappearance of the muon neutrino flavor in the path from kek to kamioka .
the k2k collaboration is being understandably rather conservative on the oscillation explanation of the data , pending more accurate estimates of the k2k neutrino spectrum and interaction uncertainties by means of the near detector @xcite .
however , it is tempting to use just the minimum amount of spectrum - integrated data ( i.e. , the total event rate , for which the error estimate appears to be relatively stable ) , in order to study the compatibility of k2k and sk data @xcite . in this paper
we attempt to investigate such a compatibility issue , by comparing predictions and data in both k2k and sk , within the simplest ( two - family ) @xmath0 oscillation framework .
we show that the k2k and the sk data are consistent , and that their combination starts to be useful in constraining further the oscillation parameters . in particular , the 99% c.l .
range for @xmath2 allowed by sk only , @xmath3\times10^{-3}$ ] ev@xmath4 , is reduced to @xmath5\times 10^{-3}$ ] ev@xmath4 by including k2k data .
finally we show that , by halving the k2k total rate error , the @xmath2 range would be ulteriorly reduced to @xmath6\times10^{-3}$ ] ev@xmath4 .
analyses of k2k data have to consider the energy spectrum of parent neutrinos at the far detector @xmath12 , i.e. , the spectrum of neutrino that have produced a detected interaction at the kamioka site .
such spectrum is , in general , given by the product of the unoscillated neutrino energy spectrum @xmath13 at the far ( sk ) detector , times the interaction cross section @xmath14 , times the detection efficiency @xmath15 ( all in differential form ) , integrated over all the final state parameters ( simbolically , @xmath16 ) .
the spectrum @xmath12 is important since , in the presence of neutrino mass and mixing , the oscillated event rate can be expressed as @xmath17 [ to be compared with the unoscillated rate , @xmath18 , where @xmath19 is the @xmath7 survival probability . at present
, there is enough public information on @xmath13 @xcite , but not yet on the product @xmath20 relevant for the far detector specifications .
therefore , a direct and accurate reconstruction of the spectrum @xmath12 and of its uncertainties is not possible outside the k2k collaboration , at least at present .
however , an indirect , approximate reconstruction can be made , by considering that : ( a ) the neutrino spectrum is practically zero above 5 gev @xcite ; and ( b ) according to the k2k montecarlo ( mc ) simulation , for the current exposure ( @xmath21 p.o.t . )
one has @xmath22 events , while @xmath23 , 27.4 , and 23.1 events for @xmath24 , 5 , and @xmath25 ev@xmath4 , respectively , assuming maximal mixing @xcite .
we have then introduced an empirical functional form for the spectrum , @xmath26 with @xmath27 , which is found to reproduce well the above mc results for @xmath28 . in the following analysis
we use such interacted neutrino spectrum , since it gives by construction results close to the k2k mc simulation , after integration over @xmath29 .
a more accurate ( and less tentative ) parametrization of @xmath12 shall be possible after the experimental specifications and selection cuts ( relevant to determine @xmath30 and its uncertainties ) will be measured and described in detail . concerning the uncertainties , the current systematic error on the total rate
is estimated to be 10% by the k2k collaboration @xcite .
spectral bins may be affected by larger ( and correlated ) errors @xcite , which , however , do not affect our spectrum - averaged analysis .
therefore , we attach a @xmath31 fractional error to the oscillated number of events @xmath32 , while the statistical ( poisson ) error @xmath33 is attached to the number of observed events , @xmath34 .
the @xmath35 statistics for k2k is then simply given by @xmath36 , where @xmath37 .
with such definition , the no oscillation hypothesis ( @xmath38 ) is disfavored at 97% c.l .
( @xmath39 for @xmath40 ) , consistently with the current claim of the k2k collaboration @xcite .
figure 1 shows the number @xmath41 of k2k events for the current exposure ( @xmath42 p.o.t . ) , as a function of @xmath2 , for maximal mixing @xmath43 . the solid curve , corresponding to our calculation of @xmath44 , starts from @xmath45 events in the no - oscillation limit ( @xmath46 ) , and tends to the asymptotic value of 63.9/2 in the fast oscillation limit ( @xmath47 ) , after a few `` wiggles '' associated to the first oscillation cycles .
the curve is very close to the four k2k mc points used to benchmark our calculation .
the horizontal gray band represents the @xmath48 interval for the total number of observed events in k2k .
the allowed band disfavors no oscillations at @xmath49 , and the first deep oscillation minimum at @xmath50 .
it is instead in very good agreement with the oscillated predictions for @xmath51 ev@xmath4 , the best fit being reached at @xmath52 ev@xmath4 .
such a value is very close to the one independently quoted by the sk collaboration as the current best fit to their atmospheric neutrino data ( @xmath53 ev@xmath4 ) @xcite , under the same @xmath0 oscillation hypothesis .
ev@xmath4 @xcite .
] therefore , it makes sense to combine k2k and sk data , as we do in the following section .
given the very good agreement between the k2k and sk allowed ranges for @xmath2 at maximal mixing , it is interesting to study their compatibility for unconstrained @xmath54 mixing . concerning the sk atmospheric @xmath54 analysis , we make use of the results recently reported by us in @xcite ( for 79.5 kty sk data ) , specialized to the simplest scenario of pure @xmath0 oscillations . to draw iso-@xmath55 contours , while we used @xmath56 for the more general @xmath57 and @xmath58 cases considered in @xcite .
] figure 2 shows the regions allowed at 90% and 99% c.l . by sk , k2k , and their combination , in the @xmath54 mass - mixing plane @xmath8 .
the upper left panel corresponds to sk atmospheric data only , for which we find two degenerate best - fit points ( stars ) at @xmath59 ev@xmath4 and at octant - symmetric mixing values , @xmath60 , corresponding to slightly nonmaximal oscillation amplitude , @xmath61 . the upper right panel in fig .
2 shows the fit to k2k only . the locus of best - fit points is a continuous , octant - symmetric curve ( not shown ) passing through @xmath62 .
the k2k constraints in the mass - mixing plane are weaker than those placed by sk , especially on @xmath63 .
therefore , one can not expect a significant improvement on @xmath63 limits from the sk+k2k combination .
concerning @xmath2 , the no - oscillation limit @xmath64 is still allowed at 99% c.l . , while values around @xmath65 ev@xmath4 are excluded ( for large mixing ) .
such values correspond to the first ( deep ) oscillation minimum in fig . 1 , which is disfavored by the data at @xmath50 , as previously noted .
therefore , in the sk+k2k combination , we expect `` high '' values of @xmath2 to be more disfavored than `` low '' values . the lower left panel in fig .
2 shows the combination of sk and k2k data .
the best - fit points are located at the same values of @xmath63 as for sk alone and , in general , the bounds on @xmath63 are not significantly modified , as expected . the best - fit value of @xmath2 is only slightly lowered ( @xmath66 ev@xmath4 ) but , most importantly , the 90% and 99% c.l
. ranges of @xmath2 are appreciably reduced both from below and ( more strongly ) from above .
such results show that the k2k experiment is already having a nonnegligible impact in the determination of the neutrino squared mass difference @xmath2 relevant for the @xmath0 channel and for the leading oscillations in ( very ) long baseline experiments . the lower right panel in fig .
2 shows a prospective sk+k2k combination , with the same sk data and k2k data but with k2k total uncertainty hypothetically reduced by a factor of two .
the @xmath2 range is significantly narrowed , while the @xmath63 range is still basically unchanged . in summary ,
the various panels of fig . 2 demonstrate that k2k data are relevant for the determination of @xmath2 , rather than of @xmath63 .
therefore , it makes sense to discuss in more detail the results of fig . 2 in terms of @xmath2 only , with @xmath63 unconstrained .
figure 3 shows the results of such an exercise , in terms of the function @xmath67 .
the 90% and 99% c.l .
ranges for @xmath2 ( @xmath68 ) are explicitly shown for the sk and sk+k2k fits .
numerically , we find that the 99% c.l . range for @xmath2 allowed by sk only , @xmath3\times10^{-3}$ ] ev@xmath4 , is reduced to @xmath5\times 10^{-3}$ ] ev@xmath4 by including k2k data . by halving the uncertainties of the k2k total rate , the @xmath2 range would be ulteriorly reduced to @xmath6\times10^{-3}$ ] ev@xmath4 ( 99% c.l . ) .
the reduction of the minimum values of @xmath2 in fig . 3 ( at a given c.l . ) is relevant for future ( very ) long baseline experiments .
for instance , in the opera experiment @xcite downstream the cern - to - gran sasso neutrino beam , the @xmath69 appearance rate is approximately proportional to @xmath70 @xcite ; therefore , the increase of @xmath71 from @xmath72 ev@xmath4 ( sk ) to @xmath73 ev@xmath4 ( sk+k2k ) at 99% c.l
. implies an increase of the minimum expected @xmath69 event rate by a factor @xmath74 ( at the same c.l . ) ; the increase would be as high as a factor @xmath75 for the case with halved k2k errors ( @xmath76 ev@xmath4 ) .
the reduction of the @xmath2 allowed range is also important to refine the energy or baseline optimization in future neutrino factory experiments , where the current @xmath2 uncertainty plays an important role ( see , e.g. , @xcite and references therein ) .
finally , the increase of @xmath71 strengthens the accuracy of approximations based on the hierarchy of squared mass differences , which have long been used in global solar+terrestrial neutrino analyses ( see , e.g. , @xcite ) .
a final remark is in order .
our current analysis is based on initial k2k data and on an approximate description of the k2k detector specification .
therefore , it can not be a substitute of the ( joint ) official oscillation analysis that will be performed by the k2k ( sk+k2k ) collaboration(s ) .
however , we think that our results , although necessarily approximate , are sufficiently indicative of the k2k potential in improving our current knowledge of @xmath0 oscillations .
we have shown that , within the simplest ( two - family ) @xmath0 oscillation scenario , the initial evidence for a @xmath7 flux suppression in k2k is perfectly consistent with the atmospheric @xmath7 flux anomaly .
the range of the neutrino squared mass difference indicated by the sk atmospheric @xmath77 experiment is reduced by including the k2k data .
as a consequence , k2k starts to be important in narrowing the range of predictions for future ( very ) long baseline experiment .
we thank t. hasegawa and t. maruyama for very useful information about the k2k experiment . e.l .
thanks the organizers of the taup conference , where preliminary results of this work were presented , for kind hospitality .
this work was supported by the italian istituto nazionale di fisica nucleare ( infn ) and ministero dellistruzione , universit e ricerca ( miur ) within the `` astroparticle physics '' project .
t. hasegawa , in the proceedings of _ taup 2001 _ , 7th international workshop on topics in astroparticle and underground physics ( laboratori nazionali del gran sasso , italy , sept .
2001 ) , to appear .
transparencies available at taup2001.lngs.infn.it . fig . 1 .
number of events in k2k for the current exposure ( @xmath78 p.o.t . ) , as a function of @xmath2 ( at @xmath79 ) .
our calculation ( solid curve ) is benchmarked by the k2k mc simulation ( square markers ) .
the horizontal gray band represents the current k2k data within one standard deviation .
fig . 2 .
separate two - flavor oscillation analyses of sk and k2k , together with their combination ( with eventual halving of k2k errors ) .
see the text for details .
fig . 3 . bounds on @xmath2 at 90% and 99% c.l .
( @xmath68 ) from the two - flavor @xmath35 analysis of the sk and k2k data . | we briefly discuss the impact of initial data from the kek - to - kamioka ( k2k ) neutrino experiment on the @xmath0 oscillation parameters ( @xmath1 ) currently indicated by the super - kamiokande ( sk ) atmospheric neutrino experiment . after showing the very good agreement between k2k and sk
, we combine the two separate pieces of information .
we find that the 99% c.l.range for @xmath2 allowed by sk only , @xmath3\times10^{-3}$ ] ev@xmath4 , is reduced to @xmath5\times 10^{-3}$ ] ev@xmath4 by including k2k data . by halving the uncertainties of the k2k total rate ( with central value unchanged ) ,
the @xmath2 range would be ulteriorly reduced to @xmath6\times10^{-3}$ ] ev@xmath4 .
such information appears to be already useful in planning ( very ) long baseline neutrino oscillation experiments . |
circumstellar disks are an ubiquitous outcome of the stellar formation process and they are believed to be the birth place of planetary systems .
the growth of dust particles towards planetesimal sizes along with their vertical settling due to gas drag are processes that are believed to be the first steps towards planet formation .
hydrodynamical models have shown that these processes can be efficient early in the disk evolution ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . to test these models , it is necessary to obtain an observation - based description of the structure and dust content of protoplanetary disks as a function of the age of the system and other relevant parameters ( e.g. , stellar mass ) .
the dust component of protoplanetary disks has long been studied via its thermal emission from near - infrared to millimeter wavelengths which is frequently associated with low - mass pre - main sequence ttauri stars @xcite . both grain growth and dust
settling can alter the overall shape of the spectral energy distribution ( sed ) of a ttauri star @xcite .
indeed , several studies that analyzed ( elements of ) the sed of young stars have concluded that both grain growth and settling is occuring in protoplanetary disks ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) .
unfortunately , such studies suffer from the absence of spatial information inherent to photometric measurements and the high optical thickness of disks in the near- to mid - infrared regime . as a result , comparing an object s sed to radiative transfer models leaves many ambiguities @xcite .
for instance , the inferred total dust mass and the maximum size of the dust grains are inversely correlated because of the dependency of dust opacity on grain size . in addition , most of these studies , which focus on a single type of observations ( e.g. , millimeter fluxes , silicate emission feature ) , only probe a limited region of the disk and a small fraction of the entire grain size distribution . to solve for the ambiguities inherent to sed studies , it is critical to obtain spatially - resolved observations .
such observations include thermal emission mapping with ( sub)millimeter interferometers ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and scattered light imaging with optical and near - infrared high - resolution instruments ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
disks are generally optically thin at long wavevengths , so the former type of observations can probe the entire disk structure . furthermore , they are very sensitive to the presence of millimeter - sized particles .
on the other hand , scattered light images , which only probe dust grains at the disk surface , are very sensitive to the size distribution of micronic grains especially when images at multiple wavelengths are analyzed simultaneously ( * ? ? ? * and references therein ) .
both types of observations have already yielded important pieces of evidence supporting both grain growth and dust settling in disks ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
while grain growth and settling appear to occur in protoplanetary disks , detailed _ quantitative _ tests of hydrodynamical models can only be achieved with a detailed view of the entire structure of a disk .
this can only be obtained via a multi - technique , panchromatic approach .
unfortunately , observational and computational limitations have so far limited the number of objects for which such an analysis could be conducted to a handful .
the most notable examples are the studies of the `` butterfly star '' @xcite , imlup @xcite and iras04158 + 2805 @xcite . in the former two cases ,
these studies have unambiguously shown that the dust population is stratified , possibly indicating that dust settling is already occurring . increasing the number of disks studied in such detail is necessary to disentangle individual peculiarities from genuine trends associated with disk evolution .
hv tau is a triple system located in the taurus star - forming region .
it consists of a 550au wide pair whose optically brightest component is itself a tight ( 10au ) visual binary @xcite .
spectroscopic and photometric measurements revealed that this subsystem does not currently experience accretion nor does it show infrared excess .
they further establish an age of about 2myr for the system @xcite .
the third component of the system , hv tau c , is much fainter yet bluer than hv tau ab . while @xcite first thought that this source was an herbig - haro object , @xcite later proposed that hv tau c is a normal m0 ttauri star surrounded by an opaque edge - on disk similar to that found in hh30 by @xcite .
subsequent high resolution imaging confirmed this hypothesis @xcite . @xcite
produced the first model of high resolution 0.8 and 2.2@xmath1 m scattered light images of hv tau c , finding that dust properties similar to those of interstellar dust grains can account for these images .
their 0.8@xmath1 m image also revealed the presence of a roughly spherical envelope , producing a symmetric halo that is more extended than the disk itself , that is likely the remnant of the core from which the system formed .
as a consequence of their particular viewing geometry , edge - on protoplanetary disks offer a unique opportunity to determine their geometry and dust content . as such , they may be the best candidates to study vertical stratification in protoplanetary disks .
they also are comparatively easy targets for high - angular resolution instruments since the contrast requirement is strongly relaxed . on the other hand , they are challenging from the modeling point of view because of the difficulty of accurately solving for radiative transfer including anisotropic scattering in high optical depth regions .
previous modeling efforts of edge - on protoplanetary disks have therefore focused on interpreting one type of observation at a time @xcite .
while these studies proved highly valuable to constrain some of the disk properties , no self - consistent model was used to model all data at once , leaving unexplained contradictions . our objective in this work is to perform a global analysis of the hv tau c disk , combining scattered light images , millimeter interferometric data and the overall sed into a single fit .
we present a series of new observations of the system in section [ sec : obs ] and discuss our empirical results in section [ sec : results ] . in section [ sec : models ] , we present radiative transfer models of the hv tau c disk and discuss their implications in section [ sec : discus ] .
section [ sec : concl ] summarizes our results .
on 2002 november 24 , we observed hv tau using the naos adaptive optics system and the conica instrument @xcite installed on the yepun 8.2m - unit telescope at eso s very large telescope , as part of the naos guaranteed time observing program .
we used the 00133 pixel scale for the @xmath4 ( @xmath5@xmath1 m , @xmath6@xmath1 m ) and @xmath7 ( @xmath8@xmath1 m , @xmath9@xmath1 m ) images and the 00270 plate scale for the @xmath10 ( @xmath11@xmath1 m , @xmath12@xmath1 m ) images .
hv tau ab , a @xmath13 source , was used as adaptive optics guide star with the visible wavefront sensor . from narrow - band observations of single stars throughout the night ,
the measured fwhm of point - like sources is about 010 , 007 and 008 at @xmath4 , @xmath7 and @xmath10 band , respectively .
these images , similarly , have strehl ratios of approximately 5% , 25% and 40% .
we obtained deep images in which hv tau ab is saturated , as well as shallow images with an additional neutral density filter and shorter exposures to record unsaturated images of the primary .
total integration times for the long exposures of 750s , 480s and 165.5s were recorded at @xmath4 , @xmath7 and @xmath10 , respectively , split in 8 to 20 dithered independent images .
total integrations times for the shallow images were 100s , 50s and 45s , respectively at @xmath4 , @xmath7 and @xmath10 .
for each sequence of images , a sky was estimated by medianing all images , which was then subtracted from each image prior to cosmetic cleaning , which included bad pixels and cosmic ray correction , and flat - fielding .
all resulting images were then shift - and - added to produce final images .
on 2002 november 26 , we re - observed hv tau with naos and conica using the 00270 pixel scale , this time using the 14-diameter coronagraph mask to block out the starlight from hv tau ab and record deeper exposures on hv tau c. three 120s @xmath10 band images were recorded . a sky was subtracted off each image before they were cosmetically cleaned and flat - fielded .
finally , they were averaged to yield the final coronagraphic image .
no point source is detected in the 30 field - of - view , preventing us to estimate the achieved fwhm and strehl ratios . on 2002 december 13 ,
we observed hv tau using the nirc2 camera ( p.i . k. matthews ) installed behind the keckii adaptive optics system @xcite to record images using the @xmath14 ( @xmath15@xmath1 m , @xmath16@xmath1 m ) and @xmath17 ( @xmath18@xmath1 m , @xmath19@xmath1 m ) filters and the 000996 plate scale @xcite .
hv tau ab was used as a natural guide star ( ngs ) for the adaptive optics system . to reduce thermal background and provide a more symmetric and smoother psf
, we used the `` inscribed circle '' pupil stop , resulting in an effective primary mirror diameter of 9 m . at @xmath14 , four images , consisting of 300 coadded 0.181s individual integrations each ( for a total of 217.2s on source integration ) , were acquired with the sources dithered on - chip between images .
a fifth image was acquired with all stars moved out of the detector field - of - view ( 10 ) to estimate the sky level . at @xmath17
, 500 individual 0.150s integrations were coadded using a reduced 6-wide field - of - view , necessary to avoid background saturation on the intense thermal background .
a total of 21 such images were obtained with the stars moved about in the available field - of - view , representing a total integration time of 1575s .
the corresponding sky was estimated by median - combining all frames .
on 2004 november 03 , we re - observed hv tau with nirc2 , this time using the newly available laser guide star ( lgs ) module on the keckii adaptive optics system @xcite .
the laser system was run at 6 watts output power , which produced a guide star with an equivalent magnitude of @xmath20 , and hv tau ab was used as the tip / tilt correction point source . to take full advantage of the better image quality , the `` largehex '' pupil mask , which does not block any section of the primary mirror , was used throughout the observations .
hv tau was imaged in the @xmath14 filter with an 0.2s integration coadded 100 times .
this observing cycle was repeated 27 times , with the system alternatively located in opposite detector quadrants , providing a total integration time of 540 secs . for each image , the subsequent one , with the star located in the opposite quadrant , was used as a sky frame .
all datasets were reduced using a similar strategy to that used in processing the naco images ( see also * ? ? ? * ) .
first of all , the sky thermal emission was subtracted .
these subtracted images were then flat fielded and had any bad pixels interpolated over . in the case of the lgs dataset ,
a residual sky level in the frames was measured by taking the median value of the quadrants which do not have a star in the field .
this value ( typically corresponding to 0.05% of the initial sky value ) was then subtracted off the cleaned images .
all cleaned images in a given dataset were then median - averaged .
we measured fwhms of 009 and 011 for point - like sources observed before / after hv tau at @xmath14 and @xmath17 for the ngs dataset and 008 at @xmath14 for the lgs dataset .
we estimated the corresponding strehl ratios to be about 70% at @xmath14 ( both in ngs and lgs mode ) and 80% at @xmath17 , using the `` strehl meter '' tool developed by the keck observatory .
given the high thermal background at @xmath14 and @xmath17 , the resulting signal - to - noise on hv tau c is limited , about 15 , 20 and 5 in the peak pixel at @xmath14 ( ngs , lgs ) and @xmath17 , respectively . to enhance detection without altering the intrinsic spatial resolution of the images , we smoothed them by using a running 2-pixel - radius ( @xmath21002 ) median - filtering circular mask , and rebinned them by a factor of 3 in both directions , resulting in an approximate final sampling of 003/pixel that still oversamples the resolution of the datasets . on 2005 february 26
, we observed hv tau with iram s 6-antenna plateau de bure interferometer ( pdbi , * ? ? ?
* ) in the 6bp configuration , with baselines ranging from 71 m to 331 m .
simultaneous 110ghz ( 2.76 mm ) and 230.5ghz ( 1.31 mm ) observations of hv tau were recorded in double - sideband mode with a bandpass of 640mhz at each frequency .
hv tau was observed alternatively with phase calibrators 0415 + 379 and 0528 + 134 through an entire 11h - transit , resulting in beam sizes of 11@xmath2209 ( along position angle 38 ) and 21@xmath2216 ( along position angle 62 ) at 1.3 and 2.8 mm , respectively .
the average weather conditions resulted in r.m.s .
phase noises on the order of 1540 at 2.8 mm and 815 at 1.3 mm ( using self - calibration from the 2.8 mm data ) , equivalent to an atmospheric `` seeing '' on the order of 03 and 06 at 1.3 and 2.8 mm , respectively .
the absolute pointing uncertainty is on the order of 0102 .
the quasars nrao150 and 3c273 were used as absolute flux calibrators , resulting in a 10% uncertainty on all quoted fluxes .
the data were reduced using the gildas package and selecting individual baseline visibilities for which the phase noise was less than 40 and flux variations were less than 20% based on calibrators measurements .
simultaneous observations of the @xmath2co 21 transition ( 230.538ghz rest frequency ) with a 20mhz bandpass were obtained to probe the gaseous component of the disk .
data reduction followed the same method as the continuum data , providing a 3-dimension reconstructed datacube with a 0.1km.s@xmath23 spectral resolution .
we obtained follow - up observations of hv tau with the 15-antenna carma array in its c configuration , with baselines ranging from 24 to 300 m . on 2008
may 6 and 9 , we tuned the receivers to a central frequency of 110ghz ( 2.76 mm ) , while on 2008 may 29 we tuned them to 90ghz ( 3.33 mm ) .
the total continuum bandwidth is 2.8ghz , split in 6 separate bands .
observing conditions were average and parts of the observations had to be flagged out because of poor phase coherence .
overall , the useful integration times on hv tau were 3h45 at 110ghz and 2h30 at 90ghz .
observations of hv tau were interleaved with pointings at 3c111 and j0530 + 135 which served as phase calibrators ; flux calibration was performed by observing a planet ( uranus , neptune ) at the beginning of each track .
the systematic uncertainty in carma s absolute flux scale is @xmath2420% ( w. kwon , priv .
the data were reduced using the miriad software package .
the final 110ghz map corresponds to the combination of both observing periods , and is characterized by a beam size of 18@xmath2214 ( along position angle 116 ) ; the 90ghz map has a beam size of 31@xmath2217 ( along position angle 124 ) . on 2008
january 28 , we observed hv tau at 350@xmath1 m with the 32@xmath2212 sharc - ii bolometer array @xcite installed at the caltech submillimeter observatory .
the array was scanned at an 82/s rate over a 60@xmath22 field centered on hv tau .
this scanning was repeated until a total integration of 600s was achieved .
ceres and hltau , two secondary flux calibrators , were observed immediately after with total integration times of 180s and 300s , respectively .
the fluxes of both these sources are known to within 10% ; their measured fluxes agreed to within this uncertainty .
conditions were excellent for submillimeter observations ( @xmath25 ) .
all data were reduced using the cso - developed crush software , using a 6 smoothing for hv tau ( a 4 smoothing was used for the flux calibrators ) , to produce a final image that has an effective resolution of approximately 10 .
at all wavelengths , our adaptive optics images clearly reveal the typical morphology of hv tau c as an opaque edge - on disk , namely two parallel , horizontal nebulae separated by a dark lane ( figure [ fig : newimgs ] ) .
they further show hv tau ab to be systematically extended along the north - western direction , although the pair is barely resolved due to its tight separation ( see figure [ fig : orbit ] ) . here
we discuss the relative astrometry of all 3 components .
focusing first on the wide pair hv tau c hv tau ab , we estimate its separation and position angle based on the location of the centroid of both components . despite saturation of hv tau ab in some images , all images yield consistent estimates ( see table [ tab : binary ] ) , with an average separation of 404@xmath26002 and a position angle of 446@xmath2606 .
there is marginal ( @xmath27 ) evidence for a variation of the binary position angle as a function of wavelength , though this could be due to underestimated uncertainties on the absolute orientation of the detectors .
we do not find any significant trend as a function time at our level of precision . based on the proper motion of hv tau ab measured by @xcite
, the projected separation and position angle of the pair would have changed in the almost 5 years between the observations of @xcite and ours by a total of 012 and 22 , respectively , if hv tau c had not been comoving with hv tau ab .
this can be rejected at the @xmath28 level in our datasets , and is further confirmed by the relative astrometry obtained at earlier epochs by @xcite and @xcite .
we therefore conclude that hv tau is a common proper motion pair with a projected separation of 565au , making it a bona fide triple system .
to obtain reliable relative photometry and astrometry for the tight hv tau ab pair , we used psf - fitting .
we choose to analyze the short @xmath7- and @xmath10-band vlt exposures and both @xmath14-band keck images , as they offer the best image quality and the most favorable ratio between binary separation and achieved resolution .
for the vlt @xmath7-band and keck @xmath14-band images , we searched for adequate psfs among images of single stars obtained on the same night as our hv tau images . for the vlt @xmath10-band
, we used a set of single stars from observations taken with the same set - up presented in @xcite .
psf - fitting was performed using iraf s _ daophot _ package .
the resulting relative astrometry and photometry is listed in table [ tab : binary ] .
the vlt and keck ngs data have been taken within a few weeks of each other , and no measurable orbital motion is expected over such a short timescale . averaging these three datasets , we find a separation of 00595@xmath2600025 and a position angle of 3125@xmath2618 .
the keck lgs data , obtained two years later , yields a relative astrometry that is marginally consistent ( @xmath29 ) with this estimate .
the projected separation in our images appears to be 45@xmath30 smaller than earlier measurements @xcite , but there is no significant change in position angle as a function of time ( see figure [ fig : orbit ] ) .
this relative displacement is most likely due to orbital motion .
the observed plane - of - the - sky velocity of hv tau ab , @xmath211.5km.s@xmath31 , is roughly one order of magnitude too low considering the 10au projected separation of the binary compared to other ttauri binaries ( e.g. , * ? ? ? * ) .
the number of resolved measurements and the total amplitude in orbital motion are insufficient to attempt an orbital fit for hv tau ab at this point .
nonetheless , the unexpectedly low measured orbital velocity suggests a highly elliptical orbit observed around apoastron passage , or a large out - of - the - plane separation which , combined with the nearly radial observed motion , would in turn imply that the orbital plane is almost perpendicular to the plane of the sky .
monitoring of the system in the next few years will help disentangle these two possibilities .
our psf fitting also yields relative photometry for the hv tau ab binary . considering the near - simultaneous @xmath7 , @xmath10 and ngs @xmath14 images
, we find evidence that hv tau b is somewhat redder than hv tau a in the near - infrared ( see table [ tab : binary ] ) . in the framework in which none of the components possesses circumstellar material ,
this is an indication that hv tau b is cooler than hv tau a , consistent with it being fainter .
surprisingly , @xcite found a flux ratio in the visible that is closer to unity than our near - infrared measurements .
temporal variability of either component could be an explanation , although we note that both our @xmath14 flux ratios are consistent with one another despite being taken two years apart .
further monitoring is required to better understand the intrinsic colors of both components .
the near - infrared images presented here are of higher spatial resolution than those presented in @xcite and @xcite , which both had a 013 resolution , and are comparable to the @xmath32 subaru images of @xcite .
our observations extend the wavelength coverage of hv tau c to @xmath17 for the first time and , combined with the hst f814w image of @xcite , offer an almost uniform spatial resolution ( 007011 , or 1015au ) view of the disk over the entire 0.84.7@xmath1 m range .
this is critical to conduct unbiased studies of the wavelength dependence of the disk images . to quantify the basic morphological properties of hv tau c
, we adopt the following method . at each wavelength
, we first estimate the position angle of the dark lane , @xmath33 , as the average of the position angles of each nebulae , as determined from fits of elliptical gaussian intensity profiles . averaging all estimates
, we find a position angle of 1083@xmath2604 , which we take as the orientation of disk midplane .
the dark lane width , @xmath34 , is measured as the projection of the vector joining the light centroid of the two nebulae on the disk minor axis .
peak - to - peak flux ratios ( @xmath35 ) are readily estimated , whereas integrated flux ratios ( @xmath36 ) are obtained by summing the flux within areas that encompass all pixels whose surface brightness is at least 5% ( 20% for the @xmath17 image ) of the peak surface brightness in the image .
finally , we measure the total extent of the disk along its major - axis , @xmath37 , defined by the horizontal extent of the contour at the 5@xmath30 noise level . we note that the spherical halo identified by @xcite dominates this measurement in the f814w image .
all of these quantities are given in table [ tab : observables ] .
several wavelength - dependent features can be noted in our images of hv tau c. as figure [ fig : contours ] illustrates , @xmath34 decreases by about 30% from 0.8 to 4.7@xmath1 m .
we further find that both @xmath35 and @xmath36 smoothly decline towards longer wavelengths without ever reaching unity ; rather , both flux ratios actually plateau longwards of 2.2@xmath1 m . finally , the intensity profile of the bright nebula along the disk major axis is remarkably invariant from 0.8 to 4.7@xmath1 m ( see figure [ fig : profiles ] ) .
the measured fwhm is about 035 at all wavelengths , i.e. , a factor of 34 larger than the resolution of our images .
this achromatic behavior is therefore not influenced by observational limitations but is rather intrinsic to the disk . one main feature of the hv tau c disk that was pointed out by @xcite is the lateral asymmetry : the northeastern ( bright ) nebulae extends further on one side while the southwestern ( faint ) nebula extends further on the other side .
this can be readily seen in figure [ fig : contours ] .
in addition to this low - intensity asymmetry , the location of the nebulae centroids is also non symmetric , suggesting a global asymmetry : the line joining the two centroids is misaligned by about 10 with respect to the disk symmetry axis as determined by the orientation of the dark lane .
we do not find any significant variation of the asymmetry with wavelength nor time within our uncertainties .
@xcite also discussed narrow structures , which they dubbed `` rays '' , extending along the disk major axis beyond the main / bright nebula which they associate to the disk , with total extent up to 14 in their optical images .
we can track scattered light up similar distances at both @xmath10 and @xmath14 in our images , a regime in which the halo is undetected .
although the exact nature of these features remains uncertain , this suggests that these `` rays '' trace the surface of the disk instead of being a mere shadow of the disk on the spherical halo . in that case
, the actual disk size may be larger than the 50au assumed in the past .
both our pdbi and carma observations have a sufficient spatial resolution to disentangle the wide hv tau pair . in all maps ,
a single source is detected , at the location of hv tau c. hv tau ab appears to have no significant emission at millimeter wavelengths with 3@xmath30 upper limits in our pdbi data of 6mjy and 1.5mjy at 1.3 and 2.8 mm , respectively . at 1.3 mm ,
hv tau c is spatially resolved in our pdbi data , with a sharp decrease in correlated flux at the longest baselines along the position angle of the disk as defined from the scattered light images ( see figure [ fig : fcor ] ) .
we therefore fit an elongated gaussian profile to the visibilities , assuming that its minor axis is unresolved which is consistent with the data .
this yields a total flux density of [email protected] , a fwhm of 117@xmath26008 along the major axis , at pa 111@xmath263 east of north , in excellent agreement with the position angle derived from the scattered light images .
this analytical fit is shown in figure [ fig : fcor ] ; deviations between observations and our simple fit result from the fact that the distribution of surface brightness in the disk is not a perfect gaussian .
the measured 1.3 mm flux agrees well with the previous single - dish measurement by @xcite . at 2.8 mm ,
the source is unresolved with the pdbi , with a flux density of [email protected] .
the carma observations do not resolve the disk either , and we extracted point - source fluxes of [email protected] at 2.8 mm and [email protected] at 3.3 mm .
flux calibration uncertainties of 10% and 20% must be added for the pdbi and carma observations , respectively .
a single point - like source is detected in our 350@xmath1 m cso map of hv tau with a flux of [email protected] ; a 10% uncertainty for flux calibration must be quadratically added .
the resolution of this dataset , roughly 10 , does not allow us to spatially resolve the wide pair .
however , taking advantage of the strong detection of the system ( @xmath38 ) , we use the nearly simultaneous observation of the bright source hltau with the same set - up to determine an accurate astrometric positioning @xcite .
the 350@xmath1 m source is found to be located at ( 04h38m35.50s , + 2610406 , j2000 ) with an uncertainty of about 1 in both directions .
this is in excellent agreement both with the position of the lone millimeter source in our pdbi and carma maps and with the location of hv tau c in optical and near - infrared images .
we therefore conclude that hv tau c is the dominant source of emission at 350@xmath1 m and assign all of the observed flux to that component .
as shown in figure [ fig : co ] , @xmath2co 21 line emission is detected at the position of hv tau c in our pdbi observations .
the integrated spectrum shows two distinct peaks , with a near - zero trough separating them .
the trough in the middle of the line is likely due to contamination by large - scale emission from the surrounding molecular cloud that is filtered out by the interferometer .
@xcite detected @xmath39co emission in the 5.57 km s@xmath23 from the molecular cloud at the location of hv tau , consistent with this interpretation .
we further find that the blue- and red - shifted parts of the line emission are spatially distinct and symmetric about the continuum peak .
the two line emission peaks are separated by about 15 , or 200au , spatially and 4km.s@xmath23 in velocity .
finally , we note that the co emission appears to extend beyond both the millimeter continuum and the scattered light images , suggesting that there is more to the system than meets the eye in continuum datasets . to draw a complete picture of the hv tau c disk , we compiled its sed by combining published fluxes with our new measurements .
from the same references ( see below ) , we also built the sed of hv tau ab ( figure [ fig : sed ] ) , which is well reproduced from the optical to the mid - infrared by a 3600k , @xmath40 , nextgen model from @xcite assuming @xmath41mag , in agreement with previous spectrophotometric estimates of the stellar properties of that component @xcite . constructing the sed of hv tau
c is a delicate matter because of the large scatter in its published optical and near - infrared photometry ( see also * ? ? ?
* ; * ? ? ?
figure [ fig : sed ] includes all published data and illustrates this variability . to construct a `` representative '' sed for hv tau c
, we elected to aim for smoothness .
for instance , we adopt the most recent @xmath42 photometry from @xcite , which matches well with the _
spitzer_/irac data from @xcite .
we discard the @xmath43 band measurements from @xcite as they appear to overestimate the flux of both components by about 1mag . based on their agreement at @xmath44 band with the mccabe et al .
photometry , we then adopted the @xmath45 fluxes from @xcite .
finally , we adopted the visible photometry from @xcite which offers a smoother extension of the sed to short wavelengths than that of @xcite , although it was obtained almost a decade earlier .
apart from the @xmath44 band measurement by @xcite , our selection is equivalent to consistently selecting the brightest measurement available at every wavelength . at longer wavelengths , we used the mips 24@xmath1 m and 70@xmath1 m detections and a 170@xmath1 m upper limit from the _ spitzer _ taurus legacy survey ( rebull et al . ,
submitted ) . at 24@xmath1 m
the astrometric accuracy is good enough to identify the detected source with hv tau c with little or no contribution from hv tau ab . in the longest wavelength regime , we adopt fluxes measured from our data and from @xcite .
the resulting sed , the most complete to date for an edge - on disk system , is shown as filled diamonds in figure [ fig : sed ] and presented in table [ tab : photom ] .
the sed of hv tau c can be characterized as a `` double hump '' ( see figure [ fig : sed ] ) , which is typical of other edge - on disks ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the short wavelength hump is dominated by photons emitted from the central star and the innermost region of the disk and scattered to the observer at the outer radius of the disk , whereas the long wavelength part represents the disk thermal emission propagated through the mostly optically thin disk to the observer .
the wavelength of the turnover between the two humps is driven by the total column density of dust along the line of sight to the central star .
we find this turnover to be between 8 and 11.8@xmath1 m although a more precise estimate would require analysis of a mid - infrared spectrum of the source . at longer wavelengths
, @xcite noted that hv tau c has an abnormally flat [email protected] mm slope . with our new observations
, it appears that the 850@xmath1 m flux is indeed too low compared to extrapolation from longer wavelength fluxes . limiting our analysis to the 1.33.3 mm regime
, we find a spectral index of @xmath46 , where @xmath47 .
this value is on the low end of the distribution observed for other protoplanetary disks @xcite .
here we interpret some of the observational results listed above to make qualitative inferences about the hv tau c system .
we will then revisit in a quantitative way these conclusions on the basis of the radiative transfer modeling presented in section [ sec : models ] .
the @xmath481mag variability of hv tau c in the near - infrared is larger than has been recorded for other ttauri stars in the past @xcite .
it is therefore unlikely that this variability results from intrinsic fluctuations in the emission of the star and inner disk .
a plausible alternative scenario could invoke a self - shadowing pattern in the disk that moves as a result of differential keplerian rotation . in this framework
, the structure responsible for the lateral asymmetry in the images would be located a few au away from the central star to account for the absence of variability over the course of 2 years .
this could be linked to the observed asymmetry in the scattered light images .
such a phenomenon has been documented in the hh30 edge - on disk @xcite .
one possibility is that the disk is warped as a consequence of the gravitational forces exerted by hv tau ab . the wavelength dependent features of the scattered light images inform us about the absorption and scattering properties of the dust grains in the disk .
for instance , the decreasing @xmath49 towards longer wavelengths is a consequence of the declining dust opacity @xcite .
the fact that the intensity profile along the bright nebula does not change with wavelength is suggestive of a dust population whose phase function is equally anisotropic at all wavelengths . a dust population similar to that of the interstellar medium , whose scattering asymmetry parameters drops rapidly beyond 1@xmath1 m ( e.g. , * ? ? ? * ) , seems inconsistent with our observations of hv tau c. we note , however , that the flux ratio between the nebulae is not constant across the 0.84.7@xmath1 m range , even though it is also believed to be driven by the scattering asymmetry .
the slope of the sed of an edge - on disk between the near- and mid - infrared is a function of the disk geometry and of the grain size distribution and composition in the disk scattering surface . in their analysis of the `` flying saucer ''
edge - on disk , @xcite concluded that the shallow 210@xmath1 m slope in that system implies the presence of a `` significant amount of 510@xmath1 m grains '' in that disk .
such grains are necessary to produce a high albedo in the mid - infrared . while it is tempting to apply this argument to hv tau c , uncertainties about the disk geometry prevents from reaching a firm conclusion just yet .
if the disk were optically thin to its own emission in the millimeter regime , the observed spectral index could be readily converted into an opacity power law index of @xmath50 , characteristic of evolved dust grain populations that extend up to a few millimeter in size @xcite . however , because the disk is very compact and viewed almost exactly edge - on , the optically thin assumption may be incorrect , so that grain growth can not be claimed on this sole basis in this particular system .
if the disk were indeed optically thick , one would expect the millimeter spectral index to steepen at longer wavelengths .
we find @xmath51 and @xmath52 , a difference that is not statistically significant .
flux measurements at even longer wavelengths are ultimately needed to conclude on this possibility . in any case
, there are several independent hints of the presence of grains at least a few microns in size , and maybe up to millimeter sizes , in the hv tau c disk . however , none of these inferences are robust enough as they depend on some assumptions about the disk geometry , among other factors .
the goal of the modeling presented in the next section is to simultaneously fit for the disk structure and the dust properties so as to solve for these ambiguities . in the absence of a gas tracer
whose emission from the cloud is negligible , we do not perform a complete keplerian rotation analysis of our @xmath2co observations .
qualitatively , _ the measured amplitude of the velocity and positional offsets are consistent with an 0.51@xmath3 central star _ ( see figure [ fig : co ] ) .
observations in other molecular tracers are necessary to go beyond this simple analysis , however .
we also note that the larger outer disk radius suggested by the gas emission is not a unique property of the hv tau c disk .
indeed , this phenomenon is rather common among protoplanetary disks ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , although few objects have been mapped at very high resolution in both the continuum and line emission .
@xcite have recently demonstrated that this can be reproduced if the surface density profile of disks is tapered , rather than sharply truncated , outside of a certain radius , for instance .
based on the gas emission alone , the disk outer radius might be as large as 100au , twice the estimate from the continuum and scattered light images .
since our entire analysis focuses on the dust component of the disk , we do not consider such large disk radii in our modeling .
the objective of this section is to compare our broad set of observations of hv tau c to predictions of a radiative transfer code in order to constrain the main properties of the disk . since this is the first attempt at simultaneously reproducing scattered light images , thermal emission maps and the sed of hv tau c , our approach consists in considering a simplified parametrized disk structure in an effort to search for a model that would represent a reasonable global fit to , rather than an exact representation of , all observations of this disk .
we use the mcfost radiative transfer code @xcite which computes synthetic observables , such as seds and images , by propagating photon packets through the disk .
scattering , absorption and reemission by dust grains are taken into account following the mie theory valid for homogeneous spherical grains .
we assume that the dust grains are at the local thermal equilibrium with the surrounding radiation field throughout the disk .
synthetic temperature maps , seds and images are computed simultaneously at all inclinations , allowing us to fit for that parameter as well . for each model ,
the temperature , seds and images computation are performed using typically 1 , 4 and 32 million packets , respectively . the disk is assumed to be passive , i.e. , its only source of heating is radiation from the central star .
the disk geometry is described by its inner and outer radii , @xmath53 ( which we fix at 0.15au ) and @xmath54 , as well as by two independent power laws .
the surface density profile of the disk follows @xmath55 .
vertically , we assume that @xmath56 $ ] , appropriate for a vertically isothermal , non - self - gravitating disk in hydrostatic equilibrium .
finally , we adopt a flaring law for the scale height , namely @xmath57 .
the dust content of the disk is characterized by a power law size distribution , @xmath58 from @xmath59 to @xmath60 .
we adopt the optical properties of the commonly - used `` astronomical silicate '' mixture from @xcite with @xmath61 and @xmath62@xmath1 m . in all our models ,
we adopt a default distance of 140pc , consistent with the system s parallax estimated by @xcite .
the central star is described by an effective temperature of 3800k based on it spectral type @xcite , and we use the corresponding @xmath40 nextgen synthetic spectrum @xcite .
although @xmath63 is difficult to constrain in hv tau c since we can not measure the total bolometric luminosity of the object , we consider it as a free parameter since there is no simple method to fix it a priori .
@xcite showed that adding an envelope to the system yields a much improved fit to the visible scattered light images of hv tau c , especially to account for the roughly spherical halo that is detected well above the disk midplane . to implement this
, we add a spherically symmetric envelope to our model , which is characterized by @xmath64au and @xmath65au ( maximum extent of the rays identified by * ? ? ?
* ) to mimic the halo seen in the hst images .
for this envelope , we use a total dust mass of @xmath66 , a radial density profile @xmath67 and interstellar - like dust grains ( @xmath68@xmath1 m and @xmath61 ) . the optical depth through the envelope at @xmath69@xmath1mis @xmath70 , i.e. the maximum acceptable considering the morphology of the visible images of the disk .
while a spherical envelope has little physical relevance , as its free - fall time would be much shorter than the system age , we do not explore more sophisticated envelope models ( e.g. , rotation - supported ) .
indeed , we consider our approach as a reasonable `` placeholder '' to avoid systematically poor @xmath71 values for our model scattered light images .
images of hv tau c show that the envelope is only significantly detected shortward of 1@xmath1 m , anyway , and even then the region of highest signal - to - noise ratio , which contributes most to the total @xmath71 is dominated by scattering off the disk .
we therefore do not expect our simplistic parametrization to induce any strong bias on our results .
finally , we note that the asymmetry of the disk seen in scattered light image produces such strong deviations from our simple model that attempting to adjust the envelope profile is of little interest at this point .
while we have set a number of parameters of the model , we still have to deal with an 8-dimensional parameter space : the total dust mass in the disk ( @xmath72 ) , @xmath54 , @xmath73 ( defined at @xmath74au ) , @xmath75 , @xmath76 , @xmath60 , @xmath63 , and the inclination to our line of sight ( @xmath77 , where @xmath78 corresponds to a face - on disk ) are all free parameters . when computing synthetic seds , we also added a foreground extinction @xmath79 following the interstellar extinction law which we considered as an additional free parameter , with values ranging from 0 to 5mag .
this extinction represents attenuation by material located around or between the hv tau system and us and does not include attenuation by the edge - on disk and spherical envelope themselves . for all model parameters except the inclination , we adopt a coarse sampling that encompasses reasonable estimates either from the previous modeling by @xcite and @xcite or our own estimates above ( see table [ tab : models ] ) . for instance , we chose a range of values for @xmath60 that extends from 1@xmath1 m to 1 cm .
as shown in figure[fig : opacity ] , @xmath80@xmath1 m results in an opacity law that is in reasonable agreement with the measured interstellar extinction law through the mid - infrared . by extension
, we will refer to this model as `` interstellar dust '' .
the only parameter for which we use a large number of possible values is the inclination .
mcfost produces synthetic disk observables at all inclinations at once .
the final images and sed are stored in several `` inclination bins '' which are equally spaced in cos(@xmath77 ) , appropriate for a random orientation prior . by using 91 independent inclination bins in all our simulations ,
we obtain a 06 sampling in the vicinity of the perfectly edge - on geometry . exploring
the parameter space with such a coarse sampling is computationally manageable , although it comes at the cost of failing to find a model that fits perfectly all data .
our original goal was to use this grid to identify a small region of the parameter space that produced a good fit to all observations and to run a second , finer grid in that smaller regions .
however , as we discuss below , our analysis revealed that no single region of the parameter space could be selected that way , as different observations point towards different parts of the parameter space .
we discuss possible avenues for improvement for future modeling efforts in section [ sec : discus ] .
with 8 free parameters in our model grid , we have computed over a million independent models , which required about 126,000h of cpu time on a 400-processor cluster , 92% of which was devoted to the synthetic sed calculation . because we chose a fixed value for @xmath81 ,
the temperature at the inner edge of the disk varies from 800 to 1300k in the grid .
similarly , the temperature at the outer edge ranges from 15 to 30k .
the radial dependence of the dust temperature in the midplane outside of @xmath82au closely follows a power law , @xmath83 , whose index is in the range @xmath840.6 and mostly depends on the amount of flaring in the disk .
the dust temperature at a given location depends primarily on @xmath85 , hence @xmath63 and , to a lesser extent , on @xmath60 ( via the total dust opacity ) and other geometric parameters .
our modeling strongly favors @xmath86 ( see below ) , for which the range in maximum temperature in the disk is 11001350k , close to the sublimation temperature of dust grains . while setting @xmath81 on the basis of this sublimation temperature for all models
may be more physically relevant @xcite , it is a computationaly expensive iterative process , and we believe that it would not dramatically change our conclusions .
we decided to model 5 independent observables of hv tau c , which we consider as representative of the entire dataset available to us : its broadband sed ( table [ tab : photom ] ) , the f814w , @xmath7 and @xmath14 ( lgs ) scattered light images , and our iram 1.3 mm visibilities as a function of projected baselines .
our choice of wavelengths for the scattered light images to fit for is a trade - off between considering as wide a wavelength range as possible while securing data that has high enough spatial resolution and signal - to - noise per pixel .
the latter criterion led us to discard our new @xmath17 image in this modeling . in any case , a model that reproduces the f814w , @xmath7 and @xmath14 images reasonably well is likely to also reproduce scattered light images at other wavelengths throughout the entire 0.85@xmath1 m range .
mcfost synthetic scattered light images were computed as monochromatic images at the effective central wavelength of each filter and with a 2 total field - of - view .
the thermal emission from both the disk and the envelope is neglected in computing these images , as most of the emission arises from the inner most regions , which is virtually unresolved as seen from the disk outer edge , where scattering towards the observer occurs .
as discussed below , we only aim at reproducing the morphology and we have checked that this morphology is virtually unchanged if we take into account the disk emission , which otherwise induces a heavy computational cost . in the millimeter regime
, we computed 1.3 mm thermal emission maps with a very fine spatial sampling to avoid aliasing in the fourier transform ( 005/pixel ) .
while we are interested in finding the best possible model to account for the observations of hv tau c , we also aim at determining the range of acceptable models around it . for this purpose , we adopt a bayesian inference method ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , in which each model is assigned a probability that the data are drawn from the model parameters . in cases where the prior has a uniform probability distribution , as is the case here either on a linear , cosinus - like or logarithmic scale ,
this probability is @xmath87 , where @xmath71 is the reduced chi square .
the normalization constant , @xmath88 , is chosen so that the sum of the probabilities over all models in the grid is unity .
once this is done for all models in our grid , we can derive the probability distribution for a given parameter by marginalizing the 8-dimensional probability hypercube against the other 7 dimensions .
the computation of @xmath89 has 12 degrees of freedom . to take into account the noise intrinsic to monte carlo simulations , we ran 10 independent realizations of one of the best - fitting model and quadratically added the resulting standard deviation to the observational uncertainties . for the number of packets we used , this monte carlo noise is about 10% in the optical / near - infrared and ( sub)millimeter regimes but reaches about 40% in the mid - infrared where photons have the hardest time escaping their deeply embedded emission region .
even though it is likely that this overestimates the uncertainties for non - edge - on models ( for which photons escape more easily to the observer ) , it is of little importance since these models have very poor @xmath71 values to start with , given that they do not produce the characteristic double - hump sed . in computing @xmath90 , we chose to fit for @xmath91 instead of @xmath92 , as this better handles the case of datasets that are dominated by calibration uncertainties ( i.e. , all measurements beyond 100@xmath1 m ) , which are multiplicative rather than additive in nature . for scattered light images ,
we perform a pixel - by - pixel computation , though with a few adjustments .
first of all , we resampled the @xmath7 band image to a pixel scale of 0027 for a higher signal - to - noise without compromising the spatial resolution .
all synthetic images were then convolved with the same psfs as used for the deconvolution of hv tau ab ( and a tinytim - generated f814w psf ) .
we then aligned the model and empirical images using cross - correlation in the fourier domain to determine the offsets with which the model images are best aligned with the observed ones .
we further normalized all images to a peak value of unity since our goal is to reproduce the morphology of the images , the flux being fitted for in the sed . finally , to avoid including many pixels where both the data and model are indistinguishable from zero and that would artificially improve our @xmath71 , we selected to use square binary masks encompassing the area where there is significant flux from hv tau c. these masks are 2 , 16 and 14 on a side at f814w , @xmath7 and @xmath14 , respectively , resulting in 2017 , 3713 and 2201 degrees of freedom .
tests conducted without these masks showed that this does not affect the ranking of the models from best to worse while ensuring that the best @xmath71 are not artificially low , which would result in too relaxed constraints .
similar to the sed @xmath71 computation , we estimated monte carlo noise maps by computing 10 realization of the best - fitting model at all three wavelengths .
for the number of packets we used here , relative uncertainties range from less than 2% at the image peak to 812% at the disk s outer edge . at 1.3 mm ,
the available data are a series of correlated fluxes as a function of projected baselines .
to reduce the number of visibilities to compute for each model and to improve their signal - to - noise ratio , we first flagged out all correlated fluxes whose uncertainty was larger than 0.1jy , i.e. , twive the total flux from the source itself .
we then averaged the correlated fluxes in the ( u , v ) plane using a 50 m bin size .
this yields 60 independent measurements , or 52 degrees of freedom , at 1.3 mm .
the model images were first rotated so that the disk midplane lies at a position angle of 1083 , and convolved with a 03 gaussian `` seeing '' .
after padding the images with zeros to obtain the adequate total field of view , we computed their fast fourier transform to obtain a set of synthetic correlated fluxes .
as for the treatment of the scattered light described above , here we do not wish to fit for the total flux in the 1.3 mm map but for its morphology .
therefore , we convert the correlated fluxes into visibilities using the observed 49.5mjy flux for the input data and the measured total flux in each map for the models .
the 1.3 mm synthetic maps are essentially noiseless and we neglect the monte carlo noise in this regime . the @xmath71 values for the models that best fit each observation of hv tau c are listed in table [ tab : chi2 ] .
the combination of observational and numerical uncertainties used in computing the goodness - of - fit of a given model is unlikely to follow exactly a normal distribution .
caution should therefore be used when interpreting individual @xmath71 values as they could be systematically biased .
nonetheless , the ranking of the models is likely correct and the model with the absolute lowest @xmath71 value should be close to what would have been the best possible fit within the explored parameter space in the absence of non - gaussian noise .
first of all , we note that the best reduced @xmath71 for the sed , @xmath7- and @xmath14-band images fit are in the 3.74.9 range . considering the intrinsic asymmetry of the disk and the coarse sampling of the parameter space , we consider that these observations are well reproduced .
while the intrinsic structure of protoplanetary disks is likely to be much more complex than the power law used here , the simplified parametrization adopted here appears reasonable and can therefore provide valuable insight about the disk properties .
the fact that the best @xmath93 is only 11.5 , i.e. , much worse than the other two scattered light images , results in part from its higher signal - to - noise ratio which amplifies any departure from the models . in addition , the relative undersampling of the wfpc2 image results in a poorer ability to register images , thereby globally increasing all values of @xmath93 .
lastly , we note that the range of @xmath94 in our entire grid is very narrow , about a factor of 2 from best to worst , implying that this observation of hv tau c is only mildly constraining for our model .
this was to be expected considering that the disk is only marginally resolved in our pdbi data .
figure [ fig : fit_sed ] compares the observed sed of hv tau c to several `` best '' models .
the best overall model has a foreground extinction of @xmath95mag , marginally lower than the extinction we estimated for hv tau ab .
both the model that best fits the f814w image and the best overall model with @xmath68@xmath1 m ( i.e. , interstellar - like dust ) fail badly at reproducing the photometric data , with @xmath89 of 139.9 and 20.7 , respectively .
although these models produce a millimeter flux that is a factor of at least a few lower than the observed flux , it is interesting to note that they yield a shallow millimeter spectral index as a result of their high optical depth .
indeed , all models that adequately reproduce at least one of the hv tau c observations have @xmath96 in the disk midplane .
this is a clear reminder that millimeter spectral indices can be strongly affected by optical depth effects in the case of edge - on disks .
another shortcoming of an interstellar - like dust model , as discussed by @xcite , is that it yields a near- to mid - infrared slope that is far too steep compared to observations if they are required to produce the right flux ratio between the two humps of the sed .
this is a result of the vanishingly small mid - infrared albedo of any dust model that does not include grains of several microns in size .
therefore , the sed of hv tau c points to the presence of at least intermediate - size grains in the disk .
let us now consider the model scattered light images .
figure [ fig : fit_imgs ] compares the observed images to several key models .
in addition to the factors listed above , the poorer fit to the f814w image likely arises from our simplistic treatment of the spherical halo , which is too bright at large distances above the disk midplane in the models .
the models that best reproduce individual images each provide a good fit to the main features of the hv tau c disk , such as the distance and flux ratio between the two nebulae .
the model that offers the best trade - off between the three scattered light images nicely reproduces the wavelength dependency of @xmath34 but has roughly achromatic flux ratios between the nebulae .
this is likely a consequence of the fact that this model , which has @xmath68 cm , has an almost wavelength - independent scattering phase function , which is required to reproduce the intensity profiles shown in figure [ fig : profiles ] .
interestingly , while the best model for each individual image corresponds to @xmath97au , the best scattered light and best overall model both have an outer radius of 50au .
again , the disk asymmetry prevents us from unambiguously estimating the disk outer radius based on the images only .
finally , we note that the model that best fits the sed is a very poor fit to the scattered light images ( @xmath71 ranging from 34.3 to 68.1 ) .
the inclination of that model is 763 , giving the observer an almost grazing angle line - of - sight , resulting in a counternebula that is almost undetectable and an almost point - like appearance at @xmath14 which does not match the observed image at all .
although the 1.3 mm emission map is approximately reproduced by virtually all models in our grid , it is noteworthy that the best - fitting models have a face - on inclination .
this stems from the fact that the disk , if viewed edge - on , is unresolved along its minor axis given the beam of our pdbi observations .
however , our data show a slight drop of the correlated fluxes at the longest baselines along that position angle .
therefore , our modeling prefers a lower inclination in order to explain this feature . since we know that the hv tau c disk is indeed edge - on
, this may indicate that the atmospheric phase noise increases with baseline in a way that differs from our gaussian `` seeing '' approximation .
alternatively , it could be that the intrinsic thermal emission from the spherical envelope is stronger than in our model , which only contains small dust grains that are inefficient emitters at this regime . while this could in principle
induce a bias in our modeling , we consider this effect as negligible considering that this dataset is the least important in constraining the disk model .
figure [ fig : bayes_all ] shows the inferred probability distributions for each of our 8 free parameters , after marginalization against all 7 other parameters .
fitting for each scattered light image separately yields probability distributions that are similar to each other and we only show the probability distribution for fitting all three images at once , for clarity purposes .
the only differences between fits to individual images , associated with inclination and maximum grain size , are discussed below .
the results from the fit to the f814w image tends to drive the results when grouping all three scattered light images because the other images are comparatively easier to reproduce , as shown by the lower best @xmath98 and @xmath99 compared to the best @xmath93 .
we remind the reader that the derived probabilities that a given parameter takes a certain value are only valid within the framework of our modeling and are based on our approximate treatment of monte carlo noise in the simulations .
nonetheless , they represent a more reliable metric in our analysis than @xmath71 values for individual models and they therefore better highlight how each observation informs our model of the disk .
we note that the best individual model ( shown as filled diamonds in figure[fig : bayes_all ] ) has parameter values that correspond in most cases to the most likely value as determined from the bayesian method , or to a value whose global probability is at least 20% .
this demonstrates that noise in both data and model is well - behaved , providing support to the results of the bayesian analysis .
the scattered light images place some constraints on @xmath54 , @xmath60 , @xmath77 and , to a smaller extent , @xmath72 .
since we have used flux - normalized images , @xmath63 remains unconstrained .
similarly , @xmath76 , @xmath75 and @xmath73 are not well constrained as a result of trade - off between these parameters and @xmath72 .
a scale height of 45au is preferred when fitting only the f814w but it is not conclusive and the other images have much flatter probability distributions for @xmath73 .
our modeling of the scattered light images finds @xmath100au as the most likely value , although an outer radius of 75au can not be formally excluded considering the disk asymmetry and total extent . the probability distribution for @xmath72 peaks at the lowest end of the range sampled here , and @xmath101% .
it is worth noting that the probability distributions based on fitting each image individually are almost flat , but combining all three in a single fit yields a significant constraint as it helps solving for some of the ambiguities between parameters .
similarly , we find that @xmath102% when fitting all three images at once , but there is an underlying trend as a function of wavelength .
indeed , @xmath103 drops from 45% ( most likely ) at 0.8@xmath1 m , to 20% at 1.6@xmath1 m and to 7% ( least likely ) at 3.8@xmath1 m , respectively .
this gradual and smooth evolution with wavelength is probably indicative of a real physical phenomenon .
finally , the combined fit to the scattered light images suggests an inclination of 827@xmath104 .
however , there is a marginal ( 2@xmath30 ) trend as a function of wavelength : the most likely inclinations are 808@xmath105 , 846@xmath10625 and 859@xmath107 for the f814w , @xmath7 and @xmath14 images , respectively .
this trend stems from the inability of our model to produce a wavelength - dependent flux ratio between the nebulae without generating chromatic variations of the intensity profile along the major axis .
this may indicate that our dust model does not possess the adequate chromatic behavior .
considering now the sed , our modeling mostly provides constraints on the same parameters as the scattered light images .
the sed fits favor the presence of large grains and a high total dust mass , with @xmath108% and @xmath109% .
both conditions are necessary to produce sufficient fluxes at the long wavelength end of the sed .
while interstellar - like grains can produce the observed spectral index , all models with such a dust population fall short in the 13 mm range by at least an order of magnitude . while sed fitting is generally not sensitive to the outer radius of a disk
, here we find that a small outer radius is strongly preferred : @xmath110 .
this is because the relative height of the two humps in the sed and the depth of the trough that separates them is directly influenced by the disk geometry , via the total optical depths along our line of sight to the star on one hand and from the star to the upper layers of the outer disk on the other hand . finally , the modeling of the sed yields a best fit inclination of 750@xmath111 .
the fact that the sed provides a weaker constraint on the disk inclination than the scattered light images is typical of all disk analyses .
nonetheless , the particular viewing angle of edge - on disk allows for a reasonably narrow range of inclination to be defined , and it is reassuring to note that the difference between the inclinations derived from the scattered light images and the sed is at the insignificant 1.6@xmath30 level .
the fit to the 1.3 mm yields much weaker constraints on the model parameters : all associated probability distributions are consistent with being flat except for @xmath54 , @xmath76 and , to a smaller extent , @xmath77 . as discussed above , this is a result of the fact that the disk is only marginally resolved in our pdbi data .
to best reproduce the observed visibilities , the model favors a face - on geometry , a large outer radius and a flat surface density that places a lot of mass , hence of millimeter emission , at large radii .
as far as inclination is concerned , the inferred most likely value is 0@xmath112 , illustrating the weak preference for a face - on geometry .
the constraints on both @xmath54 and @xmath76 are much stronger with a ratio of the most- to least - likely value of at least 4:1 .
the strength of our modeling approach is to be able to fold all independent datasets in a single coherent fit , which both yields much sharper constraints and a view of the intrinsic contradictions between observations .
for some parameters , such as @xmath75 , @xmath73 and @xmath63 , the combined fit provides better constraints than any individual observation as a consequence of the fact that each observation is associated to a different set of ambiguities .
we thus find that the disk is significantly flared ( @xmath113 ) , that its scale height at 50au is 5au at most , and that @xmath114 , which implies @xmath115 .
assuming that the accretion luminosity is negligible , this is consistent with a 1myr 0.71@xmath3 central star @xcite , sightly larger than the mass estimated for hv tau a @xcite .
although uncertainties are large , this is in rough agreement with the dynamical mass that we have estimated from the rotating co disk . in the case of @xmath54 and @xmath76 ,
the constraints set by the millimeter mapping are superseded by those from the scattered light images and the sed because the former is comparatively too easy to reproduce in our grid . while a @xmath116 surface density profile is preferred in our model , all other values tested in our grid have non - negligible probabilities .
similarly , our overall fit favors @xmath100au only by a factor of 2 over the larger radius . excluding the 1.3 mm map from our analysis increases the preference for the smaller outer radius to a factor of 4:1 , however .
future higher resolution millimeter observations are needed to better constrain both of these parameters . also , the final probability distribution for the disk inclination is largely driven by the scattered light images .
our global fit suggests an inclination of 821@xmath117 .
the most interesting results of our analysis concern @xmath60 and @xmath72 , for which scattered light images and sed lead to contradictory predictions ( see figure [ fig : bayes_all ] .
the latter favors large grains and a high dust mass whereas the former suggests small grains and a low dust mass .
our global fit , which aims at finding the best possible trade - off between all constraints points to a very high dust mass ( @xmath118 ) and an intermediate maximum grain size ( @xmath119@xmath1 m ) .
since both of these values are essentially rejected ( @xmath120% ) by at least one of the observations , this `` best '' model should not be considered as a good fit .
rather , this apparent contradiction between the various observations , which probably mirrors the trend in preferred @xmath60 as a function of wavelength for the scattered light images , indicates that our model is too simplistic in at least some aspects .
possible explanations are explored in section [ sec : discus ] .
there are three published models of the hv tau c disk . in their discovery study , @xcite derived a disk inclination of @xmath121 based on a single scattering model .
that inclination has later been confirmed by subsequent models of the disk , including the present work .
the second , most sophisticated to date , model of the disk was put forth by @xcite who analyzed scattered light images at 0.8 and 2.2@xmath1 m .
more recently , @xcite derived a total disk mass for hv tau c based on its sub - millimeter flux although they did not conduct a complete sed fit for this source .
here we compare our model results to these previous efforts .
the model constructed by @xcite was based on the same f814w image we have used here , and a lower quality 2.2@xmath1 m adaptive optics image that was characterized by a poorer resolution ( 013 compared to 008 for our vlt / naos image ) and a smaller achieved strehl ratio .
their conclusions are similar to ours in several ways .
first of all , they found that it is easier to fit for the @xmath44 band image , as demonstrated by the lower achieved @xmath71 value .
second , fitting for the envelope - free @xmath44 band image alone points towards an inclination that is closer to edge - on than if the f814w image is included in the fit .
conversely , when fitting both images simultaneously , they also note that they do not succeed in reproducing the wavelength dependence of the flux ratio between the nebulae . if we temporarily focus on the f814w image alone , which is the most constraining dataset in their fit , our bayesian analysis yields @xmath68@xmath1 m , @xmath122 and @xmath123au and @xmath1248 .
these values are consistent with those of @xcite within our uncertainties .
the derived inclination , somewhat further away from edge - on , may be explained by our different treatment of the spherical halo , which plays a non - negligible role in shaping up the morphology of hv tau c at 0.8@xmath1 m .
finally , we note that our simultaneous fit to the f814w , @xmath7 and @xmath14 images favors @xmath68@xmath1 m .
this model has a ratio of opacity of @xmath125 and scattering asymmetry parameters of @xmath126 and @xmath127 .
these properties are very similar to those derived by @xcite , namely @xmath128 and @xmath129 .
overall , we find that our fit to scattered light images is in good agreement with the previous modeling effort for this source , despite differences in strategy and parameter space covered .
however , the models that best reproduce the f814w image produce millimeter fluxes that are one to two orders of magnitude too low in the millimeter regime , in apparent contradiction with the conclusion of @xcite , who found agreement within a factor 2 .
the explanation for this discrepancy is that these authors assumed a ratio of dust opacity of @xmath130 .
however , with our assumed dust composition , this ratio is about 7.10@xmath131 for @xmath68@xmath1 m , accounting for most of the difference in predicted millimeter flux between the two models .
@xcite presented submillimeter observations of hv tau c , which we have included in our fit . because the source showed an unusual spectral index , they did not attempt to fit a full - fledged model but rather used an empirical recipe to convert the measured 850@xmath1 m flux into a disk mass .
they derived @xmath132 , almost two orders of magnitude lower than the value we have derived here based on either the sed or global fit .
the overall shape of the sed for hv tau c suggests that the 850@xmath1 m flux is underestimated , possibly a factor of 3 or so , accounting for part of this discrepancy .
in addition , the empirical law that @xcite used was mostly derived from sources which are not edge - on and for which optical depth effects are negligible .
however , the model defined by the combination of each of the most likely parameter value has @xmath133 in the disk midplane .
we believe that this factor is the most important in explaining the difference in the disk masses inferred here and by @xcite .
while our model has been successful at reproducing the sed of hv tau c , as well the morphology and several key chromatic behavior of its scattered light images , it falls short in several aspects which are worth exploring .
as expected , a significant shortcoming of our model is its built - in axisymmetric assumption , which prevents us from finding perfect matches to any of the scattered light images . since the nature of the asymmetry
can only be speculated upon , no clear path to resolution can be provided until further monitoring clarifies its origin .
this peculiarity is however not sufficient to account for the poor @xmath134 of our overall best fitting model .
the most glaring limitation of our model is the fact that it can not account simultaneously for the sed on one hand and the scattered light images on the other .
this fact , which most likely is independent of the axisymmetry of the model , implies that more complexity has to be included in the model .
the physical properties of dust grains are subject to improvement . based on our modeling experience , the necessary requirements for an improved dust model in hv tau c would be an interstellar - like opacity law and achromatic scattering phase function from 0.8 up 5@xmath1 m and an optical - to - millimeter opacity ratio @xmath135 . for one ,
one could devise a dust model that contains a fraction of c - rich grains to better conform to interstellar abundances .
departures from a simple power law size distribution are also likely .
such a phenomenon has been observed in the interstellar medium @xcite and multi - modal distributions are predicted by dynamical models of grain growth , migration and fragmentation in protoplanetary disks ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
another direction to explore is the possibility of non - spherical grains .
recently , @xcite successfully reproduced the wavelength - independent anisotropic scattering observed in cometary dust by considering aggregates of small particles .
since the dust in hv tau c also has achromatic scattering properties , aggregate grains are a candidate that deserves consideration in the future . in this context , it is interesting to note that fractal grains are expected to have a much flatter optical - to - millimeter opacity law @xcite , akin to our models with @xmath136@xmath1 m , which are preferred in our analysis .
since it appears difficult to simultaneously match all of these constraints with a single dust model , it will likely be necessary to drop the assumption that the dust properties are uniform throughout the disk .
since each type of observation probes a different region of the disk , _ decoupling the dust properties in various regions of the disk would likely provide sufficient leeway .
_ in this picture , the disk would be quite massive and would contain large grains , up to millimeter - sized particles , in the midplane in order to explain the observed sed .
however , grains in the surface layers would not exceed a few microns in size to explain the scattered light images . to limit the number of free parameters , we have only considered fully mixed models here but we suggest that this hypothesis be further studied in future modeling of hv tau c. future modeling efforts on hv tau c should consider these various possibilities in improving on the model presented here .
a robust strategy to move forward could consist in conducting detailed modeling of a particular type of observations , such as that conducted by @xcite on the scattered light images of hh30 for instance .
these specific analyses would allow one to extract the maximum amount of information from each observation and to feed it into a refined global model .
our objective in this work was to conduct the first `` global '' fit to the structure and dust content of the hv tau c disk , one of the few objects in which such an effort is conducted in a self - coherent way .
although some uncertainty remains , our analysis has led to some robust conclusions about the hv tau c disk .
for instance , the total mass of the disk must be at the high end of the range probed here . even in the presence of large grains , which are efficient emitters in the millimeter regime ,
the total dust mass has to be on the order of @xmath137 .
indeed , such a high dust mass is necessary to account for the observed fluxes and relatively high optical depths up to 3.3 mm . assuming a canonical 100:1 gas - to - dust mass ratio , this implies that the disk is quite massive compared to the central star ( @xmath138 )
this is at the high end of the distribution observed for disks surrounding ttauri stars @xcite .
it is plausible that the disk is only marginally stable , in which case the presence of spiral density waves could be responsible for the observed asymmetry and/or variability of the system .
one respect in which the hv tau c disk is special is its compact radius , a probable consequence of tidal forces induced by hv tau ab . with @xmath100au and @xmath139
, we derive a total surface density of 2800 g @xmath140 1au away from the star , comparable to the surface density inferred for the early solar nebula as well as for extra - solar planetary systems , albeit with a possibly flatter surface density profile @xcite .
indeed , the surface density at large radii in the disk , almost 60g.@xmath140 at 50au , appears substantially higher than those derived for other protoplanetary disks @xcite . _
the disk around hv tau c is a clear example of the fact that binary systems can host circumstellar disks massive enough to form planets for timescales of several million years_. indeed , hv tau may well be a prototype for the earliest evolutionary stages of field stars that are found to host both an extrasolar planet and at least one stellar companion ( e.g. , * ? ? ?
* ; * ? ? ?
* ) . as a test of the self - consistency of our model , it is interesting to compare the disk scale height we have derived , @xmath141au , to the simple assumption of vertical hydrostatic equilibrium that is embedded in our parametric disk structure . assuming molecular hydrogen , the scale height can be written as @xmath142au .
our best global model has a midplane temperature of 19k at 50au , so that our upper limit on @xmath73 implies @xmath143 .
this is not a stringent constraint on the mass of the central star , but we note that this lower limit does not violate the mass inferred from the spectral type and luminosity of hv tau c nor the rough kinematic estimate we derived from the rotating co disk .
both estimates are in the 0.71@xmath3 range , for which the derived scale height would be 3.34au at 50au , which is consistent with our modeling .
we therefore conclude that the outer disk can reasonably be described by a disk in hydrostatic equilibrium .
in addition , _ we suggest that hv tau c may be the most massive component in the triple system . _ finally , while large uncertainties remain regarding the dust properties , _ it is clear that dust grains much larger than 1@xmath1 m are found in at least some parts of the disk_. this could be evidence for grain growth in the disk , although it is plausible that growth to a few microns may have predated the formation of the disk itself @xcite .
indeed , analyses of extinction laws have already pointed to the presence of small quantities of grains a few microns in size in the interstellar medium and even more so in molecular clouds @xcite . on the other hand , growth to mm - sized grains , which seems supported by the analysis of the sed of hv tau c ,
has most likely occurred in the disk itself .
if the disk is indeed vertically stratified and no such grains are found in the disk surface , it remains unclear whether this is because of a very efficient settling process or a mere consequence of the difficulty to grow such large particles in the lower density regions of the disk .
as outlined in the introduction , evidence for grain growth and dust stratification in protoplanetary disks has been mounting over the last two decades .
it is therefore not surprising that our modeling of the hv tau c disk calls for both processes .
however , one must keep in mind that a given observation only probes a limited region of the disk .
for instance , mid - infrared emission is dominated by emission from the inner few aus at most , providing no information on dust stratification in the outer disk .
in addition , there is an intrinsic observational bias in all studies whereby only a small range of grain sizes , those with the largest effective cross - section , actually contribute to the observed fluxes .
therefore longer wavelength observations tend to systematically call for larger grain sizes , even if one limits the analysis to scattered light images ( e.g. , * ? ? ?
similarly , the mid - infrared silicate features arise from micronic and submicronic grains , providing no constraint on much larger particles . to draw a _
complete _ picture of dust properties throughout a given disk , it is therefore necessary to gather wide and homogeneous datasets , as we have done here .
as we have already mentioned , few objects have been studied in as much detail as our present analysis of hv tau c. a clear dichotomy in terms of grain size was also found in the case of iras04302 + 2247 @xcite but the interstellar - like dust grains favored by the scattered light images were located in the relatively massive envelope surrounding this embedded object and not in the disk itself .
vertical stratification was demonstrated in the massive ggtau ring @xcite , although the available evidence only applies to grain smaller than 10@xmath1 m .
the hktaub edge - on disk also appears to have a vertically stratified structure @xcite , although a self - consistent modeling of its sed , thermal emission maps and scattered light images is still required to make direct comparison with hv tau c. on the other hand , @xcite were able to reproduce a wide range of observations of iras04158 + 2805 without the need to include vertical stratification .
a similar conclusion was recently reached by @xcite for cb26 .
overall , these detailed analyses confirm the general trends outlined above that both grain growth and stratification are common , but not ubiquitous , in protoplanetary disks .
they further offer the unique opportunity to test the detailed physics of dust evolution , such as the amount of dust stratification or its radial dependency ( e.g. , * ? ? ? * ) .
an interesting head - to - head comparison can be drawn between hv tau c and imlup , for which @xcite recently conducted a similar panchromatic modeling analysis , including scattered light images , millimeter emission maps and the system s sed .
the central stars have the same spectral type , and similarly high disk masses were inferred , making these two systems very similar .
contrary to the conclusion reached here however , @xcite were able to place stringent constraints on all disk parameters and one may wonder what the reasons for this difference are .
for one , the imlup disk is not seen edge - on , which allows the authors to fix some of the model parameters ( e.g. , @xmath63 , dust composition , @xmath144 ) in an unambiguous way .
the smaller dimensionality of the parameter space to explore allowed @xcite to sample it more finely .
another key feature of the imlup disk is that it is much less compact than hv tau c and is well resolved at millimeter wavelengths , providing more stringent constraints on the model .
last but not least , a qualitative analysis of the sed of imlup readily demonstrated the need for a stratified structure in the disk , which introduces an extra degree of freedom that we could not afford in our analysis of hv tau c. nonetheless , our study has shown that _
significant constraints on the disk properties can be obtained in an edge - on geometry_. no global modeling of hh30 or hktaub , other prototypical edge - on disks , are yet available , but conducting such efforts should yield some interesting results and are needed to move forward in our understanding of protoplanetary disks .
we have obtained new 15@xmath1 m adaptive optics images of the hv tau triple system at the vlt and keck observatories , including the first 4.8@xmath1 m scattered light image of the edge - on hv tau c disk .
all of our images reveal a steady lateral asymmetry in the disk that indicates a departure from pure axisymmetry in the disk .
this could be related to the known variability of hv tau c. we extact precise relative astrometry from our new images and find that hv tau ab
c constitutes a common proper motion pair .
we also resolved the tight binary system hv tau ab and found a surprisingly slow orbital motion compared to its projected separation , probably due to orbit eccentricity and/or a large deprojected physical separation .
we have also obtained the first spatially resolved 1.3 mm continuum and @xmath2co 21 millimeter maps of the hv tau c disk with iram s pdbi interferometer .
the continuum emission is resolved along the same position angle as the scattered light images , as expected from dust thermal emission .
the co map shows evidence of keplerian rotation about an 0.51@xmath3 central star .
we also completed the sed of that component with new _
spitzer_/mips 24 , 70 and 160@xmath1 m observations as well as observations at 350@xmath1 m , 2.8 and 3.3 mm at the cso , pdbi and carma facilities .
the 13 mm spectral index of hv tau c is relatively flat ( @xmath46 ) , indicative of the presence of millimeter - sized grains in the disk and/or high optical depth even at millimeter wavelengths . to interpret these data along with previously published hst images of hv tau c ,
we have computed a grid radiative transfer models that produced synthetic disk observations .
while we can reproduce most observational properties of hv tau c , a bayesian analysis of our model grid reveals that the sed and scattered light images of hv tau c provide mutually exclusive constraints .
as found previously , the scattered light images are best fit with a relatively small total mass and interstellar - like dust properties .
fitting the sed , however , requires almost two orders of magnitude more mass as well as a grain size distribution that extends to millimeter sizes .
the mismatch between the two sets of constraints reveals an intrinsic shortcoming of our parametrized model , which assumes perfectly mixed dust .
indeed , the small maximum dust grain size inferred from fitting the scattered light images is impossible to reconcile with the long - wavelength end of the sed . in turn
, this suggests that both grain growth and vertical stratification are present in the hv tau c disk .
while these phenomena have been suggested for other protoplanetary disks in the past , hv tau c is one of a handful of objects that have been studied in sufficient details to eventually provide a complete picture of these processes .
future observations of hv tau c can help refine the model proposed here .
for instance , high - resolution ( 02 or better ) millimeter mapping of the disk would better resolve it and provide more stringent constraints .
in particular , obtaining observations at even longer wavelengths , 7 mm or 1.3 cm , would probe a regime in which the optical depth through the disk is much smaller and enable a more direct interpretation in terms of disk properties . on the longer run , mapping of the disk with alma in the submillimeter regime and in scattered light at 810@xmath1 m with the next generation of large ground - based telescopes
will help bridge the gap between the scattering and thermal emission regimes at a roughly constant spatial resolution .
in particular , very high resolution maps with alma may resolve this and other disks along the vertical axis , in order to probe their vertical stratification and , ultimately , to outline an evolutionary sequence .
we are indebted to all members of the geode group for many fruitful discussions about modeling of edge - on protoplanetary disks , and in particular to marshall perrin for developing some of the model analysis tools used in this project .
we are grateful to darren dowell for his help with the astrometric calibration of our cso data and to the carma staff for conducting the observations presented in this paper .
the work presented here has been funded in part by national science foundation science and technology center for adaptive optics , managed by the university of california at santa cruz under cooperative agreement no .
ast - 9876783 , by the programme national de physique stellaire of cnrs / insu ( france ) and by the agence nationale de la recherche through contract anr-07-blan-0221 .
c. p. acknowledges the funding from the european commission s seventh framework program as a marie curie intra - european fellow ( pief - ga-2008 - 220891 ) .
the authors wish to acknowledge the contribution from intel corporation , hewlett - packard corporation , ibm corporation , and the national science foundation grant eia-0303575 in making hardware and software available for the citris cluster which was used in producing some of the model computations presented in this paper .
support for carma construction was derived from the gordon and betty moore foundation , the kenneth t. and eileen l. norris foundation , the associates of the california institute of technology , the states of california , illinois , and maryland , and the national science foundation . ongoing carma development and operations
are supported by the national science foundation under a cooperative agreement , and by the carma partner universities .
some of the data presented herein were 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 .
the authors wish to recognize and acknowledge the very significant cultural role and reverence that the summit of mauna kea has always had within the indigenous hawaiian community .
we are most fortunate to have the opportunity to conduct observations from this mountain .
alves de oliveira , c. , & casali , m. 2008 , , 485 , 155 akeson , r. l. , ciardi , d. r. , van belle , g. t. , & creech - eakman , m. j. 2002 , , 566 , 1124 andrews , s. m. , & williams , j. p. 2005
, , 631 , 1134 andrews , s. m. , & williams , j. p. 2007
, , 659 , 705 appenzeller , i. , bertout , c. , & stahl , o. 2005 , , 434 , 1005 baraffe , i. , chabrier , g. , allard , f. , & hauschildt , p. h. 1998 , , 337 , 403 barrire - fouchet , l. , gonzalez , j .- f . , murray , j. r. , humble , r. j. , & maddison , s. t. 2005 , , 443 , 185 beckwith , s. v. w. , sargent , a. i. , chini , r. s. , & guesten , r. 1990 , , 99 , 924 beckwith , s. v. w. , & sargent , a. i. 1991 , , 381 , 250 bertout , c. , basri , g. , & bouvier , j. 1988 , , 330 , 350 bertout , c. , & genova , f. 2006 , , 460 , 499 burrows , c. j. , et al .
1996 , , 473 , 437 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , , 345 , 245 chiang , e. i. , joung , m. k. , creech - eakman , m. j. , qi , c. , kessler , j. e. , blake , g. a. , & van dishoeck , e. f. 2001 , , 547 , 1077 cotera , a. s. , et al .
2001 , , 556 , 958 dalessio , p. , calvet , n. , & hartmann , l. 2001 , , 553 , 321 dalessio , p. , calvet , n. , hartmann , l. , franco - hernndez , r. , & servn , h. 2006 , , 638 , 314 dowell , c. d. , et al .
2003 , , 4855 , 73 draine , b. t. 2003 , , 598 , 1017 duchne , g. , mnard , f. , stapelfeldt , k. , & duvert , g. 2003 , , 400 , 559 duchne , g. , mccabe , c. , ghez , a. m. , & macintosh , b. a. 2004 , , 606 , 969 duchne , g. , bontemps , s. , bouvier , j. , andr , p. , djupvik , a. a. , & ghez , a. m. 2007 , , 476 , 229 ducourant , c. , teixeira , r. , pri , j. p. , lecampion , j. f. , guibert , j. , & sartori , m. j. 2005 , , 438 , 769 dullemond , c. p. , & dominik , c. 2004 , , 421 , 1075 dullemond , c. p. , & dominik , c. 2005 , , 434 , 971 dutrey , a. , guilloteau , s. , duvert , g. , prato , l. , simon , m. , schuster , k. , & menard , f. 1996 , , 309 , 493 eggenberger , a. , udry , s. , chauvin , g. , beuzit , j .- l . ,
lagrange , a .- m . ,
sgransan , d. , & mayor , m. 2007 , , 474 , 273 eiroa , c. , et al .
2002 , , 384 , 1038 flaherty , k. m. , pipher , j. l. , megeath , s. t. , winston , e. m. , gutermuth , r. a. , muzerolle , j. , allen , l. e. , & fazio , g. g. 2007 , , 663 , 1069 furlan , e. , et al .
2006 , , 165 , 568 ghez , a. m. , weinberger , a. j. , neugebauer , g. , matthews , k. , & mccarthy , d. w. , jr .
1995 , , 110 , 753 ghez , a. m. , et al .
2008 , , 689 , 1044 glauser , a. m. , mnard , f. , pinte , c. , duchne , g. , gdel , m. , monin , j .- l . , & padgett , d. l. 2008 , , 485 , 531 guilloteau , s. , et al . 1992 , , 262 , 624 hartmann , l. , megeath , s. t. , allen , l. , luhman , k. , calvet , n. , dalessio , p. , franco - hernandez , r. , & fazio , g. 2005 , , 629 , 881 hayashi , c. 1981 , progress of theoretical physics supplement , 70 , 35 hughes , a. m. , wilner , d. j. , qi , c. , & hogerheijde , m. r. 2008 , , 678 , 1119 indebetouw , r. , et al .
2005 , , 619 , 931 isella , a. , testi , l. , natta , a. , neri , r. , wilner , d. , & qi , c. 2007 , , 469 , 213 keene , j. , & masson , c. r. 1990 , , 355 , 635 kenyon , s. j. , & hartmann , l. 1987 , , 323 , 714 kenyon , s. j. , & hartmann , l. 1995 , , 101 , 117 kessler - silacci , j. , et al .
2006 , , 639 , 275 kim , s .- h . , martin , p. g. , & hendry , p. d. 1994 , , 422 , 164 kimura , h. , kolokolova , l. , & mann , i. 2003 , , 407 , l5 kitamura , y. , momose , m. , yokogawa , s. , kawabe , r. , tamura , m. , & ida , s. 2002 , , 581 , 357 kuchner , m. j. 2004 , , 612 , 1147 laibe , g. , gonzalez , j .- f . ,
fouchet , l. , & maddison , s. t. 2008 , , 487 , 265 lay , o. p. , carlstrom , j. e. , hills , r. e. , & phillips , t. g. 1994 , , 434 , l75 lay , o. p. , carlstrom , j. e. , & hills , r. e. 1997 , , 489 , 917 lenzen , r. , et al .
2003 , , 4841 , 944 magazzu , a. , & martin , e. l. 1994 , , 287 , 571 mannings , v. , & emerson , j. p. 1994
, , 267 , 361 mccabe , c. , duchne , g. , & ghez , a. m. 2003 , , 588 , l113 mccabe , c. , ghez , a. m. , prato , l. , duchne , g. , fisher , r. s. , & telesco , c. 2006 , , 636 , 932 mizuno , a. , onishi , t. , yonekura , y. , nagahama , t. , ogawa , h. , & fukui , y. 1995 , , 445 , l161 monin , j .- l .
, & bouvier , j. 2000 , , 356 , l75 mugrauer , m. , & neuhuser , r. 2009 , , 494 , 373 mundy , l. g. , et al .
1996 , , 464 , l169 muzerolle , j. , calvet , n. , hartmann , l. , & dalessio , p. 2003
, , 597 , l149 natta , a. , grinin , v. , & mannings , v. 2000 , protostars and planets iv , 559 natta , a. , testi , l. , calvet , n. , henning , t. , waters , r. , & wilner , d. 2007 , protostars and planets v , 767 pani , o. , hogerheijde , m. r. , wilner , d. , & qi , c. 2009 , , in press ( astro - ph/0904.1127 ) pitu , v. , guilloteau , s. , & dutrey , a. 2005 , , 443 , 945 pinte , c. , mnard , f. , duchne , g. , & bastien , p. 2006
, , 459 , 797 pinte , c. , fouchet , l. , mnard , f. , gonzalez , j .- f . , & duchne , g. 2007 , , 469 , 963 pinte , c. , et al .
2008 , a&a , 489 , 633 pontoppidan , k. m. , stapelfeldt , k. r. , blake , g. a. , van dishoeck , e. f. , & dullemond , c. p. 2007
, , 658 , l111 roddier , c. , roddier , f. , northcott , m. j. , graves , j. e. , & jim , k. 1996 , , 463 , 326 rodmann , j. , henning , t. , chandler , c. j. , mundy , l. g. , & wilner , d. j. 2006 , , 446 , 211 rousset , g. , et al .
2003 , , 4839 , 140 sauter , j. , et al .
2009 , , in press simon , m. , chen , w. p. , howell , r. r. , benson , j. a. , & slowik , d. 1992 , , 384 , 212 simon , m. , holfeltz , s. t. , & taff , l. g. 1996 , , 469 , 890 stapelfeldt , k. r. , krist , j. e. , menard , f. , bouvier , j. , padgett , d. l. , & burrows , c. j. 1998 , , 502 , l65 stapelfeldt , k. , & moneti , a. 1999 , the universe as seen by iso , 427 , 521 stapelfeldt , k. r. , mnard , f. , watson , a. m. , krist , j. e. , dougados , c. , padgett , d. l. , & brandner , w. 2003 , , 589 , 410 strom , k. m. , newton , g. , strom , s. e. , seaman , r. l. , carrasco , l. , cruz - gonzalez , i. , serrano , a. , & grasdalen , g. l. 1989 , , 71 , 183 strom , k. m. , & strom , s. e. 1994 , , 424 , 237 terada , h. , tokunaga , a. t. , kobayashi , n. , takato , n. , hayano , y. , & takami , h. 2007 , , 667 , 303 van dam , m. a. , et al .
2006 , , 118 , 310 watson , a. m. , & stapelfeldt , k. r. 2004 , , 602 , 860 watson , a. m. , & stapelfeldt , k. r. 2007 , , 133 , 845 watson , a. m. , stapelfeldt , k. r. , wood , k. , & mnard , f. 2007 , protostars and planets v , 523 weidenschilling , s. j. 1997 , icarus , 127 , 290 weingartner , j. c. , & draine , b. t. 2001 , , 548 , 296 white , r. j. , & ghez , a. m. 2001 , , 556 , 265 wizinowich , p. , et al .
2000 , , 112 , 315 wizinowich , p. l. , et al . 2006 , , 118 , 297 woitas , j. , & leinert , c. 1998 , , 338 , 122 wolf , s. , padgett , d. l. , & stapelfeldt , k. r. 2003 , , 588 , 373 wood , k. , wolff , m. j. , bjorkman , j. e. , & whitney , b. 2002 , , 564 , 887 wright , e. l. 1987 , , 320 , 818 zsom , a. , & dullemond , c. p. 2008
, , 489 , 931 | we present new high spatial resolution ( @xmath01 ) 15@xmath1 m adaptive optics images , interferometric 1.3 mm continuum and @xmath2co 21 maps , and 350@xmath1 m , 2.8 and 3.3 mm fluxes measurements of the hv tau system .
our adaptive optics images unambiguously demonstrate that hv tau ab
c is a common proper motion pair .
they further reveal an unusually slow orbital motion within the tight hv tau ab pair that suggests a highly eccentric orbit and/or a large deprojected physical separation .
scattered light images of the hv tau c edge - on protoplanetary disk suggest that the anisotropy of the dust scattering phase function is almost independent of wavelength from 0.8 to 5@xmath1 m , whereas the dust opacity decreases significantly over the same range .
the images further reveal a marked lateral asymmetry in the disk that does not vary over a timescale of 2 years .
we further detect a radial velocity gradient in the disk in our @xmath2co map that lies along the same position angle as the elongation of the continuum emission , which is consistent with keplerian rotation around an 0.51@xmath3 central star , suggesting that it could be the most massive component in the triple system . to obtain a global representation of the hv tau c disk
, we search for a model that self - consistently reproduces observations of the disk from the visible regime up to millimeter wavelengths .
we use a powerful radiative transfer model to compute synthetic disk observations and use a bayesian inference method to extract constraints on the disk properties .
each individual image , as well as the spectral energy distribution , of hv tau c can be well reproduced by our models with fully mixed dust provided grain growth has already produced larger - than - interstellar dust grains . however , no single model can satisfactorily simultaneously account for all observations .
we suggest that future attempts to model this source include more complex dust properties and possibly vertical stratification .
while both grain growth and stratification have already been suggested in many disks , only a panchromatic analysis , such as presented here , can provide a complete picture of the structure of a disk , a necessary step towards quantitatively testing the predictions of numerical models of disk evolution . |
tubular neighbourhoods are used in differential and symplectic geometry to reveal the structure around submanifolds . as is well known
, a submanifold of a complex manifold will in general not admit a tubular neighbourhood that is holomorphic , and with this in mind we investigate here the structure around a complex submanifold of a manifold that is khler . to state our results ,
let @xmath0 be the normal bundle of a complex submanifold @xmath1 of a complex manifold @xmath2 and @xmath3 be the inclusion of @xmath1 as the zero section .
note that @xmath4 is a holomorphic vector bundle that admits a holomorphic @xmath5-action obtained by rotating the fibres .
a smooth tubular neighbourhood of @xmath6 is a diffeomorphism @xmath7 between a neighbourhood @xmath8 of @xmath9 and a neighbourhood @xmath10 of @xmath6 such that @xmath11 [ thm : tubularmain ] suppose that @xmath1 is a complex submanifold of a khler manifold @xmath12 .
then there exists a smooth tubular neighbourhood @xmath13 of @xmath6 with the following properties : 1 .
@xmath8 is @xmath5-invariant and the pullback @xmath14 is an @xmath5-invariant khler form on @xmath8 .
2 . for any @xmath15
the function @xmath16 given by @xmath17 extends to a holomorphic function @xmath18 from the unit disc @xmath19 such that @xmath20= u,\ ] ] i.e. the holomorphic disc @xmath21 is centered at @xmath22 and points in the normal direction determined by @xmath23 .
when @xmath1 is compact there is a canonical choice of @xmath24 .
in general , the germ of the tubular neighbourhood we construct is local and canonical , in the sense that at any point @xmath25 it depends only on the local structure of @xmath26 around @xmath27 .
of course the map @xmath24 need not be holomorphic , but the two properties above describe ways in which it interacts with the holomorphic structure ( for example the statement that @xmath14 is khler is not immediate given that @xmath24 is not holomorphic ) . even in cases
where holomorphic tubular neighbourhoods do exist they will not in general have this property so what is produced above will be different .
in fact the existence of such a tubular neighbourhood is highly non - trivial even when the submanifold is a single point in @xmath28 . in the very special case
that there exists a holomorphic @xmath5-action on @xmath2 that fixes @xmath1 pointwise inducing the usual action on @xmath4 obtained by rotating the fibers there is a natural holomorphic tubular neighbourhood of @xmath1 which encodes this symmetry .
if in addition this action preserves @xmath29 then the germ of our @xmath24 will agree with this holomorphic tubular neighbourhood .
since our construction is local this is also true locally around any point @xmath27 ( see proposition [ prop : hologerm ] ) .
we point out that there is a standard way to produce tubular neighbourhoods that only satisfy the first property in the above theorem .
in fact , for any choice of symplectic form @xmath30 on a neighbourhood @xmath8 of @xmath31 such that @xmath32 there exists , after shrinking @xmath8 if necessary , a tubular neighbourhood @xmath33 such that @xmath34 ( this follows by using an arbitrary tubular neighbourhood to pull back @xmath29 and then applying the relative version of moser s theorem ( * ? ? ? * theorem 7.4 ) ) .
so by choosing such an @xmath30 that is khler and @xmath5-invariant we get a tubular neighbourhood with property ( 1 ) .
this will of course depend on several choices , and from the point of view of khler geometry it is not clear why any particular choice is more natural than any other .
we are not aware of any standard construction that yields tubular neighbourhoods with property ( 2 ) , but we suspect that they are relatively easy to produce .
thus the significance of the above theorem is that there exist tubular neighbourhoods with both properties simultaneously , which as we shall see can be made canonical
. really the main result of this paper is a proof of existence of regular solutions to a dirichlet problem for a certain homogeneous monge - ampre equation ( hmae ) , and the tubular neighbourhood of theorem [ thm : tubularmain ] will then be constructed using the associated foliation . recall the _ deformation to the normal cone _
@xmath35 of @xmath1 in @xmath2 is the blowup @xmath36 of @xmath37 along @xmath38 .
the fiber of @xmath35 over @xmath39 has two components , one which is isomorphic to the blowup @xmath40 of @xmath2 along @xmath1 and the other being the exceptional divisor @xmath41 which isomorphic to the projective completion @xmath42 of the normal bundle @xmath4 . letting @xmath43 denote the proper transform of @xmath44 we have that @xmath45 . similarly if @xmath27 we let @xmath46 denote the proper transform of @xmath47 , so @xmath48 .
the holomorphic @xmath5-action on @xmath37 given by @xmath49 lifts to @xmath35 ; it preserves @xmath4 and induces the action on @xmath4 obtained by rotating the fibers .
[ defconethm ] there exists an @xmath5-invariant neighbourhood @xmath10 of @xmath50 in @xmath35 and a smooth closed @xmath5-invariant real @xmath51-form @xmath52 on @xmath10 that gives a regular solution to the homogeneous monge - ampre equation with boundary data induced by @xmath29 .
that is , 1 .
@xmath53 is khler for all @xmath54 where @xmath55 , 2 .
@xmath56 for all @xmath57 , and 3 .
@xmath58 on @xmath10 .
the solution @xmath52 is cohomologous to @xmath59 i.e. there exists a ( unique ) smooth real valued @xmath5-invariant function @xmath60 on @xmath10 which is zero on @xmath61 for @xmath57 and such that @xmath62 when @xmath1 is compact there is a canonical choice of @xmath63 . in general
the germ around @xmath50 of any such @xmath5-invariant regular solution is unique .
we observe that since the central fibre of @xmath35 is of a different topological type to the general fibre , there can never be a regular solution defined on all of @xmath35 . in this sense
a local result such as theorem [ defconethm ] is the best one could hope for .
the uniqueness part of theorem [ defconethm ] is also novel since it does not involve any hypothesis on the behaviour of the solutions near the boundary of @xmath10 .
we explain briefly how such an @xmath52 gives rise to a tubular neighbourhood . since the work of bedford - kalka @xcite it has been known that regular solutions to the hmae generate associated _ monge - ampre foliations _ by holomorphic curves as follows : the kernel of the form @xmath52 defines an integrable distribution of complex lines in the tangent bundle of @xmath10 . by frobenius theorem
it induces a foliation by holomorphic curves , which because of property ( 1 ) is transverse to the fibers @xmath64 . in the case under consideration , @xmath52 is @xmath5-invariant and hence so is the foliation ( see figure 1 ) .
we shall say that a leaf of the foliation is _ complete _ if it covers the base @xmath65 ( i.e. the restriction to this leaf of the projection to @xmath65 is a surjection onto @xmath65 ) . from the @xmath5-invariance it follows that @xmath66 for all @xmath27 ,
thus each @xmath46 is a complete leaf of the foliation . by continuity
we can then find an @xmath5-invariant neighbourhood of @xmath50 consisting solely of complete leaves that pass through @xmath67 , so after shrinking @xmath10 we can assume that @xmath10 is foliated by such leaves .
note that @xmath68 is an @xmath5-invariant neighbourhood of @xmath31 in @xmath4 . by flowing along the leaves of this foliation we get for each @xmath54 a diffeomorphism @xmath69 which remarkably is a symplectomorphism ,
i.e. @xmath70 the map @xmath71 is the desired tubular neighbourhood , and thanks to ( [ eq : pullbacknew ] ) it has property ( 1 ) of theorem [ thm : tubularmain ] , while the existence of the holomorphic leaves assures that it has property ( 2 ) ( the details are provided in section [ sec : tubular ] ) .
we say that a solution @xmath63 to the hmae is _ complete _ if @xmath10 is foliated by complete leaves of the monge - ampre foliation , @xmath72 and whenever @xmath73 then @xmath74 for all @xmath75 . given a regular solution @xmath63 as in theorem [ defconethm ] we get a complete regular solution by shrinking @xmath10 if necessary .
let @xmath76 denote the vector field which generates the @xmath5-action on @xmath10 .
from the fact that @xmath52 is cohomologous to @xmath77 it follows that the @xmath5-action is hamiltonian , in the sense described in the next theorem .
[ thm : ham ] let @xmath63 be a regular solution to the hmae as in theorem [ defconethm ] .
then the function @xmath78 is a hamiltonian for the @xmath5-action , in that it satisfies @xmath79 moreover @xmath80 is constant along the leaves of the monge - ampre foliation associated to @xmath52 . if @xmath63 is complete then @xmath81 with equality precisely on @xmath50 .
at least when @xmath1 is compact we can use this hamiltonian to select a canonical regular solution @xmath82 to the hmae . given a complete solution @xmath63 with hamiltonian @xmath80 we define the radius @xmath83 of @xmath63 to be the supremum of all @xmath84 such that @xmath85 .
we then define the _
canonical radius _
@xmath86 to be the supremum of @xmath83 over all complete solutions @xmath63 .
note that the radius of a complete solution could be zero , and hence the canonical radius @xmath86 could also be zero .
but at least when @xmath1 is compact we clearly have that @xmath87 and thus @xmath88 .
[ thm : canonicalhmae ] assume that @xmath1 is compact ( or more generally @xmath88 )
. then there exists a unique complete solution @xmath82 to the hmae as in theorem [ defconethm ] with @xmath89 and @xmath90 this canonical solution is maximal in the following sense .
if @xmath63 is any other complete solution to the hmae as in theorem [ defconethm ] then for any @xmath91 we have that @xmath92 and there @xmath93 .
thus when @xmath88 the function @xmath94 gives a canonical smooth function defined on a neighbourhood of @xmath1 in @xmath2 ( that depends on the khler structure of @xmath2 near @xmath1 ) , whose level sets give a canonical `` flow '' away from @xmath1 , which we will revisit below . for topological reasons one can not find global regular solutions of the hmae on @xmath95 but instead of looking for local regular solutions one can consider global weak solutions . using a variant of the perron envelope , for any fixed number @xmath96 one can construct a closed positive @xmath51-current @xmath97 on @xmath95 cohomologous to @xmath98 $ ] which restricts to @xmath29 on @xmath99 for @xmath100 and which is maximal in the sense of pluripotential theory .
moreover when @xmath2 is compact and @xmath96 is small @xmath101 in the sense of bedford - taylor , thus @xmath97 is a globally defined weak solution to the hmae .
[ thm : mainhmae2 ] suppose @xmath1 is compact ( or more generally @xmath88 )
. then @xmath97 is equal to @xmath102 on @xmath103 . in particular
@xmath97 is smooth and regular on a neighbourhood of @xmath104 .
in particular this gives a local regularity result for the weak geodesic rays that are naturally associated to the deformation to the normal cone ( see section [ sec : previous ] ) .
as an application of the above ideas we shall give a new regularity result for some naturally defined envelopes that occur in pluripotential theory .
let @xmath1 be a compact complex submanifold of a ( not necessarily compact ) khler manifold @xmath12 and let @xmath105 be a parameter .
we consider the envelope @xmath106 where @xmath107 denotes the set of @xmath29-plurisubharmonic functions ( i.e. upper semicontinuous @xmath108 functions @xmath109 such that @xmath110 is a positive current ) and @xmath111 denotes the lelong - number of @xmath109 along @xmath1 . given this data , the _
equilibrium set _ is defined to be @xmath112 whose complement @xmath113 is a neighbourhood of @xmath1 .
we introduce the following definition of regularity that captures what we shall prove : we say that @xmath114 has
_ optimal regularity _ if 1 .
@xmath115 is smoothly bounded .
2 . on @xmath116
the envelope @xmath114 is smooth and also @xmath117 3 .
there is a family @xmath118 of holomorphic discs in @xmath119 parametrized by points @xmath120 in the exceptional divisor in @xmath121 such that the restriction of @xmath122 to each @xmath123 is zero .
moreover each @xmath123 passes through @xmath1 , the boundary of @xmath123 lies in @xmath124 and the family @xmath125 foliates the blowup of @xmath113 along @xmath1 ( see figure 3 ) .
recall that @xmath86 denotes the canonical radius .
[ optimalregthm ] assume that @xmath1 is compact ( or more generally @xmath88 ) . then for @xmath126 small enough ( i.e. @xmath127 ) the envelope @xmath114 has optimal regularity .
moreover the corresponding holomorphic discs @xmath128 all have area @xmath126 , and the boundaries @xmath124 vary smoothly with @xmath126 .
in fact @xmath129 where @xmath130 is the hamiltonian function of the canonical solution @xmath82 to the hmae associated to the data @xmath131 .
the above result is interesting even in the simplest case that @xmath1 is a point in an open set in @xmath132 .
this is then a purely local statement and we shall use the optimal regularity to prove the existence of a flow in @xmath132 by sets with a certain reproducing property : [ thm : flowpoint ] let @xmath133 be a smooth strictly plurisubharmonic function on the unit ball @xmath134 in @xmath135 . define @xmath136 and @xmath137 then for small @xmath126 ( i.e. @xmath138 ) the envelope @xmath114 has optimal regularity .
also for any bounded holomorphic function @xmath139 on @xmath140 we have @xmath141 here @xmath86 is the canonical radius of the hmae associated to the data @xmath142 the reproducing property in arises in the theory of complex moments and the planar hele - shaw flow in fluid mechanics ( see section [ sec : previous ] ) and gives a generalization of this flow to all complex dimensions . the key tool that we will use to prove the existence of local regular solutions to the hmae is an openness theorem of donaldson @xcite .
this states that on a product @xmath143 where @xmath2 is a compact khler manifold , the set of boundary conditions for which there exists a regular solution to the hmae is open ( in a suitable topology ) .
recall that such a boundary condition consists of a family of cohomologous khler forms @xmath144 on @xmath2 for @xmath145 . if @xmath146 were independent of @xmath147 then the trivial solution to the hmae obtained by pulling back @xmath29 to the product is regular .
thus we have regular solutions to any small perturbation of such a boundary condition .
of course we are not interested in the product , but in the deformation to the normal cone @xmath35 . by working locally around a fixed point @xmath27
it is possible by a simple change of coordinates to translate the hmae on a neighbourhood of @xmath46 to one on a product .
the cost in this change is that whereas we originally were seeking a solution with @xmath5-invariant boundary data , the new boundary data will no longer be @xmath5-invariant .
we shall show , however , that by shrinking attention to a sufficiently small neighbourhood of @xmath148 any boundary data can be made sufficiently close to an invariant one ; and thus we can apply donaldson s openness theorem .
thus for any point @xmath27 we get a local regular solution to the hmae around @xmath46 .
we then observe that regular solutions to hmae enjoy some strong uniqueness properties ( even if they are defined locally ) coming from the existence of the associated foliation .
we use this to show that these local solutions glue together to give a regular solution in a neighbourhood of the proper transform @xmath50 of @xmath44 .
the fact that the solutions we produce are cohomologous to @xmath77 gives the existence of a hamiltonian function , and we use this to define the notion of canonical radius @xmath86 and to get a canonical regular solution to the hmae . to prove that , if @xmath1 is compact ( or more generally if @xmath88 ) , the weak solution agrees with the canonical regular one in a neighbourhood of @xmath104 we use the legendre transform of the potential @xmath60 to deduce that on some neighbourhood of @xmath104 this @xmath60 is bounded from above by a local potential for the weak solution .
the fact that @xmath149 is harmonic along the leaves of the monge - ampre foliation allows us to use the maximum principle to deduce the reverse inequality . for the result about the envelopes @xmath114 ,
the key point is that these can be shown to agree with the legendre transform of the potential for the canonical regular solution .
the associated foliation on @xmath35 can then be translated to this optimal regularity result on @xmath2 .
* tubular neighbourhoods : * it is well known that holomorphic tubular neighbourhoods around a complex submanifold @xmath6 need not exist .
in fact if one does exist then the exact sequence @xmath150 must split holomorphically .
there are stronger results on non - existence , for instance it is a theorem of van de ven @xcite that the only connected submanifolds of projective space that admit holomorphic tubular neighbourhoods are linear subspaces .
the local structure around a holomorphic submanifold @xmath1 is studied in seminal works of grauert @xcite and griffiths @xcite , in which some basic notions in algebraic geometry are developed , for example infinitesimal neighbourhoods , notions of positivity of vector bundles and subvarieties , and some vanishing theorems ( see @xcite for a survey ) . in this , and related works , what is often sought after is a `` transverse foliation '' by which is meant a family of disjoint subvarieties @xmath151 for @xmath152 such that @xmath153 .
we have avoided the use of this terminology , but what we produce is clearly analogous only we do not require that these subvarieties vary holomorphically in @xmath154 .
this highlights the key difference : the existence of such a foliation varying holomorphically places strong restrictions on the normal bundle , but by allowing a weaker notion we have existence always .
of course one of the main uses of the deformation to the normal cone in algebraic geometry is to get around the non - existence of holomorphic tubular neighbourhoods ( as used , for example , in intersection theory ) .
thus it is not particularly surprising that this same deformation appears here .
* geodesics in the space of khler metrics : * the dirichlet problem for the complex hmae has a long history , going back at least as far as the fundamental work of bedford - taylor @xcite .
the existence of smooth ( or regular ) solutions is a difficult and much studied problem and can depend in a subtle way on the boundary data ( see @xcite for a survey ) .
following work of semmes @xcite , mabuchi @xcite , donaldson @xcite , it is known that in the case of compact fibres the existence of solutions can be interpreted as geodesic rays in the space of khler metrics , and the existence of regular solutions has deep implications to the theory of extremal metrics ( for example @xcite among others ) . in the above
we consider this problem for the deformation to the normal cone of a submanifold @xmath1 .
this family is the simplest non - trivial example of a _ test - configuration _ that lies at the heart of the yau - tian - donaldson conjecture connecting the existence of a constant scalar curvature khler metric with the algebro - geometric notion of k - stability , and has been studied from a number of points of view ( for example @xcite ) .
so another interpretation of theorem [ thm : mainhmae2 ] is that the phong - sturm geodesic in the space of khler metrics associated to the test - configuration given by the deformation to the normal cone of a submanifold @xmath1 is regular near the orbit of @xmath1 .
* envelopes , equilibrium set and the hele shaw flow : * the kind of envelopes we consider above are basic objects in the study of the monge - ampre equation ( see for example ( * ? ? ?
they are related to the asymptotic behaviour of the partial bergman kernel involving holomorphic sections that vanish to a particular order along the submanifold @xmath1 .
this was first studied in detail in the toric case by schiffman - zelditch @xcite who introduced the name _ forbidden region _ essentially for the complement of what we call the equilibrium set .
this was then taken up by berman @xcite in the general projective case ( from whom we have taken the terminology _ equilibrium set _ ) who proves , among other things , that the envelopes are @xmath155 , and then again by berman - demailly @xcite in the setting of big khler classes .
we refer the reader also to ross - singer @xcite and ross - witt - nystrm @xcite where these envelopes are considered further . in the particular case
that @xmath1 is a single point in a riemann surface @xmath156 , the tubular neighbourhoods produced here describe a model of the hele - shaw flow with empty initial condition and permeability encoded by @xmath29 .
this has been studied by hedenmalm - shimorin @xcite for real - analytic hyperbolic riemann surfaces . using the hele - shaw flow
they construct an exponential map @xmath157 which has the property that the image of any concentric circle intersects orthogonally the image of any ray emanating from zero .
one can check that this implies that their exponential map differs from ours , and does not share the properties listed in theorem [ thm : tubularmain ] .
the hele - shaw flow on a general riemann surface is the topic of previous work of the authors @xcite in which we prove the @xmath158 case of theorem [ thm : flowpoint ] .
the link is made through the fact that the hele - shaw flow is characterized by its complex moments , or said another way the reproducing property . in this way we can consider theorem [ thm : flowpoint ] as a generalization of the hele - shaw flow to a point in @xmath132 ( but
do not claim that this has any fluids interpretation ) .
* acknowledgements : * we wish to thank robert berman and bo berntdsson for their encouragement and input , and to sbastien boucksom who at an early stage pointed out the connection to tubular neighbourhoods in the symplectic category . during this work
jr was supported by an epsrc career acceleration fellowship ( ep / j002062/1 ) .
dwn has received funding from the people programme ( marie curie actions ) of the european union s seventh framework programme ( fp7/2007 - 2013 ) under rea grant agreement no 329070 , and previously by chalmers university and the university of gothenburg .
* deformation to the normal cone : * let @xmath1 be a complex submanifold of a complex manifold @xmath2 ( which we always take to be connected , closed as a subset and not equal to @xmath2 ) .
the _ deformation to the normal cone _
@xmath35 of @xmath1 is the blowup @xmath159 of @xmath37 along @xmath160 .
the _ exceptional divisor _
@xmath161 is isomorphic to the projective completion @xmath42 of the normal bundle @xmath4 of @xmath1 .
the map @xmath162 is an isomorphism away from @xmath41 and the _ proper transform _ of a submanifold @xmath163 not contained in @xmath160 is the closure of @xmath164 in @xmath35 .
we also let @xmath165 and @xmath166 denote the projections to @xmath2 and @xmath65 respectively . * plurisubharmonic functions : * the following notions can be found in @xcite . for an open subset @xmath8 of a complex manifold @xmath2
we denote by @xmath167 the space of plurisubharmonic functions @xmath168 on @xmath8 .
we say a function @xmath133 on @xmath8 is _ pluriharmonic _ if both @xmath133 and @xmath169 are plurisubharmonic , and @xmath133 is _ strictly plurisubharmonic _ if its curvature @xmath170 is a strictly positive current .
any khler form @xmath29 can be written locally as @xmath170 for some smooth strictly plurisubharmonic function @xmath133 , which is uniquely defined up to addition by pluriharmonic functions .
given a closed real @xmath51-form @xmath171 on @xmath2 we let @xmath172 be the set of upper - semicontinuous @xmath108 functions @xmath109 such that @xmath173 is a positive current .
so if @xmath174 on @xmath8 then @xmath175 if and only if @xmath176 .
* foliations : * a complex @xmath177-dimensional foliation on a @xmath178-dimensional complex manifold @xmath2 consists of a set @xmath179 of @xmath177-dimensional complex submanifolds that cover @xmath2 , that are pairwise disjoint , and so that any @xmath180 is contained in a neighbourhood @xmath8 for which there is a smooth mapping @xmath181 of maximal rank such that @xmath182 .
we call the members of @xmath179 the _ leaves _ of the foliation . by the frobenius theorem ,
any involutive @xmath177-dimensional distribution of the tangent bundle of @xmath2 is integrable , and thus this distribution is in fact the tangent space to the leaves of a uniquely defined foliation .
moreover if this distribution is complex ( i.e. consists of complex subspaces of the tangent bundle ) then the leaves of this foliation are complex submanifolds by the theorem of levi - civita ( see the appendix of @xcite ) .
* @xmath183-norms : * when @xmath139 is a smooth function on an open set @xmath184 we will let the @xmath183-norm of @xmath139 be defined as @xmath185 on a complex manifold @xmath2 we choose an open cover of coordinate patches @xmath186 together with a partition of unity @xmath187 , and the @xmath183-norm of a smooth function @xmath139 on @xmath2 will be computed as @xmath188 the @xmath183-norm of functions defined on @xmath189 is then defined in the obvious way .
suppose that @xmath190 is a surjective map from a complex @xmath191 dimensional manifold @xmath10 with boundary to the closed unit disc @xmath65 such that all fibres @xmath192 are manifolds
. in this paper @xmath10 will always be either a product @xmath193 or a subset of the deformation to the normal cone @xmath35 of a submanifold @xmath6
. we shall refer to a smooth family of khler forms @xmath146 on @xmath64 for @xmath57 as _ boundary data _ for @xmath10 .
if @xmath10 is a subset of @xmath95 then the khler form @xmath29 on @xmath2 induces the boundary condition @xmath194 which we refer to as the _ boundary condition induced by _ @xmath29 .
we say that a smooth closed real @xmath51-form @xmath52 on @xmath10 is a _ regular solution to the hmae _ with boundary data @xmath146 if 1 .
@xmath53 is khler for all @xmath195 2 .
@xmath196 for all @xmath57 , and 3 .
@xmath58 on @xmath10 .
the distribution defined by the kernel of a regular solution @xmath52 to the hmae is integrable ( since @xmath52 is closed ) , complex ( since @xmath52 is @xmath51 ) , and one - dimensional and transverse to the fibers ( since @xmath52 is nondegenerate along the fibers ) ( see @xcite or @xcite ) . thus by the frobenius integrability theorem there is @xmath197-dimensional complex foliation of @xmath10 , and by construction the restriction of @xmath52 to each such leaf vanishes .
we shall refer to this as the _ monge - ampre foliation _ determined by @xmath52 .
[ thm : donaldsonopenness ] let @xmath12 be compact khler and set @xmath193 .
suppose that for some function @xmath198 on @xmath10 the form @xmath199 is a regular solution to the hmae with boundary data @xmath200 .
then any smooth function @xmath201 on @xmath189 that is sufficiently close to @xmath202 with respect to the @xmath183-norm can be uniquely extended to smooth function @xmath203 on @xmath10 such that @xmath204 is a regular solution to the hmae with boundary data @xmath205 .
clearly this applies when @xmath206 and in fact we will only use the above theorem in this form ( i.e. to obtain perturbations of this trivial solution ) . for convenience we record here the precise consequence that we need in the non - compact case .
[ thm : donaldsonopennesslocal ] there is an @xmath207 such that the following holds : if @xmath208 is a smooth family of smooth strictly plurisubharmonic functions on the unit ball @xmath209 such that 1 . for all @xmath57 the function
@xmath210 is equal to @xmath211 near @xmath212 2 .
@xmath213 then there exists a unique smooth plurisubharmonic function @xmath60 on @xmath214 with @xmath215 for any @xmath100 @xmath60 being equal to @xmath211 near @xmath216 and such that @xmath217 is a regular solution to the hmae on @xmath214 with boundary data @xmath218 . moreover if @xmath219 is a holomorphic @xmath5-action on @xmath214 which covers the usual rotation on the disc ( or its inverse ) , leaving the boundary data @xmath210 and the function @xmath211 fixed , then the solution @xmath60 is also invariant under this action .
one approach to prove this is simply to observe that donaldson s proof of theorem [ thm : donaldsonopenness ] works in this noncompact setting because we require @xmath210 to be equal to @xmath211 near the boundary .
but instead of relying on this fact , we shall show how this non - compact version can actually be deduced from the compact case .
let @xmath220 , and consider the inclusions @xmath221 .
then the khler form @xmath222 on @xmath223 can clearly be extended to a khler form @xmath29 on @xmath224 . as mentioned above
, the trivial solution to the hmae on @xmath37 with boundary data induced by @xmath29 is regular .
now let @xmath210 be a family of smooth plurisubharmonic functions that are equal to @xmath211 near the boundary of @xmath223 .
the function @xmath225 then extends by zero to @xmath226 . by donaldson s openness theorem there
exists an @xmath207 such that if the @xmath183-norm of @xmath227 is less than @xmath228 then @xmath227 extends to a smooth function @xmath203 on @xmath229 such that @xmath230 is a regular solution to the hmae .
we can now set @xmath231 .
the fact that @xmath203 is zero near @xmath232 relies on aspects of donaldson s proof of theorem [ thm : donaldsonopenness ] .
the first step in that proof is a construction of a fibre bundle @xmath233 a khler manifold ( that is locally modelled on the cotangent bundle ) so that there is a correspondence between khler forms of the form @xmath234 and the images of certain sections of @xmath235 ( the requirement being that this image is a `` ls - manifold '' in donaldson s terminology ) .
thus a boundary data @xmath146 defines a family of ls - submanifolds of @xmath235 parameterized by @xmath57 . in our case
the ls - submanifold is locally the graph of @xmath236 .
the second step is to show that the foliation defined by a regular solution to the hmae with this boundary data lifts to a family of holomorphic discs in @xmath235 parameterized by @xmath2 whose boundary lie in these ls - submanifolds .
moreover any holomorphic disc with this boundary condition that is sufficiently close to this family is actually a leaf of this foliation ( * ? ? ? * prop 2 ) .
we now let @xmath237 be the regular solution with boundary values @xmath238 @xmath239 $ ] .
the ls - submanifolds at time @xmath240 will thus be locally defined as @xmath241 .
also note at time @xmath242 we have the trivial solution , and thus foliation at that time is given by the constant discs @xmath243 .
since these ls submanifolds do not change with @xmath240 over a neighbourhood of @xmath244 it follows that the constant discs for @xmath245 near @xmath244 remain to be part of the foliation , which forces @xmath203 to be zero near @xmath232 .
any solution @xmath246 with the stated properties will extend to a global solution on @xmath247 which by the uniqueness part of theorem [ thm : donaldsonopenness ] is equal to @xmath248 thus showing that @xmath249 that @xmath60 is invariant under a holomorphic action which fixes the boundary data and @xmath211 now follows from the uniqueness of @xmath250 since any rotation of @xmath60 gives a regular solution to the hmae with the same boundary values .
[ lemma : invariantleaves ] if @xmath52 is a regular solution to the hmae on @xmath214 which is invariant under the action @xmath219 of @xmath5 where @xmath251 then the constant discs @xmath252 are all leaves of the associated monge - ampre foliation .
we know that the monge - ampre foliation is @xmath219-invariant so since any point of the form @xmath253 if fixed by @xmath219 the leaf passing through that point must be @xmath219-invariant itself .
we let @xmath254 @xmath255 be the @xmath256:th coordinate of the point of that leaf in the fiber over @xmath195 then @xmath139 is holomorphic at least in some small neighbourhood of the origin . from the invariance
follows that for any @xmath257 @xmath258 which immediately implies that @xmath139 must be constantly zero ( e.g. from looking at the winding number ) .
thus since the leaf is locally constant it must be globally constant , and hence equal to the constant disc @xmath252 .
let @xmath1 be a complex submanifold of a khler manifold @xmath12 and consider the family @xmath259 given by the deformation to the normal cone of @xmath1 .
we recall that we say that a leaf of a one dimensional foliation on a subset of @xmath35 is _ complete _ if the projection from this leaf to the base @xmath65 is a surjection .
[ def : localregularsolution ] a _ local regular solution _ to the hmae with boundary data induced by @xmath29 for a subset @xmath260 consists of a pair @xmath261 such that 1 . @xmath10 is an open @xmath5-invariant neighbourhood of the proper transform of @xmath262 in @xmath35 such that all fibers @xmath64 are smooth manifolds , 2 .
@xmath60 is a smooth real valued @xmath5-invariant function on @xmath10 which is zero on @xmath263 @xmath100 3 .
@xmath264 is a regular solution to the hmae on @xmath10 , 4 .
@xmath10 is the union of complete leaves of the monge - ampre foliation determined by @xmath52 , and if a leaf intersects @xmath104 then it must be equal to @xmath265 for some @xmath27 .
note that it follows from the definition that if @xmath261 is a local regular solution to the hmae for @xmath266 then @xmath265 is part of the monge - ampre foliation for each @xmath267 .
[ prop : hmaelocal ] let @xmath27 .
then there exists a local regular solution @xmath261 to the hmae for the subset @xmath268 . as discussed previously ,
the idea is to reduce the problem to finding regular solutions to the hmae over a neighbourhood of @xmath265 which looks like a ball times the disc , i.e. the product case , but with a twisted boundary condition . to set this up ,
let @xmath269 be coordinates centered at @xmath148 chosen so that @xmath1 is given locally by @xmath270 ( these coordinates will later be chosen with additional properties ) . for simplicity we shall write @xmath271 where @xmath272 and @xmath273
. let @xmath274 denote the corresponding holomorphic embedding of the ball of radius @xmath275 where @xmath276 is chosen small enough so that @xmath1 is given by @xmath270 on the chart @xmath277 .
let also @xmath278 .
consider the map @xmath279 given by @xmath280 from the definition of the blowup one easily sees that this map extends to a biholomorphism from @xmath281 to an open set @xmath282 that contains @xmath265 .
henceforth we shall let @xmath283 denote this extended biholomorphism .
thus @xmath284 is a holomorphic chart on @xmath35 containing @xmath46 .
recall the holomorphic action @xmath219 of @xmath5 on @xmath285 @xmath286 under the biholomorphism @xmath287 the action @xmath219 is the same as the @xmath5 action on @xmath95 considered above restricted to @xmath288 .
we shall therefore refer to this data as a @xmath219-chart . by abuse of notation
we will let the holomorphic action of @xmath5 on @xmath223 given by @xmath289 also be denoted by @xmath219 .
[ prop : translation ] there exists a local regular solution to the hmae for the subset @xmath268 if and only if there exists a @xmath219-chart @xmath290 centered at @xmath148 together with a smooth real @xmath219-invariant function @xmath246 on @xmath281 such that @xmath291 is a regular solution to the hmae with boundary condition @xmath292 .
if @xmath261 is a local regular solution to the hmae for the subset @xmath268 then pick a @xmath219-chart @xmath290 centered at @xmath148 such that @xmath293 .
let @xmath294 where @xmath295 .
since the usual @xmath5-action on @xmath95 restricts to @xmath219 on the coordinate chart @xmath296 @xmath246 is @xmath219-invariant , and since @xmath287 is a biholomorphism @xmath297 is a regular solution to the hmae on @xmath281 .
one easily checks that it has boundary values @xmath292 . for the other direction ,
let us assume that we have a @xmath219-chart and an @xmath219-invariant function @xmath246 on @xmath281 such that @xmath291 is a regular solution to the hmae with boundary condition @xmath292 .
let @xmath298 . thanks to lemma [ lemma : invariantleaves ] we know that @xmath299 is part of the monge - ampre foliation on @xmath296 so by continuity there exists a neighbourhood @xmath300 of @xmath299 consisting of complete leaves .
since the the foliation is @xmath219-invariant we can take the union of all @xmath301 which is then a @xmath219-invariant neighbourhood @xmath300 of @xmath299 consisting of complete leaves , and we now call this neighbourhood @xmath300 . because @xmath287 is a biholomorphism , @xmath302 on @xmath303 and @xmath219 corresponds to the restriction of the usual @xmath5-action on @xmath95 to @xmath304 we get that @xmath261 where @xmath305 is a local regular solution to the hmae for the subset @xmath268 .
thus it is enough to find a @xmath219-chart @xmath290 centered at @xmath148 together with a @xmath219-invariant function @xmath246 such that @xmath291 is a regular solution to the hmae on @xmath281 with boundary data @xmath306 .
the first observation is that , given a @xmath219-chart , if @xmath307 happened to be @xmath219-invariant then the boundary data is independent of @xmath147 , and so we have the trivial regular solution @xmath308 where @xmath133 is @xmath219-invariant and @xmath295 .
of course @xmath307 will not in general be @xmath219-invariant , but we shall show that one always can find charts such that @xmath307 is approximately @xmath219-invariant .
thus donaldson s openness theorem can be applied to give the desired local regular solution .
[ lemma : coordinates ] there exists a @xmath219-chart centered at @xmath148 such that @xmath309 with @xmath310 for completeness we include the standard argument .
let @xmath311 with @xmath312 and @xmath313 be coordinates centered at @xmath148 defining some @xmath219-chart , and thus @xmath1 is given locally by @xmath314 . also choose a potential @xmath109 for @xmath307 .
the fact that @xmath109 is real valued implies that the taylor expansion around @xmath242 takes the form @xmath315 for some real coefficients @xmath316 with @xmath317 clearly @xmath318 is pluriharmonic since it is the real part of a holomorphic function . as @xmath109 is strictly plurisubharmonic
, @xmath287 is a positive definite hermitian .
thus by choosing an orthonormal basis with respect to the hermitian form given by @xmath287 one gets a linear change of coordinates @xmath319 so that @xmath320
. moreover this can be achieved whilst preserving a given subspace ( say by first picking such an orthonormal basis for this subspace ) .
thus there exits a matrix with block form @xmath321 so that @xmath322 . letting @xmath323 we get a new @xmath219-chart @xmath324 and @xmath325
now @xmath326 where @xmath327 which clearly is of the right form .
a chart as in lemma [ lemma : coordinates ] will be called an admissible @xmath219-chart .
we see that in such a chart the local potential @xmath133 is close in @xmath183 to @xmath211 near @xmath242 .
the idea is now to find a smooth strictly plurisubharmonic function @xmath328 on @xmath223 that is equal to @xmath133 near @xmath242 while ( after a scaling and a constant shift ) being @xmath228-close in @xmath183 to @xmath211 on the whole of @xmath223 and actually equal to @xmath211 near @xmath244 .
this will subsequently allow us to apply the local version of donaldson s openness theorem ( theorem [ thm : donaldsonopennesslocal ] ) to get a local regular solution for @xmath268 . to get @xmath328 we will glue together @xmath133 with @xmath329 where @xmath330 and for this we need the following lemma : [ lem : regmax ] let @xmath331 and @xmath332 be smooth strictly plurisubharmonic functions on some domain @xmath333 containing the origin , and assume that for some @xmath334 we have that @xmath335 while @xmath336 on @xmath337 .
then there exists a smooth strictly plurisubharmonic function @xmath338 on @xmath8 which is equal to @xmath171 near @xmath339 equal to @xmath340 near @xmath341 and such that @xmath342 where @xmath343 denotes the maximum of @xmath344 and @xmath345 for @xmath346 note that @xmath347 is strictly plurisubharmonic , equal to @xmath171 near @xmath242 and equal to @xmath340 near @xmath341 but not smooth .
instead we let @xmath139 be a smooth even non - negative bump function on the real line with unit integral , support in @xmath348 which is bounded from above by @xmath349 and then define @xmath338 to be @xmath350 that @xmath23 is smooth , strictly plurisubharmonic and has the right behaviour near @xmath242 and @xmath337 is easily checked .
note that @xmath351 and the bound of the @xmath183-norm of @xmath352 on @xmath8 follows from simply differentiating this expression and using the obvious bounds .
[ prop : approxharmonic ] given an admissible @xmath219-chart @xmath277 centered at @xmath148 and an @xmath207 we can find a smooth strictly plurisubharmonic function @xmath328 on @xmath223 such that for some constant @xmath353 @xmath354 near @xmath339 while at the same time @xmath355 near @xmath244 and @xmath356 let @xmath133 be a local potential on @xmath357 such that @xmath295 and @xmath358 .
we are going to glue @xmath133 with @xmath359 where @xmath360 are some small numbers yet to be determined . clearly @xmath361 .
since @xmath362 when @xmath363 we get that for @xmath364 small enough @xmath365 on @xmath366 where say @xmath367 .
for this to make sense we need to choose @xmath364 small enough so that @xmath368 in order for @xmath133 to be defined on @xmath369 . by lemma
[ lem : regmax ] there is a smooth strictly plurisubharmonic function @xmath23 on @xmath369 which is equal to @xmath133 near @xmath339 equal to @xmath340 near @xmath366 and with @xmath370 we note that as @xmath364 and thus @xmath371 tends to @xmath339 @xmath372 will converge to @xmath373 . also for small @xmath364 we have that @xmath374 put together this means that by letting @xmath364 be small enough we can get the @xmath183-norm of @xmath23 to approach @xmath375 .
so then we pick @xmath376 such that @xmath377 and we will be able to get a smooth strictly plurisubharmonic function @xmath23 defined on some small ball @xmath369 which is equal to @xmath133 near @xmath339 equal to @xmath340 near @xmath366 and with @xmath378 now let @xmath328 be defined as @xmath379 on @xmath380 and as @xmath328 now is equal to @xmath211 near @xmath366 we can extend it to @xmath223 by simply letting it be @xmath211 on @xmath381 .
it is now immediate that this @xmath328 has the desired properties .
we are now ready to prove our local existence result .
pick an admissible @xmath219-chart @xmath277 centered at @xmath148 , let @xmath207 be that from theorem [ thm : donaldsonopennesslocal ] , and let @xmath328 be a function on @xmath223 with the properties stated in proposition [ prop : approxharmonic ] .
thus @xmath356 let @xmath382 as is easily seen @xmath383 so it follows from theorem [ thm : donaldsonopennesslocal ] that there exists a @xmath219-invariant function @xmath246 such that @xmath384 when @xmath57 and such that @xmath291 is a regular solution to the hmae on @xmath214 .
now on some small ball @xmath385 we have that @xmath354 and thus on @xmath386 we have that @xmath387 .
it follows that @xmath388 restricted to @xmath389 is a @xmath219-invariant regular solution to hmae with the right boundary values .
proposition [ prop : translation ] now tells us that we get a local regular solution to the hmae for @xmath268 .
we now show how the local solutions provided in section [ sec : local ] patch together .
this will follow from a local uniqueness property for these regular solutions . [
prop : localuniq ] if @xmath390 is a local regular solution to the hmae for a subset @xmath266 and @xmath391 is a local regular solution to the hmae for a subset @xmath392 then in fact @xmath393 in some neighbourhood of the proper transform of @xmath394 .
pick a point @xmath395 and a @xmath219-chart @xmath277 centered at @xmath148 such that @xmath396 .
let @xmath133 be a function such that @xmath295 and let @xmath397 and @xmath398
. then @xmath246 and @xmath399 are two @xmath219-invariant regular solutions to the hmae on @xmath281 with the same boundary values .
let @xmath400 be a complete leaf of the foliation associated to @xmath401 clearly we have that @xmath246 is harmonic along @xmath400 .
since @xmath402 on the boundary of @xmath400 while @xmath399 is subharmonic on @xmath400 it follows from the maximum principle that @xmath403 on @xmath400 .
we know that the constant disc @xmath404 is part of the monge - ampre foliation of @xmath291 ( as well as that of @xmath405 ) and thus a neighbourhood of @xmath299 is foliated by such complete leaves , so in this neighbourhood @xmath403 .
but arguing the same way using complete leaves of the foliation associated to @xmath406 now gives us that in fact @xmath402 in a neighbourhood of @xmath299 .
pulling back by @xmath284 yields the identity of @xmath407 and @xmath408 near @xmath409 and we are done . for spaces
@xmath410 where the fibers are compact manifolds without boundary there exist general uniqueness results for the homogeneous monge - ampre equation even for weak solutions .
however the local uniqueness result stated here is quite different in that it does not depend on the behaviour of the solution near the boundary of @xmath10 .
if @xmath261 is a local regular solution to the hmae and @xmath411 then we let @xmath412 be defined as the union of all leaves of the associated monge - ampre foliation which connect to @xmath413 .
note that this is an @xmath5-invariant subset of @xmath95 .
given @xmath207 we will let @xmath414 denote the set of points in @xmath2 within an @xmath228 distance of @xmath415 ( measured using the khler metric ) .
[ lem : locpatch ] if @xmath416 are two compact subsets of @xmath1 and @xmath390 is a local regular solution to the hmae for @xmath417 while @xmath391 is a local regular solution to the hmae for @xmath418 then there exists a local regular solution @xmath261 to the hmae for @xmath419 .
moreover if @xmath415 is a compact subset of @xmath2 such that @xmath420 then we can choose @xmath261 so that for some @xmath207 we have that @xmath421 and @xmath422 on @xmath423 . by proposition [ prop :
localuniq ] , for any point @xmath395 we can pick an @xmath5-invariant neighbourhood @xmath424 of @xmath128 in @xmath425 where @xmath426 consisting only of complete leaves of the associated foliation . let @xmath8 denote the union of those neighbourhoods .
now we pick a compact subset @xmath427 such that @xmath420 . by property
( 4 ) of a local regular solution ( see definition [ def : localregularsolution ] ) we have that @xmath428 is equal to the proper transform of @xmath429 .
this implies that when @xmath228 is small enough @xmath430 is a neighbourhood of the proper transform of @xmath431 .
we then define @xmath60 to be equal to @xmath407 on @xmath432 while letting it be @xmath408 on @xmath433 and then equal to either one on @xmath8 .
this now gives local regular solution @xmath261 to the hmae for @xmath419 .
the following gives a proof of theorem [ defconethm ] .
as before the data consists of a khler manifold @xmath12 ( not necessarily compact ) together with a complex submanifold @xmath6 .
[ thm : hmae : repeat ] there exists a local regular solution @xmath261 to the hmae for @xmath1 .
in particular it means that we have an @xmath5-invariant neighbourhood @xmath10 of @xmath50 in @xmath35 and a smooth closed @xmath5-invariant real @xmath51-form @xmath264 on @xmath10 that gives a regular solution to the homogeneous monge - ampre equation with boundary data induced by @xmath29 , i.e. 1 .
@xmath53 is khler for all @xmath195 2 .
@xmath56 for all @xmath57 , and 3 .
@xmath58 on @xmath10 .
moreover the germ around @xmath50 of any such @xmath5-invariant regular solution is unique .
using proposition [ prop : hmaelocal ] there is a collection @xmath434 ( @xmath435 with @xmath436 or @xmath437 ) of local regular solutions to the hmae for compact subsets @xmath438 such that @xmath439 is a locally finite cover of @xmath1 .
let @xmath440 . then by lemma
[ lem : locpatch ] , for some @xmath441 we can find a locally regular solution @xmath442 to the hmae for @xmath443 such that @xmath444 and so that @xmath445 there . for @xmath446 we let @xmath447 and do the same thing , and we will get a sequence of local regular solutions @xmath448 to the hmae for @xmath449 .
we note that the subsets @xmath450 are all @xmath5-invariant and increase with @xmath451 and for @xmath452 @xmath453 on @xmath450 .
let @xmath454 and define @xmath60 by letting it be @xmath455 on @xmath456 then it is immediate that @xmath261 has the properties described in the theorem .
this proves the existence part , while the uniqueness of the germ follows directly from proposition [ prop : localuniq ] .
in fact we can say a bit more than just that the germ of any regular solution is unique , but for this we need to introduce the notion of complete regular solutions . we will call a pair @xmath63 a _ complete regular solution _ to the hmae ( or in short a _ complete solution _ ) if it is a regular solution to the hmae as in theorem [ defconethm ] ( so in particular it is cohomologous to @xmath77 ) such that @xmath10 is foliated by complete leaves of the monge - ampre foliation , @xmath72 and whenever @xmath73 then @xmath74 for all @xmath75 . given a regular solution @xmath63 as in theorem [ defconethm ] , by shrinking @xmath10 we can always get a nontrivial complete solution .
[ prop : completeuniq ] if @xmath63 and @xmath457 are two complete solutions and @xmath458 then @xmath459 and @xmath460 on @xmath10 .
let @xmath60 and @xmath246 be the potentials of @xmath52 and @xmath297 respectively . pick a point @xmath73 and let @xmath461 denote the leaf in the monge - ampre foliation of @xmath52 that passes through @xmath462 for @xmath239 $ ] .
let also @xmath463 denote the leaf in the monge - ampre foliation of @xmath297 that passes through @xmath462 .
we know that @xmath464 .
let @xmath465 .
we claim that @xmath466 . indeed , if @xmath467 then there is a neighbourhood @xmath468 of @xmath461 consisting of complete leaves of the monge - ampre foliation of @xmath52 . on any such leaf
@xmath469 is subharmonic , and since this function is zero on the boundary of the leaf we get that @xmath470 . similarly , using a neighbourhood of @xmath467 consisting of leaves of the foliation associated to @xmath297
we get that in fact @xmath471 in a neighbourhood of @xmath472 and hence the foliations agree there .
this implies that the set of points @xmath239 $ ] such that @xmath467 is open . on the other hand it is closed since
the leaves vary continuously with @xmath240 .
this thus shows that @xmath473 and also that @xmath471 on that leaf .
since @xmath10 is foliated by such leaves it gives us the proposition .
note that in the proof we only used the facts that @xmath10 and @xmath300 were foliated by complete leaves , @xmath474 @xmath475 connected and @xmath458 .
let @xmath76 denote the vector field generated by the @xmath5-action on @xmath10 .
from the fact that @xmath52 is cohomologous to @xmath77 follows that the @xmath5-action is hamiltonian , in the sense described in the next theorem .
[ thm : ham2 ] let @xmath63 be a regular solution to the hmae as in theorem [ defconethm ] . then the function @xmath78 is a hamiltonian for the @xmath5-action , in that it satisfies @xmath476 @xmath80 is constant along the leaves of the monge - ampre foliation associated to @xmath52 .
furthermore if @xmath63 is complete then @xmath81 with equality precisely on @xmath50 . from the definition of the @xmath477-operator we get that @xmath478 but on the other hand clearly @xmath479 and thus @xmath78 is a hamiltonian for the @xmath5-action .
the lie derivative of @xmath80 with respect to the vector field @xmath480 obtained by the flow along a leaf of the associated foliation is @xmath481 where we have used and that by definition @xmath480 lies in the kernel of @xmath52 . thus @xmath80 is constant along leaves .
as @xmath60 is zero on @xmath50 and @xmath76 is tangential to @xmath50 we get that @xmath482 along @xmath50 .
it remains to prove that @xmath80 is strictly positive away from @xmath50 when the solution is complete . since @xmath80 is constant along the leaves of the foliation it suffices to consider @xmath80 on the discs @xmath483 .
we have that @xmath484 and since @xmath77 restricts to zero on the disc the function @xmath485 is strictly convex and @xmath242 at @xmath486 , which implies that @xmath487 as long as @xmath488 .
we will now show how to define a canonical complete solution @xmath82 , at least when @xmath1 is compact . given a complete solution @xmath63 with hamiltonian @xmath80
we define the radius @xmath83 of @xmath63 to be the supremum of all @xmath84 such that @xmath85 .
we then define the _
canonical radius _
@xmath86 to be the supremum of @xmath83 over all complete solutions @xmath63 .
note that the radius of a complete solution could be zero , and hencethe canonical radius @xmath86 could also be zero .
but at least when @xmath1 is compact we clearly have that @xmath87 and thus @xmath88 . [ lem : completerestr ] if @xmath63 is a complete solution and @xmath105 then @xmath489 is a new complete solution .
since @xmath80 is constant along the leaves we see that @xmath490 is still foliated by complete leaves , and from the proof of theorem [ thm : ham2 ] we see that @xmath491 when @xmath195 showing that the discs @xmath492 lie in @xmath493 when @xmath23 does .
[ prop : completeuniq2 ] if @xmath63 and @xmath457 are two complete solutions with @xmath494 then letting @xmath495 we have that @xmath496 and @xmath460 there .
clearly we can assume that @xmath105 since otherwise the statement is vacuous . by lemma [ lem : completerestr ] the solutions @xmath489 and @xmath497 are still complete , so for ease of notation we may as well assume that @xmath498 and @xmath499 ( and hence @xmath500 ) . pick a point @xmath501 then we know that @xmath502 .
we now use an argument similar similar to the one in the proof of proposition [ prop : completeuniq ] .
thus we let @xmath461 denote the leaf in the monge - ampre foliation of @xmath52 that passes through @xmath462 for @xmath239 $ ] .
we also let @xmath463 denote the leaf in the monge - ampre foliation of @xmath297 that passes through @xmath503 if there is any such leaf . we let @xmath465 and
again we claim that @xmath466 .
that the set of points @xmath239 $ ] such that @xmath467 is open follows exactly as before . on the other hand ,
@xmath504 which since @xmath500 implies that @xmath505
. then the closedness follows as before since the leaves vary continuously with @xmath506 which exactly as in the proof of proposition [ prop : completeuniq ] gives us the desired equalities .
armed with this result it is immediate how to construct a canonical complete solution @xmath82 when the canonical radius @xmath86 is positive .
namely , let @xmath507 be any sequence of complete solutions such that @xmath508 is increasing to @xmath86
. then we define @xmath509 and we let @xmath102 be defined to be equal to @xmath510 on @xmath511 .
that this is a well - defined complete solution now follows immediately from proposition [ prop : completeuniq2 ] .
we have thus proved theorem [ thm : canonicalhmae ] from the introduction : [ thm : canonicalhmae2 ] let @xmath1 be compact ( or more generally assume that @xmath88 ) .
then there is a unique complete solution @xmath82 such that @xmath89 and @xmath90 this canonical solution is maximal in the following sense . if @xmath63 is any other complete solution then for any @xmath91 we have that @xmath92 and there @xmath93 .
let @xmath1 be a submanifold in a khler manifold @xmath12 . recall that @xmath0 denotes the normal bundle and @xmath3 is the inclusion of @xmath1 as the zero section in @xmath4 .
a _ tubular neighbourhood _ of @xmath1 in @xmath2 is a smooth map @xmath512 from an open neighbourhood @xmath8 of @xmath31 in @xmath4 which is a diffeomorphism onto a neighbourhood @xmath513 of @xmath1 with @xmath514 .
suppose that @xmath52 is as provided by theorem [ thm : hmae : repeat ] , so is a regular solution to the hmae on a neighbourhood @xmath282 of the proper transform @xmath50 of @xmath44 with boundary data @xmath29 .
recall that @xmath515 denotes the central fibre of @xmath10 , which is a neighbourhood of @xmath9 .
we let @xmath516 be the restriction of the regular solution @xmath52 of the hmae to @xmath8 .
so @xmath516 is an @xmath5-invariant khler form on @xmath8 , and by the uniqueness property of regular solutions , the germ of this khler form around @xmath9 is independent of choice of @xmath52 .
we recall that @xmath10 was assumed to be a union of complete leaves of the monge - ampre foliation determined by @xmath52 . thus flowing along these leaves
gives an injective smooth map @xmath517 such that @xmath518 and for each @xmath15 the map @xmath519 is holomorphic .
the family of maps @xmath520 has an additional remarkable property .
[ prop : pullback ] for any @xmath54 we have that @xmath521 this is classical ( see e.g. @xcite or @xcite ) but for the convenience of the reader we give the simple argument here .
when differentiating the left hand side of ( [ eq : pullback ] ) by a vector field @xmath522 on the base one gets @xmath523 where @xmath524 is the unique lift of @xmath522 to @xmath10 parallel to the foliation , and at the same time by cartan s formula @xmath525 since @xmath52 is closed and @xmath524 lies in the kernel .
thus the left hand side of ( [ eq : pullback ] ) is independent of @xmath526 so letting @xmath147 be zero completes the proof .
we define @xmath33 by @xmath527 [ lem : jistubular ] @xmath528 is a tubular neighbourhood of @xmath1 and @xmath529 .
in particular @xmath530 is khler and @xmath5-invariant . by construction @xmath528 is a tubular neighbourhood .
the second statement follows from proposition [ prop : pullback ] .
the next result completes the proof of theorem [ thm : tubularmain ] in the introduction .
let @xmath15 and @xmath531 be given by @xmath532 for @xmath533 .
then there exists a holomorphic @xmath534 extending @xmath535 such that @xmath536 = u.\label{eq : bound}\ ] ] for @xmath15 there is a unique holomorphic leaf @xmath537 of the monge - ampre foliation that passes though @xmath23 and by definition of @xmath24 it contains the point @xmath538 .
now let @xmath21 be the lift from @xmath65 to @xmath400 composed with the projection to @xmath2 .
it is then immediate that @xmath539 @xmath540 and @xmath541 = u$ ] .
the @xmath5-invariance of the foliation implies that @xmath542 and thus @xmath543 .
thus we get that @xmath21 extends @xmath535 . when @xmath1 is compact ( or more generally @xmath88 ) we clearly get a canonical tubular neighbourhood @xmath544 with the desired properties by using the canonical complete solution @xmath82 .
we also have the following characterization of when the germ of @xmath24 is holomorphic at a given point @xmath25 .
[ prop : hologerm ] the germ of @xmath24 is holomorphic around a point @xmath25 if and only if there exists a neighbourhood @xmath424 of @xmath148 together with a holomorphic @xmath5-action on @xmath424 such that @xmath545 is invariant , @xmath546 is fixed pointwise and the induced action on @xmath547 is equal to the usual rotation of its fibers .
if the germ of @xmath24 is holomorphic at @xmath548 then we can take @xmath424 to be the image under @xmath24 of an @xmath5-invariant neighbourhood of @xmath548 , and consider the induced action of @xmath5 on the image .
this will then have the described properties since @xmath14 is @xmath5-invariant .
let now @xmath424 be a neighbourhood of @xmath148 with the properties described above .
choose holomorphic coordinates @xmath549 centered at @xmath148 such that locally around @xmath148 we have that @xmath1 is given by the equations @xmath550 @xmath255 .
let @xmath371 denote the @xmath5-action on @xmath424 .
for @xmath255 we let @xmath551 be the holomorphic function defined by @xmath552 while for @xmath553 we let @xmath554 then these coordinates @xmath551 define a @xmath219-chart @xmath555 and it is easy to see that on this chart @xmath556 .
thus by assumption @xmath307 is @xmath219-invariant , giving rise to the trivial local regular solution @xmath557 .
the associated foliation is the trivial one , and by uniqueness ( proposition [ prop : localuniq ] ) the monge - ampre foliation of any regular solution agrees with this trivial one near @xmath265 .
in particular we see that @xmath24 is holomorphic near @xmath548 .
the next definition captures another relation between @xmath24 and the holomorphic structure .
let @xmath558 be complex manifolds with fixed inclusions @xmath559 and @xmath560 for some complex manifold @xmath10 .
we say that a smooth map @xmath33 is _ holomorphic motion _ if there exists a smooth @xmath561 such that 1 .
@xmath562 and @xmath563 for all @xmath15 .
2 . for each @xmath15
the map @xmath564 is holomorphic .
3 . for each @xmath565
the map @xmath566 is injective .
so by construction @xmath567 is a holomorphic motion ( with respect to @xmath10 ) .
this terminology is adapted from the case of a family of injections @xmath568 parameterized by @xmath54 which for fixed @xmath569 are holomorphic in @xmath147 ; we refer the interested reader to @xcite and the references therein .
recall that one can not find global regular solutions of the hmae on @xmath95 but instead of looking for local regular solutions one can consider global weak solutions .
if @xmath570 is a @xmath77-psh function on @xmath95 and @xmath571 then @xmath572 $ ] is a closed positive current cohomologous to @xmath98 $ ] .
pick a @xmath573 and let @xmath574 be defined as the supremum of all @xmath77-psh functions on @xmath575 such that @xmath576 on @xmath189 and with @xmath571 .
we call the closed positive @xmath51-current on @xmath577 @xmath578\ ] ] the weak solution to the hmae on @xmath575 .
a motivation for this comes from the following theorem .
when @xmath2 is compact and @xmath96 sufficiently small the closed positive @xmath51-current @xmath578\ ] ] solves the weak hmae on @xmath104 with boundary condition induced by @xmath29 , i.e. @xmath579 this is proved by berman in ( * ? ? ?
* prop 2.7 ) which in turn relies on the original approach of bedford - taylor @xcite ( see also demailly @xcite ) .
the cited reference is stated for certain kinds of `` test configurations '' , an example of which is @xmath35 , and under some additional assumptions such as @xmath2 being fano , and @xmath580 $ ] being integral , but these are not used in the proof .
we stress that since @xmath574 is not necessarily smooth , the term @xmath581 is to be taken in the sense of bedford - taylor .
also the boundary value statement in should be read as saying that @xmath582 uniformly as @xmath583 .
[ thm : weakagree ] assume that @xmath1 is compact ( or more generally @xmath88 ) and let @xmath82 be the canonical complete solution as provided by theorem [ thm : canonicalhmae ] .
then @xmath584 on @xmath103 .
in particular @xmath97 is regular in a neighbourhood of @xmath104 . to prove theorem [ thm : weakagree ]
we will consider the legendre transform of @xmath585 , i.e. the unique @xmath5-invariant function on @xmath586 being zero on @xmath587 such that @xmath588 .
so for @xmath589 we consider the legendre transform @xmath590 note that the function @xmath591 is convex and @xmath592 where @xmath593 is the hamiltonian .
since @xmath594 as @xmath595 approaches @xmath596 it follows that the infimum in ( [ eq : legtrans ] ) is attained in @xmath10 .
[ prop : alphalambda ] for @xmath138 the function @xmath597 is @xmath29-psh , @xmath598 and @xmath599 . that @xmath597 is @xmath29-psh follows from the kiselman minimum principle @xcite , ( * ?
* chi 7b ) . also from ( [ eq : halpha ] ) we see that @xmath600 precisely on @xmath493 . for the final statement
we will work locally around one of the discs @xmath265 .
so let @xmath277 be an admissible @xmath219-chart centered at @xmath148 .
then we can write @xmath601 where @xmath358 and @xmath602 we know that @xmath603 on @xmath357 for some number @xmath604 and it follows that @xmath605 on @xmath281 . since as we recall @xmath606 it follows that @xmath607\}\leq \\
\leq c(e^t|z'|^2+|z''|^2)-\lambda t = c\delta^2-\lambda\ln(\delta^2-|z''|^2)+\lambda \ln |z'|^2.\end{aligned}\ ] ] where @xmath608 i.e. @xmath609 .
this shows that @xmath599 . in light of proposition
[ prop : alphalambda ] we can define @xmath597 to be zero on @xmath610 and thus get an @xmath29-psh function on @xmath2 with lelong number along @xmath1 greater than or equal to @xmath126 .
now assume that @xmath611 .
then the function @xmath612 is @xmath77-psh , bounded from above by zero on @xmath189 and has lelong number at least @xmath96 along @xmath41 . from the definition of @xmath574 as the supremum of such functions we get that @xmath613 on the other hand , by the involution property of the legendre
transform @xmath614 on @xmath615 and thus we get that @xmath616 on @xmath617 . pick a leaf in the foliation of @xmath618 and let @xmath109 be a function on it such that @xmath619 .
since @xmath574 has lelong number @xmath96 along @xmath41 it follows that @xmath620 defines a subharmonic function on @xmath621 which is equal to @xmath109 on the boundary . on the other hand we know that @xmath622 is harmonic and equal to @xmath109 on the boundary , which implies that in fact @xmath623 on @xmath617 .
locally on a @xmath219-chart we would get that @xmath624 on @xmath625 and since in this picture @xmath41 is the zero fiber this then implies that @xmath626 on @xmath617 .
recall that the envelope @xmath114 we wish to consider is @xmath627 and the equilibrium set @xmath115 is @xmath628 we also recall the definition of optimal regularity for such envelopes .
we say that @xmath114 has optimal regularity if @xmath115 is smoothly bounded , the function @xmath114 is smooth on @xmath629 @xmath630 and the blowup of @xmath113 along @xmath1 is foliated by holomorphic discs attached to @xmath124 passing through @xmath1 such that the restriction of @xmath122 to each such disc vanishes .
[ optimalregthm2 ] suppose @xmath1 is compact ( or more generally @xmath88 ) . then for @xmath126 small enough ( i.e. @xmath138 ) the envelopes @xmath114 have optimal regularity , the boundaries @xmath124 vary smoothly with @xmath126 , and the corresponding holomorphic discs attaching to @xmath124 and passing through @xmath1 all have area @xmath126 .
let @xmath82 be the canonical solution to the hmae with boundary data induced by @xmath631 as provided by theorem [ thm : canonicalhmae ] , and choose a number @xmath96 such that @xmath632 ( if @xmath86 is finite we can just as well set @xmath633 ) .
let @xmath574 be corresponding weak solution as defined in section [ sec : localregularity ] .
we saw in the proof of theorem [ thm : weakagree ] that @xmath623 in @xmath617 ( here as we recall @xmath60 is the unique @xmath5-invariant function on @xmath575 equal to zero on @xmath189 such that @xmath588 and @xmath80 is the hamiltonian of @xmath82 ) .
we get by definition that @xmath634 which implies that @xmath635 since @xmath636 in @xmath617 we see that the right hand side of ( [ eq : psialpha ] ) is equal to @xmath597 on @xmath617 and since it is less than or equal to zero on @xmath2 it must be equal to @xmath597 on the whole of @xmath2 . thus @xmath637 but since @xmath597 is a candidate for the supremum we get that in fact @xmath638 thus from proposition [ prop : alphalambda ] we get that the boundary @xmath124 is equal to @xmath639 .
since @xmath80 is constant along the leaves of the foliation it follows that @xmath124 is the image of the @xmath126 level set of @xmath80 on @xmath475 under the tubular neighbourhood @xmath24 .
since @xmath52 restricted to @xmath475 is khler and @xmath76 ( i.e. the vector field generating the @xmath5-action ) is nonvanishing on @xmath640 we get that @xmath641 is nonzero away from @xmath31 .
this shows that @xmath124 is smooth and varies smoothly with @xmath126 .
let now @xmath642 be a point in @xmath116 .
since @xmath643 is foliated by leaves there is one such leaf @xmath400 which intersects the open line segment @xmath644 . at the intersection point
@xmath595 we then have that @xmath645 on @xmath621 @xmath646 and thus the function @xmath647 restricted to @xmath400 is thus subharmonic , bounded from above by zero and equal to zero at an interior point @xmath648 and thus must be equal to zero .
therefore @xmath649 on @xmath643 and @xmath650 on @xmath400 .
this now shows that @xmath651 is smooth on @xmath116 .
if we then project the leaf @xmath400 to @xmath2 we get a holomorphic disc @xmath652 passing through @xmath1 which attaches to @xmath124 along its boundary .
also because of ( [ harmonicity ] ) we have that @xmath122 vanishes on @xmath652 .
it is clear that these discs foliate the blowup of @xmath113 along @xmath1 . to calculate the area of one of these discs
@xmath652 we first observe that @xmath653 now @xmath654 is not necessarily a holomorphic disc in @xmath4 but the boundary we know is a circle @xmath655 and so the symplectic area is the same as for the holomorphic disc @xmath656 .
since @xmath657 is @xmath5-invariant and @xmath658 on the boundary of this disc this means that it has symplectic area @xmath126 .
finally we will show that @xmath659 on @xmath116 .
pick a complex hyperplane in the tangent space at a point @xmath660 not containing the tangents of the holomorphic disc going through @xmath642 .
this is then the projection of a complex subspace of the tangent space of @xmath575 at @xmath648 and we can choose this subspace to lie in the tangent space of @xmath643 .
this subspace will not contain the tangents of the leaf going through @xmath648 and thus @xmath661 is a volume form on it . because of ( [ envelopeequal ] ) this implies that @xmath659 .
let @xmath133 be a smooth strictly plurisubharmonic function on the unit ball @xmath134 in @xmath135 . for small @xmath126 ( i.e. @xmath138 )
the envelope @xmath136 has optimal regularity ( here @xmath665 and the holomorphic discs will pass through the origin ) .
also , for any bounded holomorphic function @xmath139 on @xmath666 we have that @xmath667 where the @xmath668 denotes the monge - ampere measure
@xmath669 we get from the envelope @xmath114 to the kind considered earlier by taking @xmath670 .
it then follows from theorem [ optimalregthm2 ] that @xmath114 has optimal regularity as long as @xmath126 is small .
now let @xmath139 be a bounded holomorphic function on @xmath671 .
also let @xmath672 be the canonical tubular neighbourhood of @xmath242 in @xmath673 thus @xmath8 is an @xmath5-invariant neighbourhood of the origin in @xmath674 and @xmath675 is khler and invariant .
we also note that @xmath676 and so this is also invariant .
now pulling back the integral we get that @xmath677 since the map @xmath678 extends to a holomorphic disc in @xmath134 passing through the origin at its centre it follows from the mean value property that @xmath679 at which point we are done .
r berman and j - p demailly _ regularity of plurisubharmonic upper envelopes in big cohomology classes _ perspectives in analysis , geometry , and topology , 3966 .
, 296 , birkhuser / springer , new york , 2012 .
julius ross , department of pure mathematics and mathematical statistics , university of cambridge , uk . + [email protected] + david witt nystrm , department of pure mathematics and mathematical statistics , university of cambridge , uk .
+ [email protected] , [email protected] | we prove the existence of canonical tubular neighbourhoods around complex submanifolds of khler manifolds that are adapted to both the holomorphic and symplectic structure .
this is done by solving the complex homogeneous monge - ampre equation on the deformation to the normal cone of the submanifold .
we use this to establish local regularity for global weak solutions , giving local smoothness to the ( weak ) geodesic rays in the space of ( weak ) khler potentials associated to a given complex submanifold .
we also use it to get an optimal regularity result for naturally defined plurisubharmonic envelopes and for the boundaries of their associated equilibrium sets . _ dedicated to findus on his birthday _ |
multilayer neural networks ( mln ) are more powerful devices for information processing than the single - layer perceptron because of the possibility of _ different _ activation patterns , so - called internal representations ( ir ) , at the hidden units for the _ same _ input - output mapping .
it is well known that the correlations between the activities at the hidden units are crucial for the understanding of the storage and generalization properties of a mln @xcite . a particular simple situation to study these correlations
is the implementation of random input - output mappings by the network , the so - called storage problem , near the storage capacity . using the replica trick and assuming replica symmetry the correlation coefficients building up in this case
were calculated in @xcite and shown to be characteristic for the prewired boolean function between hidden layer and output .
conversely , _ prescribing _
these correlations the storage properties of the networks change @xcite . the assumption of replica symmetry ( rs ) in this calculation is somewhat doubtful .
in fact it is well known that the storage capacity of mln is strongly modified by replica symmetry breaking ( rsb ) @xcite , which is due to the very possibility of different internal representations . moreover
, even the distribution of the output field of a simple perceptron is influenced by rsb effects @xcite . in the present paper
we elucidate the impact of rsb on the correlation coefficients between the activity of different hidden units in mln with one hidden layer and nonoverlapping receptive fields .
the central quantity of interest is the joint probability distribution for the local fields at the hidden units . in the general part of this paper
we show how this distribution can be calculated both in rs and in one - step rsb . for a detailed analysis
we than specialize to mln with @xmath0 hidden units and discuss , in particular , the parity , committee and and machines . together with the corrections from one - step rsb
the rs results give insight in the division of labor between different subperceptrons in mln and the role of rsb .
calculating finally the correlation coefficients we find that although modifying the local field distribution markedly rsb gives rise to minor corrections to the correlation coefficients only .
we consider feed - forward neural networks with @xmath1 inputs @xmath2 , one hidden layer of @xmath3 units @xmath4 and a single output @xmath5 .
the hidden units have nonoverlapping receptive fields of dimension @xmath6 ( tree structure ) .
they are determined by the inputs via spherical coupling vectors @xmath7 according to @xmath8 with @xmath9 denoting the local fields .
we call an activation pattern @xmath10 of the hidden units an internal representation ( ir ) .
the output @xmath5 of the mln is a fixed boolean function @xmath11 of the ir .
examples of special interest include the parity machine , @xmath12 , the committee machine , @xmath13 , and the and machine , @xmath14 .
all ir consistent with a desired output are called legal internal representations ( lir ) .
the number of and similarity between lir to a given output specifies the division of labor taking place between the different perceptrons forming the mln .
it is quantitatively characterized by the correlation coefficients @xmath15 @xmath16 , where @xmath17 denotes the average over the inputs and the output and @xmath18 is a subset of @xmath19 natural numbers between @xmath20 and @xmath3 . for permutation symmetric boolean functions ,
the @xmath21 only depend on @xmath19 and not on the particular choice of this subset .
we focus on the so - called storage problem in which the inputs @xmath2 and the outputs @xmath22 are generated independently at random according to the probability distributions @xmath23 and @xmath24 where @xmath25 , @xmath26 and @xmath27 .
the basic quantity which gives us access to the probability of the lir and to the correlation coefficients is the distribution @xmath28 of the local fields @xmath29 at the @xmath30th hidden unit .
it is given by @xmath31 @xmath32 denotes the average over all stored input - output patterns .
@xmath33 denotes the partition function @xmath34 @xmath35 the measure on the gardner sphere @xcite @xmath36 and @xmath37 the integration measure @xmath38 we use the replica trick @xmath39 in eq .
( [ p_h ] ) to perform the average over the inputs @xmath40 and introduce the overlaps @xmath41 between different replicas @xmath42 of a coupling vector @xmath43 of hidden unit @xmath44 .
we will consider only permutation symmetric booleans @xmath45 .
hence all hidden units have the same statistical properties implying @xmath46 and @xmath47 with @xmath48 .
equation ( [ p_h ] ) takes on the form @xmath49 in terms of the @xmath50-dimensional order parameter matrix @xmath51 where @xmath52 and @xmath53 .
here @xmath54 -\sum_{k , a < b}x_k^{a}x_k^{b}q^{ab}\right ) \nonumber\\ & \times&\prod\limits_{a}\theta(\sigma f(\{{\text{sgn}}(\lambda_k^{a})\}))\delta(h_j-\lambda_1^{1 } ) \label{ph_central},\end{aligned}\ ] ] and the expression for @xmath55 is specified in the appendix , eq .
( [ g1 ] ) , together with some more details of the calculation . in the limit
@xmath56 the integral ( [ ph_1 ] ) is dominated by the saddle point values of the order parameters @xmath57 which extremize the partition function @xmath58 in the following , we simplify eqs .
( [ ph_central ] ) and ( [ sad_cond ] ) using the assumption that the order parameter matrix @xmath51 is either replica symmetric or describes one - step replica symmetry breaking .
we will always consider the saturation limit @xmath59 since the expressions then simplify and the correlations become most characteristic in this limit .
the rs case is specified by @xcite @xmath60 the saturation limit @xmath61 is characterized by the existence of a unique solution @xmath43 , e.g. , @xmath62 .
we then get @xmath63 for the conditional probability to find a specific value @xmath64 of the postsynaptical potential under the constraint of a given output @xmath5 .
the terms abbreviated by @xmath65 ensure that only lir for the respective value of @xmath5 contribute to the sum in eq .
( [ ph_rs ] ) . as usual
we have used the error function @xmath66 with @xmath67 .
let us now turn to main features of the solution within the ansatz of one - step rsb .
then the following form for the order parameter matrix is assumed @xcite : @xmath68 accordingly there are two overlap scales characterizing the similarity between coupling vectors belonging to the same and different regions of the solution space , respectively . using this ansatz we find after standard manipulations @xcite for the probability distribution of the local field for a specific output @xmath5 @xmath69 where now @xmath70 @xmath71 these expressions simplify in the saturation limit @xmath72 in which one finds @xmath73 and @xmath74 .
the remaining order parameters @xmath75 are given by the saddle point equations corresponding to the following expression for the storage capacity @xmath76 : @xmath77 + q_0 w/[1+w(1-q_0 ) ] } { -2\lim\limits_{q_1\rightarrow 1 } \left\langle\!\!\left\langle{\displaystyle \int}\prod\limits_{k}dy_k\ln\left\ { { \displaystyle \int}\prod\limits_{k}dz_k\left(\phi_{\rm lir}(\sigma)\right)^m\right\ } \right\rangle\!\!\right\rangle_{\sigma } } \right]\quad .
\label{saddle2}\end{aligned}\ ] ] as in the rs case the analytical and numerical analysis of these expressions for concrete situations needs some care ( see next section ) . to finally obtain @xmath78
we must average eqs .
( [ ph_rs ] ) and ( [ ph_rsb ] ) over the two possible outputs @xmath79 , @xmath80 from this probability distribution we find the distributions @xmath81 of the lir according to @xmath82 the correlation coefficients @xmath21 , @xmath16 , are then given by @xmath83 the kronecker @xmath84 in eq .
( [ corr1 ] ) restricts the sum to all lir of the output @xmath5 .
equation ( [ corr1 ] ) is valid as long as the pattern load of the mln does not exceed its saturation threshold @xmath76 .
in this section we apply the general formalism developed above to the analysis of simple versions of three popular examples of mln , namely , the parity , committee and and machines , each with @xmath0 hidden units .
we start with the rs results . in committee and
parity machines there is for every lir of output @xmath85 an ir with all signs reversed that realizes output @xmath86 . therefore @xmath87 and the final average over @xmath5 in eq .
( [ ph_rs_av ] ) is trivial . analyzing eqs .
( [ lir1_rs ] ) and ( [ lir2_rs ] ) in the limit @xmath62 one realizes that they depend on both the sign and values of all integration variables @xmath88 .
expression ( [ lir1_rs ] ) as well as eq .
( [ lir2_rs ] ) are either equal to one or exponentially small in some or all integration variables .
the quotient of both figuring in eq .
( [ ph_rs ] ) can hence become one , zero , or singular with respect to @xmath89 .
whenever it is one the integral in eq .
( [ ph_rs ] ) gives rise to @xmath90 for @xmath91 .
whenever the quotient is singular a contribution @xmath92 results . keeping track of the different contributions arising in this way we find for the @xmath0 committee machine @xmath93 and for the parity machine @xmath94 note that @xmath78 for the parity machine is an even function due to the additional symmetry of the boolean function @xmath45 for this case . in the and machine the output @xmath85 can be realized by one lir only whereas the output @xmath86 results from all the remaining @xmath95 ir .
hence @xmath96 and @xmath97 differ significantly .
in fact we find for the @xmath0 and machine @xmath98 and @xmath99/2 $ ] .
note that we have introduced two different singular contributions @xmath100 and @xmath101 in eqs .
( [ ph_com_rs ] ) , ( [ ph_par_rs ] ) and eqs .
( [ ph_andplus_rs ] ) , ( [ ph_andmin_rs ] ) .
the reason for this is that the weight of @xmath101 adds to the probability of positive local fields whereas the weight of @xmath100 adds to that of negative local fields .
this distinction will be important later when calculating the correlation coefficients from @xmath78 ( cf .
( [ h1 ] ) ) . the results ( [ ph_com_rs ] ) , ( [ ph_par_rs ] ) and ( [ ph_andplus_rs ] ) , ( [ ph_andmin_rs ] ) are shown as the dashed lines in figs .
[ fig1]-[fig3 ] respectively .
these rs results are in fact very intuitive and can be even quantitatively understood by assuming that the outcome of a gardner calculation corresponds to the result of a learning process in which the initially wrong ir are eliminated with least adjustment @xcite . due to the permutation symmetry between the hidden units we may consider only the local field @xmath102 of the first unit of the hidden layer . before learning the couplings @xmath43 are uncorrelated with the patterns and the local field @xmath102 is consequently gaussian distributed with zero mean and unit variance .
now consider , e.g. , the parity machine . due to
the discussed symmetries it is sufficient to analyze the case @xmath85 and @xmath103 .
if @xmath104 and @xmath105 are equal in sign , which will occur with probability @xmath106 , there is no need to modify the couplings at all .
this gives rise to the first term in eq .
( [ ph_par_rs ] ) which is just the original gaussian and describes the chance that a randomly found ir with @xmath103 is legal .
if @xmath104 and @xmath105 differ in sign the ir is illegal and the couplings @xmath43 have to be modified until one of the hidden units changes sign . in an optimal learning scenario the local field with the smallest magnitude
would be selected and the corresponding coupling vector would be modified such that the field just barely changes sign .
hence @xmath102 remains still unmodified if either @xmath104 or @xmath105 is smaller in absolute value which gives rise to the second term in eq .
( [ ph_par_rs ] ) .
finally , if really @xmath102 is selected for the sign change , which will happen with probability @xmath107 for symmetry reasons , it will after learning be either slightly smaller or slightly larger than zero , which is the origin of the last two terms in eq .
( [ ph_par_rs ] ) . with a similar reasoning it is possible to rederive the rs result for the committee machine .
again it is sufficient to consider the case @xmath85 . if @xmath103 initially it will not be modified , which gives rise to the last term in eq .
( [ ph_com_rs ] ) .
if , on the other hand , @xmath108 , prior to learning it will not be modified only if both @xmath104 and @xmath105 are either positive from the start or easier to make positive than @xmath102 .
hence a negative @xmath102 survives the learning process if the other two fields are both larger .
this is described by the first term in eq .
( [ ph_com_rs ] ) .
finally , with probability @xmath109 we find that @xmath108 and either @xmath104 or @xmath105 is even smaller than @xmath102 and therefore harder to correct . in this case
the learning would shift @xmath102 to positive values as described by the second term in eq .
( [ ph_com_rs ] ) .
the resulting distribution of local fields will hence have a dip for negative values of small absolute value clearly visible in fig .
[ fig1 ] .
the case of the and machine is the simplest .
the output @xmath85 requires all local fields to be positive .
hence positive fields are not modified , negative ones are shifted to @xmath110 resulting immediately in eq .
( [ ph_andplus_rs ] ) which is , of course , identical to the result for the single - layer perceptron @xcite . in the case of a negative output @xmath86
only the ir @xmath111 is illegal and must be eliminated which is again done by changing the sign of the smallest field .
this gives rise to eq .
( [ ph_andmin_rs ] ) .
it is finally interesting to compare the distribution of local fields found above with that for a single perceptron above saturation @xcite .
the individual perceptrons in a mln certainly operate above their storage limit even when the storage capacity of the mln is not yet reached .
the most remarkable feature of the distribution of local fields for a perceptron above saturation minimizing the number of misclassified inputs is a _
gap _ separating positive from negative values .
being intimately related to the failure of any finite level of rsb for this problem this gap is believed to exist even in the solution with continuous rsb @xcite . on the other hand , none of the distributions for mln showed a gap .
as should be clear from the above qualitative discussion the reason for this is quite simple .
the single perceptron above saturation has to reject some inputs as not correctly classifiable . in order to keep the number of these errors smallest
it chooses those with negative fields of large absolute value .
inputs with initially only slightly negative local fields will be learned whereby their local fields shift to values just above zero . in this way
the gap occurs .
in mln , on the other hand , there is no reason to shift all negative local fields of small absolute value because the correct output may be realized by the other hidden units
. therefore one will not find an interval of @xmath64 values for which @xmath78 is strictly zero . on the other hand , the tendency that predominantly fields of small absolute value will be modified in the learning process
is clearly shown by the dips of the distribution functions around @xmath112 ( cf . figs .
[ fig1]-[fig3 ] ) .
let us now discuss how the above results get modified by rsb .
the analytical and subsequent numerical analysis of eqs .
( [ ph_rsb]-[saddle2 ] ) for the @xmath0 machines under consideration needs some care in order not to miss the various singular contributions .
we have first to determine the values of the order parameters at the saddle point using eq .
( [ saddle2 ] ) . in the saturation limit @xmath113 , @xmath114
is dominated by one specific lir which is selected among all other lir by the sign and absolute value of the compound variables @xmath115 .
@xmath114 either tends to 1 or becomes exponentially small in one or more compound variables @xmath116 .
transforming the integration from @xmath117 space to @xmath116 space allows us to reduce the @xmath3-fold @xmath118 integral to a one - dimensional integral .
this is performed numerically by rhomberg integration whereas the outer @xmath88 integrals are done using gauss - legendre quadrature @xcite .
the saddle point equation ( [ saddle2 ] ) is solved with a standard minimization routine ( powells method in two dimensions @xcite ) .
the values we get for the order parameters and for the storage capacity are consistent with those obtained earlier . for the @xmath0 parity machine we find @xmath119 , @xmath120 , and @xmath121 in agreement with@xcite . in the case of the @xmath0 committee machine
we get @xmath122 , @xmath123 , and @xmath124 , a result somewhat larger than reported previously @xcite .
the @xmath0 and machine finally does not show rsb at all and we find accordingly @xmath125 , @xmath126 together with @xmath127 . in a second step
, we use this values of the order parameters @xmath75 to calculate the respective distribution of local fields ( [ ph_rsb ] ) .
the distribution functions @xmath78 obtained in this way are included as full lines in figs .
[ fig1]-[fig3 ] .
table [ tab1 ] quantifies the main changes .
the main modification of the distribution functions of local fields that occurs in one - step rsb is a redistribution of probability from the @xmath84 peaks at @xmath128 to the continuous part of the distribution around zero resulting in a reduction of the weight of the singular parts of roughly 50% .
this gives rise to a less pronounced dip of the distribution functions around @xmath112 and is qualitatively similar to the rsb modifications for a single perceptron above saturation @xcite . from the results for the parity machine
it is conceivable that the central peak may get reduced further if higher orders of rsb are included and that it might eventually disappear completely in the full parisi solution using continuous rsb .
for all machines the probability of fields with large absolute values is hardly affected by rsb . for the and machine we did not find rsb at all .
the numerical solution of the saddle point equations only gave the rsb result @xmath129 , @xmath130 .
we therefore suspect that replica symmetry is correct for the and machine .
this is also in accordance with the rule of thumb that rsb is necessary if the solution space is disconnected . in the and machine the output @xmath85
can be realized only by one lir which clearly corresponds to a connected ( even convex ) solution space .
the output @xmath86 is realized by all remaining ir , which as the complement of the previous solution space must be connected too .
we have finally to clarify how much the modifications found for the distributions of local fields will change the probabilities of the internal representations and the correlation coefficients @xmath21 depending only on the _ sign _ of the local fields .
this question is , in fact , nontrivial only in the case of the committee machine . for
the and machine no rsb occurs at all and for the parity machine the correlation coefficients are completely determined by the symmetry of the boolean function @xmath45 between hidden units and output . for the committee machine
we find that the probability of the lir @xmath111 is shifted from its rs value @xmath131 to @xmath132 , which is an increase by roughly 13% whereas the probability of the three remaining lir ( consisting of two pluses and one minus each ) is reduced by 1.9% from 0.2917 to 0.2861 .
qualitatively this means that more inputs are stored with the lir @xmath111 than the fraction @xmath133 that had this lir already by chance before learning .
the learning process hence does not shift illegal ir just up to the decision boundary of the boolean @xmath45 but in some cases the correlations between inputs @xmath134 and couplings @xmath135 neglected in rs allow even the safer lir @xmath111 .
( [ corr1 ] ) we can now also calculate the correlation coefficients and find that @xmath136 increases by 2.7% from its rs value 5/12 , @xmath137 decreases in absolute value by 13.3% from its rs value 1/6 and @xmath138 decreases in absolute value by 4.5% from its rs value 3/4 .
this confirms the prediction of @xcite that although crucial for the storage capacity rsb will have only a minor influence on the correlation coefficients in mln .
generalizing the calculation of the distribution function of local fields for the single - layer perceptron we introduced a general formalism to determine the joint probability distribution @xmath139 of local fields at the @xmath3 hidden units of a two - layer neural network of tree architecture with fixed boolean function between hidden layer and output both in replica symmetry and in one - step replica symmetry breaking .
explicit results were obtained for the parity , committee , and and machine with @xmath0 hidden units in the saturation limit @xmath140 .
although the individual perceptrons are by far overloaded there is no gap in the distribution of local fields as known from a single perceptron above saturation .
there is no rsb for the and machine which we attribute to the connected solution space for this architecture . for the parity and committee machine
we find as a result of rsb a slight redistribution of probability from the singular parts at @xmath128 to the continuous part around the origin .
the correlation coefficients @xmath21 characterizing the correlations between the legal internal representations are not modified by rsb for the parity machine since in this case they are fixed already by symmetries . for the committee machine
the changes of the correlation coefficients are rather small and the rs results derived in @xcite may serve as useful approximations .
in this appendix we give some more details on the calculation of the distribution function @xmath78 of the local fields at the hidden units following gardner s approach @xcite . introducing the replica trick @xmath39 into eq .
( [ p_h ] ) yields @xmath141 with replica index @xmath142 . in the integration measures ( [ measure_j ] ) , ( [ measure_fields ] ) we replace the @xmath84 functions by their integral form
@xmath143 @xmath144 we now perform the average over the gaussian distributed patterns @xmath145 and introduce the overlaps @xmath146 of different replicas of the same perceptron @xmath43 as well as its conjugated variable @xmath147 . from the assumed permutation symmetry of the boolean function @xmath45 with respect to all hidden units we infer @xmath148 , @xmath149 , and @xmath150 for all @xmath25 .
this gives rise to the form @xmath151 where @xmath51 and @xmath152 denote the symmetric matrices @xmath52 , @xmath153 and @xmath154 , @xmath155 . moreover , @xmath156 -\sum_{k , a < b}x_k^{a}x_k^{b}q^{ab}\right ) \nonumber\\ & \times&\prod\limits_{a}\theta(\sigma f(\{{\text{sgn}}(\lambda_k^{a})\}))\delta(h-\lambda_1^{1}),\end{aligned}\ ] ] @xmath157 -\sum_{k , a < b}x_k^{a}x_k^{b}q^{ab}\right ) \nonumber\\ & \times&\prod\limits_{a}\theta(\sigma f(\{{\text{sgn}}(\lambda_k^{a})\ } ) ) , \label{g1}\\ g_2(a)&=&\int\!\prod\limits_{k , a}\frac{dj_k^a}{\sqrt{2\pi e } } \exp\left(-\frac{1}{2}\sum\limits_{k;a , b } j_k^aa^{ab}j_k^b\right).\end{aligned}\ ] ] in the limit @xmath56 the integral in eq . ( [ ph_replic2 ] )
is dominated by the saddle point values of the order parameters @xmath158 , @xmath159 , and @xmath57 . solving the saddle point equation with respect to @xmath160 and @xmath159 yields @xmath161 .
( [ ph_replic2 ] ) takes the form @xmath162 @xmath163 can be calculated by assuming either rs or one - step rsb for the matrix @xmath51 resulting in eqs .
( [ ph_rs ] ) and ( [ ph_rsb ] ) , respectively .
the remaining saddle point condition for the matrix @xmath57 has in one - step rsb ( [ rsb ] ) the form @xmath164 } + \frac{1}{2m}\ln\left(1+\frac{m(q_1-q_0)}{1-q_1}\right ) + \frac{1}{2}\ln(1-q_1)\right .
\nonumber\\ & + & \left.\frac{\alpha}{m } \left\langle\!\!\left\langle\int\prod\limits_{k}dy_k\ln\left\ { \int\prod\limits_{k}dz_k\left(\phi_{\rm lir}(\sigma)\right)^m\right\ } \right\rangle\!\!\right\rangle_{\sigma } \right ] .
\label{saddle1}\end{aligned}\ ] ] it determines a set of order parameters @xmath165 for every pattern load below the storage capacity @xmath166 . the abbreviation @xmath167 is defined by eq .
( [ sum_lir_1 ] ) .
the angular brackets @xmath168 indicate the average over the two possible outputs @xmath169 . [
99 ] m. mezard and s. patarnello , lptens report , 1989 ( unpublished ) .
m. griniasty and t. grossman , phys .
a * 45 * , 8924 ( 1992 ) .
a. priel , m. blatt , t. grossman , e. domany and i. kanter , phys .
e * 50 * , 577 ( 1994 ) .
b. schottky , j. phys .
a * 28 * , 4515 ( 1995 ) .
r. monasson and r. zecchina , phys .
lett . * 75 * , 2432 ( 1995 ) .
a. engel , j. phys .
a * 29 * , l323 ( 1996 ) .
d. malzahn , a. engel and i. kanter , phys .
e * 55 * , 7369 ( 1997 ) .
e. barkai , d. hansel and i. kanter , phys .
lett . * 65 * , 2312 ( 1990 ) .
e. barkai , d. hansel and h. sompolinsky , phys .
a * 45 * , 4146 ( 1992 ) .
a. engel , h. m. khler , f. tschepke , h. vollmayr and a. zippelius , phys .
a * 45 * , 7590 ( 1992 ) .
p. majer , a. engel and a. zippelius , j. phys .
a * 26 * , 7405 ( 1993 ) , whyte and sherrington , _ ibid . _
* 29 * , 3063 ( 1996 ) .
g. gyrgyi and p. reimann , phys .
lett . * 79 * , 2746 ( 1997 ) .
e. gardner , j. phys .
a * 21 * , 257 ( 1988 ) , e. gardner , b. derrida , _ ibid . _
* 21 * , 271 ( 1988 ) .
m. mzard , g. parisi and m. virasoro , _ spin glass theory and beyond _
( world scientific , singapore , 1987 ) .
e. gardner , j. phys .
a * 22 * , 1969 ( 1989 ) .
m. opper , phys .
a * 38 * , 3824 ( 1988 ) .
d. j. amit , m. r. evans , h. horner and k. y. wong , j. phys .
a * 23 * , 3361 ( 1990 ) . w. h. press , s. a. teukolsky , w. t. vetterling and b. p. flannery , _ numerical recipes _ ( cambridge university press , cambridge , england , 1992 ) . .
saturated @xmath0 machines : integrated features of the probability distribution @xmath78 of the local field .
corrections by one - step rsb are given in percent of the respective rs value .
dashes indicate that a respective singular contribution does not occur ( committee ) or that we found no rsb ( and ) .
[ cols="<",options="header " , ] | we consider feed - forward neural networks with one hidden layer , tree architecture and a fixed hidden - to - output boolean function .
focusing on the saturation limit of the storage problem the influence of replica symmetry breaking on the distribution of local fields at the hidden units is investigated .
these field distributions determine the probability for finding a specific activation pattern of the hidden units as well as the corresponding correlation coefficients and therefore quantify the division of labor among the hidden units .
we find that although modifying the storage capacity and the distribution of local fields markedly replica symmetry breaking has only a minor effect on the correlation coefficients .
detailed numerical results are provided for the parity , committee and and machines with k=3 hidden units and nonoverlapping receptive fields . |
the bepposax satellite performed a survey of the galactic center region in the 110 kev energy band with its narrow field instruments during 19971998 .
a source positionally coincident with the galactic center ( hereafter gc ) was observed , together with strong diffuse emission and several point like sources with luminosity l@xmath4 erg s@xmath5 .
the results on these sources , most of which are likely low mass x
ray binaries containing neutron stars and black holes , both with transient and persistent emission , are reported in detail by sidoli et al .
the spectral results for these sources are summarized in fig . 1 , where the photon index of the fits with a power law are plotted versus the hydrogen column density
. an upper limit of l@xmath6 erg s@xmath5 has also been placed to the 210 kev luminosity from the x
ray counterpart of sgra * ( see sidoli et al .
1999 for details ) , confirming the underluminosity of this presumed supermassive black hole at high energies .
intense diffuse x ray emission is also present in the gc region , the nature of which is still poorly known . here
we present the bepposax results on the diffuse emission from the sgra complex ( sidoli & mereghetti 1999 ) . -5truecm
the sgr a region has been imaged with the mecs instruments ( 1.310 kev ) with a spatial resolution of @xmath7 ( about 2.5 pc at the gc distance ) in august 1997 ( 99.5 ksec net exposure time ) . in order to study the spectral properties of the diffuse emission coming from the sgra complex , we extracted the mecs counts from four concentric annular regions ( @xmath8,@xmath9,@xmath10,@xmath11 ) around sgra*. several emission lines are present in all the spectra , with the k - lines from iron ( e@xmath12 kev ) and sulfur ( e@xmath13 kev ) particularly bright .
the fit with a single temperature hot plasma model showed a nearly constant temperature ( @xmath14 kev ) at radii larger than 2@xmath1 , while in the innermost circle a softer spectrum ( kt@xmath15 kev ) was found .
this is probably due to the contribution from one ( or more ) point source(s ) located close to sgra * ( predehl & trmper 1994 , maeda et al .
1996 , sidoli et al .
1999 ) , that can not be spatially resolved in our data .
since the temperature profile does not show spectral variations in the region from @xmath16 to @xmath17 , we studied the overall spectrum from this entire corona .
a single temperature plasma ( mekal model ) is not adequate to describe the spectrum , leaving positive residuals at low energy and especially around 6.4 kev .
this can be due to the presence of fluorescent emission from neutral or weakly ionized iron in the nearby molecular clouds ( koyama et al .
thus we added a lower temperature plasma component plus a gaussian line at 6.4 kev .
the resulting best fit is a double temperature plasma , with [email protected] kev and kt@xmath38 kev ( n@xmath18=8@xmath19 @xmath20 ) and a gaussian line at 6.4 kev with an equivalent width of @xmath21120 ev ( see fig . 2 ) .
the total flux corrected for the absorption is f@xmath22 erg @xmath20 s@xmath5 ( 210 kev ) , which translates into a luminosity of l@xmath23 erg s@xmath5 .
about one third of the flux is contributed by the soft component . -3.5truecm
-2truecm -2truecm the spatial distribution of the diffuse emission was studied extracting two images in different energy ranges , below and above 5 kev ( see fig .
both emissions are peaked at the gc position , but they have significantly different spatial distributions : the soft emission ( 25 kev ) displays a triangular shape , while the hard one ( 510 kev ) is elliptical and elongated in the direction of the galactic plane . while the hard emission can be simply part of the diffuse emission permeating the inner 60@xmath24 of the galactic disk ( e.g. kaneda et al .
1997 ) , the soft x
rays are spatially correlated with a structure observed in the radio band , known as the sgr a east triangular halo ( pedlar et al .
this is an extended non thermal structure ( probably a snr ) which surrounds in projection sgra*. since also our spectral data are well described by a two temperature thermal model , it is tempting to give an interpretation in terms of two plasma components with different temperatures and spatial distributions . from the spectral analysis of the lower temperature plasma ( kt@xmath25 kev ) , we derive a luminosity @xmath26 ergs s@xmath5 ( 210 kev ) , an electron density n@xmath27 @xmath28 , a total mass @xmath29 and an average thermal pressure @xmath30 ergs @xmath28 , consistent with the pressure @xmath31 erg @xmath28 derived for a snr in a sedov phase .
the bepposax observation of the sgra complex revealed the presence of at least two distinct components : a soft component with [email protected] kev spatially correlated with the sgra east @xmath32 triangular halo , and a hard one with kt@xmath2179 kev elongated along the galactic plane and possibly associated with the harder component of the galactic ridge emission .
the soft component , which accounts for about one third of the 2 - 10 kev diffuse luminosity from the sgra complex , can be well explained as thermal emission from the snr responsible for the radio halo of the sgra east shell . | the @xmath0sax satellite performed a survey of the galactic center region in the 1 - 10 kev energy band with its narrow field instruments . several bright x
ray sources containing neutron stars and black holes have been observed and studied , including the possible counterpart of sgra*. here we report the results on the diffuse emission coming from the sgra complex .
the emission from within 8@xmath1 from sgra * has a double temperature thermal spectrum ( kt@xmath2 0.6 kev and kt@xmath38 kev ) and an energy
dependent morphology : the hard emission ( 510 kev ) is elongated along the galactic plane , while the soft one ( 25 kev ) shows a triangular shape , very similar to the radio halo of sgra east .
this spatial correlation and the physical parameters of the lower temperature component support the interpretation of the radio halo of the sgr a east shell as a snr . |
for more than 40 years the yakutsk array continuously records ultra - high energy cosmic rays ( uhecr ) by measuring electromagnetic , muonic and cherenkov components of extensive air showers ( eas ) @xcite . through these measurements a lateral distribution of charged particles and muons with @xmath0 gev @xcite
was studied and longitudinal development of air showers was reconstructed by measuring the lateral distribution of cherenkov light emitted in eas @xcite .
estimation of the depth of maximum shower development ( @xmath1 ) in individual showers with energy @xmath2 ev at the yakutsk array was performed using readings from integral and differential ( tracking ) cherenkov light detectors .
integral detectors provide observation of cherenkov photons lateral distribution in wide distance range , while differential detectors record time - base of the cherenkov light pulse at different core distances .
it follows from model calculations that characteristics such as @xmath3 ( a relation between the density of cherenkov light fluxes at @xmath4 and @xmath5 m from shower axis ) and @xmath6 ( the width of the cherenkov pulse at the half - height ) are sensitive to the depth of maximum shower development .
besides , the @xmath7 parameter depends weakly on conditions of cherenkov light distribution in the atmosphere ( @xmath8% ) , but depends strongly on the accuracy of shower geometry reconstruction ( @xmath9% ) .
the @xmath6 is measured by equipment that provides high precision of the cherenkov pulse shape reproduction and thus assures better accuracy in @xmath1 measurement ( @xmath10 g/@xmath11 ) .
the methodology of cherenkov light measurement and analysis is presented in a greater detail in papers @xcite .
to extract information about the parameters of interaction between the primary particle and air nuclei it is necessary to measure those characteristics of eas which are morphologically connected to the start of nuclear cascade in air .
one can test the sensitivity of eas characteristics to cross - section of inelastic interaction with monte - carlo method to simulate artificial showers using some model of hadronic interactions .
such calculations were performed in multiple works .
further we will consider the methodology and algorithm for estimation of inelastic interaction cross - section involving the data on the depth of maximum eas development ( @xmath1 ) .
total cross - section of secondary particles production in eas is one of fundamental values that characterize hadronic interactions . for showers initiated by primary protons of some energy it is possible to obtain the distribution of the depth of initial interaction ( @xmath12 ) : @xmath13 where @xmath14 is a mean free path of a proton before interaction in air .
mean depth of first interaction with respect to its fluctuations is directly connected to the cross - section of inelastic interaction between proton and air nucleus : @xmath15 here , @xmath16 is mean air `` mass '' , which is @xmath17 mb / g@xmath18 @xcite . the relations ( [ eq1 ] ) and ( [ eq2 ] ) were used to estimate the inelastic interaction cross - section by experimental data obtained at the yakutsk array . during the measurement of the @xmath14 value , a mass composition of primary cosmic rays ( pcr ) was taken into account ( contribution to generation of @xmath1 distribution at fixed energy of other nuclei ) by utilizing the points shifted by one @xmath19 from the maximum of histogram ( see fig.[fig1a ] ) . at fixed energy and the technique of @xmath20 measurement by cherenkov light data adopted at the yakutsk array .
@xmath21 events , @xmath22 ev , @xmath23 g/@xmath11 , @xmath24 g/@xmath11 , @xmath25 g/@xmath11.,scaledwidth=75.0% ]
in present work we employ measured @xmath1 distribution at fixed energy values @xmath26 ev , @xmath27 ev and @xmath28 ev which were obtained from cherenkov light data recorded by the small cherenkov setup during more than ten years of continuous observation ( see fig.[fig1b ] ) .
final results of the analysis are presented in table [ tab1 ] and fig.[fig2 ] .
distributions , from which the @xmath29 was derived.,scaledwidth=75.0% ] .cross - sections of p - air inelastic interaction [ cols="<,<,<,<,<,<",options="header " , ] fig.[fig2 ] shows a compilation of the @xmath30 data obtained in different experiments including the yakutsk array in the energy range @xmath31 ev ( see tien shan data for 1 pev @xcite ) . in this work , using similar algorithm and performing the simulation of @xmath1 measuring process with monte carlo method ( with respect to hardware uncertainties ) , we obtained the @xmath32 for energies @xmath33 ev .
new results are shown on fig.[fig2 ] .
on the same figure results of calculations with various hadron interaction models are shown @xcite .
extrapolation of accelerator data leads to a large uncertainty at energy above @xmath34 ev . to a greater extent this is tied to some discrepancy between models in description of more precise accelerator data and different approaches in description of generation of secondary particles in eas . as it is seen on fig .
[ fig2 ] , cosmic ray data together with accelerator data allow to introduce some limitations on applicability of particular hadron interaction model for description of ultra - high energy air shower development .
the yakutsk eas array is a medium sized array compared to compact and large arrays .
it effectively controls the energy range from @xmath35 ev to @xmath36 ev . in several decades of continuous observations a large amount of experimental data was accumulated .
both irregularities in cosmic ray energy spectrum are manifested in these data and the mass composition of primary particles initiating ultra - high energy air showers has been estimated using one technique .
such simultaneous analysis of the spectrum and mass composition allows one to test various hypotheses on cosmic rays origin and their propagation in the universe . to increase the precision of this analysis
it would take a choice of a single hadron interaction model from the group of models presented today .
studying the energy dependence of inelastic interaction cross - section just helps make such a choice .
for example , the choice of the model for eas development would allow experimenters to improve the precision in energy estimation and obtain the model - independent estimation of cosmic ray mass composition in ultra - high energy domain , more reliable than those adopted nowadays .
the work is supported in part by sb ras ( integral project `` modernization of the yakutsk array '' ) , rfbr ( grant # 110200158 ) and the russian ministry of education and science ( contract # 02.740.11.0248 ) . | we present estimation of proton - air inelastic interaction cross - section obtained for different energy values are .
results are compared with different hadron interaction models . |
1 . correlation between the @xmath0 mean lifetime and production branching ratio @xmath163 , needed to explain the karmen anomaly ; the solid curve is taken from ref .
@xcite and the dashed curve is its extrapolation . the @xmath37 scale is derived from eq.([eq : bf ] ) for @xmath108 .
the @xmath46 scale is derived from eq.([eq : t ] ) for the case of dominant @xmath67 visible decays with dirac @xmath0 .
the most stringent laboratory constraints on the @xmath164 values are also shown @xcite .
for majorana @xmath0 there is an additional constraint @xmath165 and the @xmath46 scale moves left by a factor of 2 .
[ fig : fig1 ] 2 .
visible energy spectra for @xmath18 decays from various sources , weighted by the inverse of the @xmath0 momentum for the relative decay probability in a given detector .
the solid curve denotes the stopped @xmath12 source ; dotted , dashed , and dot - dashed curves denote the @xmath13 , @xmath166 and @xmath167 sources , respectively .
[ fig : fig2 ] in the case that neutrino @xmath0 is mainly isosinglet ( sterile ) , we overlooked neutral - current contributions to the decay @xmath168 , as pointed out by j. peltoniemi ( hep - ph/9606228 ) .
the corrected width for small mixing is @xmath169\;{\sec}^{-1},\ ] ] where @xmath170 , @xmath171 for dirac ( majorana ) @xmath0 and the @xmath46 term includes charged - current contributions .
the visible decay width ( including @xmath23 ) and total width then become @xmath172\;{\sec}^{-1},\\ \gamma_{\total } & = & k \left[2470 |u_{ex}|^2 + 1760 |u_{\mu x}|^2 + 1760 |u_{\tau x}|^2\right]\;{\sec}^{-1}. \\ \end{array}\ ] ] the karmen event rate determines the product @xmath173 with @xmath174 as before , which allows different types of solution : \(a ) if @xmath46 dominates @xmath175 , we obtain @xmath176 . for majorana @xmath0 ,
no such solutions are compatible both with the constraint eq.(9 ) and with @xmath46 dominance of @xmath175 . for dirac @xmath0
there are acceptable solutions near @xmath177 ; but as @xmath142 decreases , @xmath62 increases and they merge into category ( b ) .
\(c ) if @xmath181 dominates @xmath175 , we obtain @xmath182 , with @xmath183 , where the upper ( unitarity ) limit can not be approached closely because @xmath0 can not be the main component of @xmath73 . solutions ( a ) and ( b ) give mean lifetimes @xmath184 , but ( c ) covers the range @xmath185 .
the neutral - current contributions now guarantee that @xmath186 is always the dominant visible mode .
eqs.(16)-(17 ) are invalidated and some parameters change , but our qualitative conclusions about the existence of solutions ( especially short - lived type-(c ) cases ) remain broadly unchanged .
concerning additional sources of @xmath0 at karmen , the @xmath187 and @xmath188 channels depend on the ratio @xmath189 as before ; this ratio is @xmath190 for ( a ) and ( b ) but can be large for type-(c ) solutions .
the @xmath191 rate is fixed and unchanged . | a recently reported anomaly in the time structure of signals in the karmen neutrino detector suggests the decay of a new particle @xmath0 , produced in @xmath1 with mass @xmath2 mev .
we discuss the constraints and difficulties in interpreting @xmath0 as a neutrino .
we show that a mainly - sterile neutrino scenario is compatible with all laboratory constraints , within narrow limits on the mixing parameters , although there are problems with astrophysical and cosmological constraints .
this scenario predicts that appreciable numbers of other @xmath0-decay events with different origins and time structures should also be observable in the karmen detector .
such @xmath0-decay events should also be found in the lsnd experiment and may be relevant to the search for @xmath3 oscillations . 6.25 in 9.75 in -.5 in = = # 1#20=1=to0@xmath40=1=to0#2#1 - 01 # 1/#2 l
@=11 tempcntc citex[#1]#2@fileswauxout tempcnta@tempcntb@neciteaciteforciteb:=#2citeo#1 citeotempcnta > tempcntbciteacitea , tempcnta = tempcntbtempcnta @=12 = cmssbx10 scaled 2 to remarks on the karmen anomaly + v. barger@xmath5 , r.j.n .
phillips@xmath6 and s. sarkar@xmath7 + _ @xmath5physics department , university of wisconsin , madison , wi 53706 , usa + @xmath6rutherford appleton laboratory , chilton , didcot , oxon ox11 0qx , uk + @xmath7theoretical physics , university of oxford , 1 keble road , oxford ox1 3np , uk _
( published in _ phys .
lett .
b352 ( 1995 ) 365 _ ; erratum added june 1995 ) the karmen collaboration , which studies the interactions of neutrinos from the stopped @xmath8 decay chain at ral , has recently reported an anomaly @xcite in the time - dependence of their signals .
this anomaly suggests the production of a new weakly - interacting neutral particle ( call it @xmath0 ) in the initial @xmath8 decays , which travels with well determined velocity @xmath9 and decays in the detector after a mean flight path of 17.5 m. the distinctive feature of the @xmath0-events is their timing , apparently at a well determined interval @xmath10s after the arrival and prompt decay of the pion pulse ( which determines @xmath11 ) ; however , the visible energy in the detector scintillator shows no anomaly , so @xmath0 decays apparently deposit visible energy similar to typical neutrino interactions .
the present note briefly discusses the interpretation of @xmath0 as a massive neutrino .
we show that a mainly - sterile neutrino scenario is compatible with all laboratory constraints , for either dirac or majorana options , within rather narrow bounds on the mixing parameters .
there are some problems with astrophysical and cosmological constraints , but it is interesting nevertheless to explore the further implications for laboratory experiments which can test this interpretation directly .
this scenario predicts that appreciable numbers of other @xmath0-decay events with different origins and time structures should also be observable in the karmen detector .
such @xmath0-decay events should also be found in the lsnd experiment at lampf @xcite and may be relevant to the ongoing search for @xmath3 oscillations @xcite .
if we postulate no other new particles below the pion mass , then the precise time structure @xcite requires @xmath0 production to go via one of the two - body modes @xmath12 or @xmath13 .
however the latter implies mass @xmath14 mev ( determined from @xmath11 ) and hence anomalously large visible @xmath0-decay energy , with mean value @xmath15 mev for @xmath16 or @xmath17 mev for @xmath18 , compared to neutrino interactions that typically give @xmath19 mev @xcite .
we must therefore presume that @xmath0 is produced via @xmath20 , with @xmath2 mev determined from @xmath11 .
assuming no new weak interactions , the standard decay mode is @xmath21 ; note that flavour - changing neutral current ( fcnc ) processes @xmath22 and @xmath23 are highly suppressed for a standard isodoublet neutrino ( we consider the isosinglet case later ) @xcite .
the neutrino charged - current eigenstates @xmath24 ( @xmath25 ) which appear in the weak interaction coupled via @xmath26 to @xmath27 , may be written as coherent superpositions of mass eigenstates @xmath28 ( @xmath29 ) using the usual mixing matrix @xmath30 ; @xmath31 ( in general @xmath32 is a @xmath33 matrix , with @xmath34 running over all @xmath35 multiplet assignments and @xmath36 running over all masses ) . then the @xmath0 production and decay processes are scaled by factors @xmath37 and @xmath38 , respectively : @xmath39 \lambda^{1\over 2}(m_\pi^2 , m_\mu^2 , m_x^2 ) }
{ m_\mu^2(m_\pi^2 - m_\mu^2)^2},\\ \frac{\gamma ( x\to e^+e^-\nu_e)}{\gamma ( \mu\to\nu_\mu\bar\nu_e e)}&= & \frac{|u_{e x}|^2 m_x^5}{m_\mu^5}\;\ { \dirac}\ , \quad \frac{2|u_{e x}|^2 m_x^5}{m_\mu^5}\;\ { \majorana}\ , \end{aligned}\ ] ] where @xmath40 and we neglect @xmath41 ; we recall that majorana neutrinos decay twice as fast as dirac neutrinos with the same coupling , because their right chiral components are not inert .
hence the production branching fraction and mean decay lifetime are simply given by @xmath42 @xmath43 the correlation between branching fraction and lifetime , needed to explain the karmen anomaly , was shown as a curve in the ( @xmath44 ) plane in ref .
@xcite .
we reproduce this as the solid curve in fig.1 , extrapolating along the dashed curve with fixed @xmath45 , and showing also the scales of @xmath46 and @xmath37 implied by eqs.([eq : bf])-([eq : t ] ) along the upper and right - hand edges of the diagram ( the dirac option is illustrated for @xmath46 ) .
the regions with @xmath47 have no physical meaning in our scenario . direct experimental constraints on the mixing elements
@xmath48 are summarized in ref .
@xcite ; they generally depend on mass and the constraints we quote below are all for @xmath2 mev .
absence of a correction to the @xmath49 parameter of the @xmath50 spectrum in @xmath51 decay gives @xcite @xmath52 absence of decay events in neutrino beams gives @xcite @xmath53 absence of anomalous contributions to @xmath54 gives @xcite @xmath55 limits from neutrinoless @xmath56-decay searches @xcite on the effective @xmath57 majorana mass + @xmath58 , would require ( see also ref .
@xcite ) @xmath59 where we have conservatively taken @xmath60 ev @xcite .
this bound would apply if @xmath0 were a majorana state and there were no substantial cancellations in the sum but not if @xmath0 were a dirac neutrino or part of a quasi - dirac pair ( with opposite cp signatures ) .
other direct laboratory constraints are weaker than these @xcite . studies of short muon tracks in @xmath61 events , from pions stopping in emulsion , would give stringent constraints on @xmath62 for @xmath63 mev @xcite ; but for the present value @xmath2 mev , the muon kinetic energy is only 1.5 kev giving an unobservable track length less than 1 micron , so no constraint can be derived on this basis .
the @xmath0 mass and mixing predict a contribution to the @xmath64 branching fraction @xcite @xmath65 a factor @xmath66 below the experimental upper limit @xcite .
the constraints of eqs.([eq : shrock])-([eq : delee ] ) are shown on fig.1 ; they leave a range of solutions " to the karmen anomaly , based on @xmath67 decay , described by @xmath68 there are also constraints on the mass and identification of @xmath0 .
the argus bound @xmath69 mev @xcite , the cleo bound @xmath70 mev @xcite and the recent aleph bound @xmath71 mev @xcite ( all at @xmath72 c.l . ) , exclude @xmath0 from being the major component of @xmath73 .
since lep experiments measure @xmath74 @xcite light neutrino species ( weighted by their isodoublet mixing factors ) , the chiral component @xmath75 participating in standard weak interactions must then be dominantly isosinglet , i.e. sterile .
an isodoublet interpretation of @xmath0 is also excluded by cosmological and astrophysical arguments concerning unstable neutrinos in the mass and lifetime range of interest @xcite .
if @xmath0 has standard weak interactions , its cosmological relic abundance would be sufficiently high that its decay products would have distorted the spectrum of the 2.73 k blackbody radiation background unless its lifetime were less than @xmath76 s @xcite .
this requires @xmath77 ( for @xmath2 mev ) and removes part of the solution range in eq.([eq : sol ] ) . at such early times
the background photons are energetic enough to be compton scattered by the @xmath78 pairs from @xmath0 decay to energies above the @xmath79h photofission threshold and may thus undo nucleosynthesis @xcite . taking into account the energy degradation due to @xmath80 scattering @xcite this
sets a lifetime bound of @xmath81 s @xcite corresponding to @xmath82 , which leaves only a tiny region of the remaining solution range .
( further constraints on isodoublets from consideration of the entropy production by the decaying particle are rather sensitive to the adopted upper limit to the primordial @xmath83he abundance @xcite . )
this loophole is closed by consideration of the production and decays of massive neutrinos in supernova 1987a .
for example the process @xmath84 operates at a rate @xmath85 relative to the decay @xmath86 and would have generated a @xmath87 ray burst which was not observed by the smm satellite @xcite . when combined with other arguments relating to energy deposition inside the supernova
@xcite , as well as experimental bounds on fast @xmath88 decays @xcite , this rules out _ all _ lifetimes shorter than @xmath89 s @xcite , forbidding the entire solution range for doublet neutrinos . in order to evade the above bounds
, it may appear adequate to require @xmath0 to be mainly sterile since its direct production , both in supernovae and in the early universe , is then suppressed .
however , sterile neutrinos can also be produced through their mixing with doublet neutrinos , modulated by matter effects . during the collapse phase of a supernova
, resonant @xmath90 conversions may cause rapid deleptonisation of the core which would probably prevent the supernova explosion @xcite ; however this does not happen for @xmath91 ev@xmath79 so will not apply to the @xmath0 particle .
secondly the energy loss due to emission of sterile neutrinos during the cooling phase would have excessively shortened the @xmath92 burst from sn 1987a for a mixing in the range @xmath93 @xcite ; this argument applies to a neutrino with mass up to @xmath94 mev @xcite ( see also ref .
@xcite ) and will rule out the allowed region for @xmath0 .
most crucially , the absence of a @xmath87 ray burst from sn 1987a would , as in the case of doublet neutrinos discussed above , rule out radiative decays @xcite with lifetimes in the range @xmath95 s , taking into account the enhancement of radiative decays for a singlet neutrino ( see below ) .
the production of sterile neutrinos through neutrino oscillations in the early universe @xcite also presents a problem since @xmath0 particles can be brought into thermal equilibrium at temperatures exceeding a few gev .
although the @xmath0 relic abundance is thus substantially diluted by the entropy release during the subsequent quark - hadron phase transition , @xmath0 decays with a lifetime @xmath96 s will have an adverse effect on primordial nucleosynthesis @xcite ( see also ref .
@xcite ) .
thus the interpretation of the karmen anomaly as a singlet neutrino is severely challenged by astrophysical and cosmological arguments , although there may be a loophole for fast decays with lifetime @xmath97 s which occur in the supernova core and , in the early universe , before nucleosynthesis .
therefore we proceed to examine the implications of this hypothesis for laboratory experiments which can test it definitively . identifying @xmath0
as mainly sterile has several distinct repercussions for the @xmath0-decay modes , since gim cancellations no longer suppress fcnc : \(i ) the @xmath22 invisible modes are now appreciable ; we obtain @xmath98/|u_{ex}|^2 , \label{eq : nununu}\ ] ] assuming that just one singlet neutrino flavour takes part and the other 3 neutrinos are much lighter than @xmath0 ; thus the invisible branching fraction is never less than @xmath99 .
but our determination of @xmath46 from the karmen @xmath45 plot is preserved , because the latter is determined by the visible decays only : @xmath100 .
adding invisible decay modes increases the total @xmath0-width and hence the total number of decays in a given detector , but the visible fraction decreases by the same factor and the number of visible decays remains unchanged ( so long as the distance to the detector remains much less than the mean decay length , which is the case here ) .
\(ii ) radiative decays @xmath23 , going via loop diagrams , also escape gim suppression for mainly - singlet @xmath0 ; we obtain @xcite @xmath101 assuming one singlet and summing 3 light final flavours as before .
radiative decays are thus between @xmath102 and @xmath103 of total decays , but can dominate visible decays as follows : @xmath104 @xmath105
\(iii ) when @xmath23 modes start to dominate over @xmath106 in the visible decays , we can no longer compensate an increase in @xmath0-production ( increase in @xmath37 ) by a decrease in @xmath46 , so we have to add eq.([eq : eendom ] ) to the solution constraints for a sterile neutrino .
this leaves a reasonable range for sterile dirac but only a small region for sterile majorana solutions in fig.1 .
\(iv ) a new class of solution is possible , in which @xmath23 dominates the visible decays , in contrast to the solutions displayed in fig.1 where @xmath67 dominates .
these solutions have @xmath107 with @xmath108 determined by eq.([eq : bf ] ) as before , but @xmath46 now contributes negligibly to the decay and we have @xmath109 for dirac @xmath0 ( double this for majorana @xmath0 ) .
hence these solutions are characterized by @xmath110 which automatically satisfies the requirement @xmath111 from eq.([eq : shrock ] ) .
they are also constrained by the preliminary bound on @xmath112 decays @xcite from the bebc - wa66 beam dump experiment @xcite , which here translates into @xmath113 the @xmath0 mean lifetime , dominated by @xmath22 decays , can range between about @xmath114 s ( at the limit where @xmath115 ) to about @xmath116 s ( where @xmath117 approaches its upper limit from eq.([eq : wa66 ] ) ) . thus the @xmath0 mean decay length is always much greater than the distance to the karmen detector , as required . also at the lower end of this lifetime range , the cosmological and astrophysical constraints may perhaps be evaded .
we now address the further implications of the @xmath0-neutrino scenario for laboratory experiments .
this scenario implies that , with the stopped @xmath8 decay chain as the neutrino source , @xmath0 will be produced not only via @xmath12 ( giving the time - signatured anomaly in the karmen detector ) , but also via @xmath118 and @xmath119 channels .
it is interesting to ask how many of these @xmath0 should decay in the karmen detector ( or other detectors ) , compared to the anomaly events , and what their signatures may be .
\(a ) the @xmath13 channel .
the rate depends on @xmath46 .
the fraction that decay in a given detector depends inversely on the momentum with which @xmath0 is produced .
hence the ratio of detected @xmath0 decays from this channel compared to
anomaly " events from @xmath12 is @xmath120 } { |u_{\mu x}|^2[m_\pi^2(m_\mu^2+m_x^2)-(m_\mu^2-m_x^2)^2 ] } \simeq 0.15\ \frac{|u_{ex}|^2}{|u_{\mu x}|^2}\ ] ] in these events @xmath0 has velocity @xmath121 and reaches the detector essentially in coincidence with the prompt @xmath122 burst from the initial pion pulse , well within the resolution defined by the 100 ns pulse length at the karmen source . for solutions with dominant @xmath67
visible decays , the spectrum of visible energy ( summed @xmath123 kinetic energies ) is shown in fig.2 , including initial @xmath0 polarization and folding in the relative decay probability in a detector . about @xmath124 of such detected decays
have visible energy greater than 40 mev , whereas each prompt @xmath122 carries only 29.8 mev total energy , so these @xmath0-decay events should be quite distinctive . for the more restricted majorana solutions , @xmath125 and the @xmath126 signal
is strongly suppressed .
solutions with dominant @xmath23 visible decays necessarily have @xmath127 ; if this is achieved with @xmath128 mev from anomaly events .
\(b ) the @xmath129 channel . once again the rate depends on @xmath46 and the fraction decaying in a given detector depends inversely on the @xmath0-momentum .
numerical calculations give the detected event ratio : @xmath130 there is a spread of velocities @xmath131 , but these events are smeared anyway by the parent muon lifetime and arrive with essentially the same time distribution as the @xmath132 and @xmath57 interaction events . the detected visible energy spectrum for solutions with dominant @xmath67 , calculated for the @xmath133 cascade with full spin - correlated decay matrix elements @xcite , is shown in fig.2 ; it has mean value @xmath134 mev .
this spectrum roughly resembles the @xmath135c@xmath136 spectrum shown in ref .
@xcite , but has a longer tail above 35 mev , so it should be possible to distinguish one from the other . as with channel ( a ) above , this channel is suppressed by @xmath125 for majorana @xmath67 solutions , but is important for a subset of @xmath23 solutions , for which the decay photon energies extend from 5 to 53 mev with mean value 26 mev .
\(c ) the @xmath137 channel . here
the rate depends on @xmath37 instead , like the anomaly " channel @xmath138 .
numerical calculations give the detected event ratio : @xmath139 as with channel ( b ) , these events have essentially the same time distribution as the @xmath132 and @xmath57 interaction events ; on the other hand , the number of these events compared to time - anomaly events is now firmly predicted .
the detected visible energy spectrum for @xmath67 solutions is shown in fig.2 ; like case
( b ) it has mean value @xmath134 mev and can be distinguished from conventional @xmath135c@xmath136 by the tail above 35 mev . for @xmath23 solutions
, the decay photon again has energies between 5 and 53 mev with mean value 28 mev . to summarize , a mainly - sterile neutrino interpretation for @xmath0 is consistent with all laboratory constraints , within limited ranges of mixing parameters , for both dirac and majorana options , although there are some astrophysical and cosmological problems .
solutions with dominant @xmath67 visible decays are constrained by eqs.([eq : shrock])-([eq : sol]),([eq : eendom ] ) ; here the mixing parameters @xmath62 and @xmath140 are adjusted to give compatible values of branching fraction @xmath108 and visible - mode lifetime @xmath141 in fig.1 ( while @xmath117 is negligible ) .
alternative solutions with dominant @xmath23 visible decays are constrained by eqs.([eq : shrock]),([eq : ngdom]),([eq : nusol])-([eq : wa66 ] ) ; here @xmath108 and @xmath141 are determined by @xmath62 and @xmath117 instead ( while @xmath142 is negligible ) .
such interpretations imply that other sources of @xmath0-production should contribute appreciable additional @xmath0-decay events in the karmen detector , with different time - structures ; in channels ( a ) and ( b ) the event rate depends on @xmath142 and is appreciable in some dirac @xmath67 solutions and some @xmath23 solutions , but in channel ( c ) the event rate is always @xmath143 of the anomaly event rate .
similar @xmath0-decay signals should also appear in the lsnd detector @xcite .
since both detectors have approximately the same density , and since both @xmath144-interaction and @xmath0-decay events have the same inverse - square dependence on distance @xmath145 ( so long as @xmath146 ) , the ratio of interactions to decays should be approximately the same in both experiments .
such @xmath0 decays could conceivably provide a background to the @xmath3 oscillation search currently under way @xcite , although the @xmath147 signal is distinguishable in principle by detecting the delayed 2.2 mev gamma from the subsequent neutron capture @xmath148 . in high - energy neutrino beams , however , where the parent pions have been boosted to energies @xmath149 with @xmath150 , the number of @xmath0-decays in a given detector volume will scale down as @xmath151 $ ] ( where @xmath152 and @xmath153 are @xmath154-restframe momentum and energy ) while the @xmath144 interactions will scale up as @xmath87 ; thus the fraction that are decays decreases as @xmath155 and rapidly becomes negligible .
hence @xmath156 decays can not contribute significantly to the apparent @xmath157 excess in atmospheric neutrino events @xcite , where the parent pions and kaons have energies of @xmath158(gev ) .
similarly , there should be a negligible @xmath156 contribution in accelerator experiments such as bnl - e776 @xcite which set an upper limit on @xmath159 and @xmath3 oscillations using gev neutrinos .
finally , we note that isosinglet ( sterile ) neutrinos occur naturally in @xmath160 and @xmath161 gut models , as members of the basic fermion families @xcite .
a very light ( @xmath162 ev ) sterile neutrino has been suggested as a possible @xcite or even necessary @xcite participant in solar neutrino oscillations , while one with a mass of @xmath158(kev ) is a good candidate for ` warm ' dark matter @xcite .
heavy singlet neutrinos could cause distinctive lepton - number - nonconserving , lepton - flavour - changing and lepton - universality - breaking effects in a wide range of laboratory processes @xcite .
* acknowledgments * rjnp thanks j. kleinfeller , i. blair and j. guy for conversations ; vb thanks r. imlay for a discussion ; ss thanks n. booth , r. mohapatra and g. raffelt for helpful comments .
this research was supported in part by the u.s .
department of energy under grant no .
de - fg02 - 95er40896 and in part by the university of wisconsin research committee with funds granted by the wisconsin alumni research foundation .
ss is a pparc advanced fellow and acknowledges support from the ec theoretical astroparticle network .
99 karmen collaboration : b. armbruster , phys .
lett . *
b348 * ( 1995 ) 19 .
r.a .
reeder , nucl .
instr . and meth .
* a334 * ( 1993 ) 353 .
d.o .
caldwell , in _ trends in astroparticle physics _ ,
stockholm 1994 , eds .
l. bergstrm , nucl . phys .
* b * , proc . suppl .
( in press ) . a. de rujula and s.l .
glashow , phys .
rev .
lett .
* 45 * ( 1980 ) 942 ; + p.b .
pal and l. wolfenstein , phys . rev .
* d25 * ( 1982 ) 766 ; + m. gronau and r. yahalom , phys . rev . * d30 * ( 1984 ) 2422 .
a. hime , r.j.n .
phillips , g.g .
ross and s. sarkar , phys .
lett . *
b260 * ( 1991 ) 381 .
particle data group , phys . rev . *
d50 * ( 1994 ) 1173 .
r.e .
shrock , phys . rev .
* d24 * ( 1981 ) 1232 , 1275 .
charm collaboration : f. bergsma , phys .
lett .
* 128b * ( 1983 ) 361 .
bebc - wa66 collaboration : a.m.
cooper - sarkar , phys . lett . * 160b * ( 1985 ) 207 .
ps-191 collaboration : g. bernardi , phys .
lett . *
166b * ( 1986 ) 479 . n. de leener - rosier , phys . rev .
* d43 * ( 1991 ) 3611 ; + t. numao , _ proc . beyond the standard model iii _ , ottawa ,
eds .
s. godfrey and p. kalyniak ( world scientific , singapore , 1993 ) ; + g. czapek , phys .
. lett .
* 70 * ( 1993 ) 17 ; + d.i .
britton , phys . rev .
* d49 * ( 1994 ) 28 .
t. bernatowicz , phys .
rev .
lett .
* 69 * ( 1992 ) 2341 ; + a. balysh , phys . lett . * b283 * ( 1992 ) 32 , preprint hep - ex/9502007 ; + j .- c .
vuilleumier , phys . rev . *
d48 * ( 1993 ) 1009 .
p. bamert , c.p .
burgess and r.n .
mohapatra , nucl phys . *
b438 * ( 1995 ) 3 .
s. petcov , sov .
j. nucl .
phys . 25 ( 1977 ) 344 , ( erratum ) 25 ( 1978 ) 698 ; + b.w .
lee , s. pakvasa , h. sugawara and r.e .
shrock , phys .
rev .
lett .
* 38 * ( 1977 ) 937 ; + t .- p . cheng and l .- f .
li , phys . rev .
* d16 * ( 1977 ) 1425 ; + w.j .
marciano and a. sanda , phys .
lett . * 67b * ( 1977 ) 303 .
argus collaboration : h. albrecht , phys . lett . *
b292 * ( 1992 ) 221 .
cleo ii collaboration : d. cinabro , phys .
rev .
lett .
* 70 * ( 1993 ) 3700 .
aleph collaboration : d. buskulic , preprint cern ppe/95 - 003 .
the lep collaborations : preprint cern ppe/94 - 187 .
g.b .
gelmini and e. roulet , preprint ucla/94/tep/36 [ hep - ph/9412278 ] ; + r.n .
mohapatra and p.b .
pal , _ massive neutrinos in physics and astrophysics _
( world scientific , 1991 ) .
d.a .
dicus , e.w . kolb and v.l .
teplitz , astrophys .
j. * 221 * ( 1978 ) 327 ; + e.w .
kolb and t. goldman , phys .
rev .
lett .
* 43 * ( 1979 ) 897 .
s. sarkar and a.m. cooper , phys .
lett . *
148b * ( 1984 ) 347 .
d. lindley , astrophys .
j. * 294 * ( 1985 ) 1 .
s. sarkar , _ proc .
xvii int .
symp .
on multiparticle dynamics _ , seewinkel , eds .
m. markytan ( world scientific , 1986 ) p 863 ; + l.m .
krauss , _ proc .
telemark minconference on neutrino masses _
( 1985 ) .
e.w . kolb and r.j .
scherrer , phys . rev .
* d25 * ( 1982 ) 1481 ; + n. terasawa , m. kawasaki and k. sato , nucl . phys . * 302 * ( 1988 ) 697 ; + s. dodelson , g. guyk and m.s .
turner , phys . rev .
* d49 * ( 1994 ) 5068 .
a. dar and s. dado , phys .
rev .
lett .
* 59 * ( 1987 ) 2368 ; + l. oberauer , c. hagner , g. raffelt and e. rieger , astropart
. phys .
* 1 * ( 1993 ) 377 ; + r.n .
mohapatra , s. nussinov and x. zhang , phys . rev .
* d49 * ( 1994 ) 3434 .
s.w . falk and d.n .
schramm , phys .
lett . *
b79 * ( 1978 ) 511 ; + r. cowsik , d.n . schramm and p. hflich ,
. lett . *
b218 * ( 1989 ) 91 . g. sigl and m.s .
turner , phys . rev .
* d51 * ( 1995 ) 1499 .
g. raffelt and g. sigl , astpart . phys .
* 1 * ( 1993 ) 165 ; + x. shi and g. sigl , phys .
lett .
b323 ( 1994 ) 360 .
k. kainulainen , j. maalampi and j.t .
peltoniemi , nucl . phys .
* b358 * ( 1991 ) 435 ; + b. mukhopadhyaya and r. gandhi , phys . rev .
* d46 * ( 1992 ) 3682 .
g. raffelt , private communication .
a. manohar , phys .
lett .
* b186 * ( 1987 ) 370 ; + r. barbieri and a. dolgov , nucl . phys . *
b349 * ( 1991 ) 743 ; + k. enqvist , k. kainulainen and m. thompson , nucl . phys . * b373 * ( 1992 ) 498 ; + x. shi , d.n .
schramm and b.d .
fields , phys . rev .
* d48 * ( 1993 ) 2563 .
p. langacker , b. sathiapalan and g. steigman , nucl . phys .
* b266 * ( 1986 ) 669 .
bebc - wa66 collaboration : a.m. cooper - sarkar , private communication ( quoted by k.s .
babu , phys .
lett . *
b321 * ( 1994 ) 140 ) .
v. barger , j. ohnemus and r.j.n .
phillips , phys . rev .
* d35 * ( 1987 ) 166 .
k.s .
hirata , phys . lett .
* b280 * ( 1992 ) 146 ; + y. fukuda , phys . lett . *
b335 * ( 1994 ) 237 .
l. borodovsky , phys .
rev .
lett .
* 68 * ( 1992 ) 274 .
v. barger , phys .
rev .
* d43 * ( 1991 ) 1759 ; + c .- s .
lim and w.j .
marciano , phys . rev .
* d37 * ( 1988 ) 1368 .
g.m .
fuller , j. primack and y .- z .
qian , preprint doe - er-40561 - 185 [ astro - ph/9502081 ] . s. dodelson and l.m .
widrow , phys .
rev .
lett .
* 72 * ( 1994 ) 17 .
see p. langacker , l. luo and a.k .
mann , rev .
mod .
phys .
* 64 * ( 1992 ) 87 ; + j. bernabeu , phys . rev . lett .
* 71 * ( 1993 ) 2695 ; + a. ilakovac and a. pilaftsis , nucl . phys . * b437 * ( 1995 ) 491 ; + p. roy , preprint tifr - th-94 - 35 [ hep - ph/9501209 ] ; and references therein . |
the phenomenon of superconductivity was discovered @xcite as a sudden drop of resistance to immeasurably small value . with the development of the topic
it was realized that the superconducting phase transition is frequently not at all sudden and the measured dependence of the sample resistance @xmath3 in the vicinity of the critical temperature @xmath4 may have a finite width .
one possible reason for this behavior and frequently the dominating factor is the sample inhomogeneity , i.e. the sample might simply consist of regions with different local critical temperatures .
however , with improving fabrication technologies it became clear that even for highly homogeneous samples the superconducting phase transition may remain broadened .
this effect is usually very small in bulk samples and becomes more pronounced in systems with reduced dimensions .
a fundamental physical reason behind such smearing of the transition is _
superconducting fluctuations_. an important role of fluctuations in reduced dimension is well known . above @xmath4
such fluctuations yield an enhanced conductivity of metallic systems @xcite . for instance
, the so - called aslamazov - larkin fluctuation correction to conductivity @xmath5 becomes large in the vicinity of @xmath4 and this effect increases with decreasing dimensionality @xmath6 .
below @xmath4 according to the general theorem @xcite fluctuations should destroy the long - range order in low dimensional superconductors .
thus , it could naively be concluded that low dimensional conductors can not exhibit superconducting properties because of strong phase fluctuation effects .
this conclusion , however , turns out to be somewhat premature .
for instance , 2d structures undergo berezinskii - kosterlitz - thouless ( bkt ) phase transition @xcite as a result of which the decay of correlations in space changes from exponential at high enough @xmath7 to power law at low temperatures . this result implies that at low @xmath7 long range phase coherence essentially survives in samples of a finite size and , hence , 2d films can well exhibit superconducting properties . can superconductivity survive also in ( quasi)-1d systems or do fluctuations suppress phase coherence thus disrupting any supercurrent ?
the answer to this question would clearly be of both fundamental interest and practical importance .
on one hand , investigations of this subject definitely help to encover novel physics and shed more light on the crucial role of superconducting fluctuations in 1d wires . on the other hand , rapidly progressing miniaturization of nanodevices opens new horizons for applications of superconducting nanocircuits and requires better understanding of fundamental limitations for the phenomenon of superconductivity in reduced dimension .
a detailed review of the present status of this field is the main purpose of this paper .
it was first pointed out by little @xcite that quasi - one - dimensional wires made of a superconducting material can acquire a finite resistance below @xmath4 of a bulk material due to the mechanism of thermally activated phase slips ( taps ) . within the ginzburg - landau theory
one can describe a superconducting wire by means of a complex order parameter @xmath8 .
thermal fluctuations cause deviations of both the modulus and the phase of this order parameter from their equilibrium values .
a non - trivial fluctuation corresponds to temporal suppression of @xmath9 down to zero in some point ( e.g. , @xmath10 ) inside the wire , see fig .
1 . as soon as the modulus of the order parameter @xmath11 vanishes the phase @xmath12 becomes unrestricted and can jump by the value @xmath13 , where @xmath14 is any integer number .
after this process the modulus @xmath11 gets restored , the phase becomes single valued again and the system returns to its initial state accumulating the net phase shift @xmath13 .
provided such phase slip events are sufficiently rare , one can restrict @xmath14 by @xmath15 and totally disregard fluctuations with @xmath16 .
( left axis , dashed line ) and phase @xmath17 ( right axis solid line ) at various moments of time : ( a ) before , ( b ) during and ( c ) after the phase slippage.,width=302 ] according to the josephson relation @xmath18 each such phase slip event causes a nonzero voltage drop @xmath19 across the wire . in the absence of any bias current
the net average numbers of positive ( @xmath20 ) and negative ( @xmath21 ) phase slips are equal , thus the net voltage drop remains zero .
applying the current @xmath22 one creates nonzero phase gradient along the wire and makes positive phase slips more likely than negative ones .
hence , the net voltage drop @xmath19 due to taps differs from zero , i.e. thermal fluctuations cause non - zero resistance @xmath23 of superconducting wires even below @xmath4 .
we would also like to emphasize that , in contrast to the so - called phase slip centers @xcite produced by a large current above the critical one @xmath24 , here we are dealing with _ fluctuation - induced phase slips _ which can occur at arbitrarily small values @xmath25 .
a quantitative theory of the taps phenomenon was first proposed by langer and ambegaokar @xcite and then completed by mccumber and halperin @xcite .
this lamh theory predicts that the taps creation rate and , hence , resistance of a superconducting wire @xmath26 below @xmath4 are determined by the activation exponent @xmath27 where @xmath28 is the effective potential barrier for taps proportional to the superconducting condensation energy ( @xmath29 is the metallic density of states at the fermi energy and @xmath30 is the bcs order parameter ) for a part of the wire of a volume @xmath31 where superconductivity is destroyed by thermal fluctuations ( @xmath32 is the wire cross section and @xmath33 is the superconducting coherence length ) . at temperatures very close to @xmath4 eq . ( [ taps ] ) yields appreciable resistivity which was indeed detected experimentally @xcite .
close to @xmath4 the experimental results fully confirm the activation behavior of @xmath3 expected from eq .
( [ taps ] ) .
however , as the temperature is lowered further below @xmath4 the number of taps inside the wire decreases exponentially and no measurable wire resistance is predicted by the lamh theory @xcite except in the immediate vicinity of the critical temperature .
experiments @xcite were done on small whiskers of typical diameters @xmath34 @xmath35 m .
recent progress in nanolithographic technique allowed to fabricate samples with much smaller diameters down to and even below
in such systems one can consider a possibility for phase slips to occur not only due to thermal , but also due to _ quantum _ fluctuations of the superconducting order parameter .
the physical picture of quantum phase slippage is qualitatively similar to that of taps ( see fig .
1 ) except the order parameter @xmath9 gets virtually suppressed due the process of _ quantum tunneling_. following the standard quantum mechanical arguments one can expect that the probability of such tunneling process should be controlled by the exponent @xmath36 , i.e. instead of temperature in the activation exponent ( [ taps ] ) one should just substitute @xmath37 , where @xmath38 is an effective attempt frequency .
this is because the order parameter field @xmath39 now tunnels under the barrier @xmath40 rather than overcomes it by thermal activation . since such tunneling process should obviously persist down to @xmath2 one arrives at a fundamentally important conclusion that _ in nanowires superconductivity can be destroyed by quantum fluctuations at any temperature including _ @xmath2 .
accordingly , such nanowires should demonstrate a non - vanishing resistivity down to zero temperature .
_ assuming _ that @xmath41 one would expect that at @xmath42 the taps dependence ( [ taps ] ) applies while at lower @xmath43 quantum phase slips ( qps ) take over , eventually leading to saturation of the temperature dependence @xmath3 to a non - zero value in the limit @xmath44 .
this behavior was indeed observed : giordano @xcite performed experiments which clearly demonstrated a notable resistivity of ultra - thin superconducting wires far below @xmath4 .
these observations could not be adequately interpreted within the taps theory and were attributed to qps .
later other groups also reported noticeable deviations from the lamh theory in thin ( quasi-)1d wires .
these experiments will be discussed in chapter 6 . it should be noted , however , that despite these developments the idea that in realistic samples superconductivity can be destroyed by quantum fluctuations was initially received with a large portion of scepticism . on one hand , this was due to a number of unsuccessful attempts to experimentally observe the qps phenomenon . on the other hand ,
some early theoretical efforts have led to the results strongly underestimating the actual qps rate . also , unambiguous interpretation of the observations @xcite in terms of qps was questioned because of possible granularity of the samples used in those experiments .
if that was indeed the case , qps could easily be created inside weak links connecting neighboring grains .
also in this case superconducting fluctuations play a very important role @xcite however in contrast to the case of uniform wires the superconducting order parameter _ needs not to be destroyed _ during a qps event .
first attempts to theoretically analyze the qps effects @xcite as well as a number of later studies were based on the so - called time - dependent ginzburg - landau ( tdgl ) equations .
unfortunately the tdgl approach is by far insufficient for the problem in question for a number of reasons : ( i ) a trivial reason is that the ginzburg - landau ( gl ) expansion applies only at temperatures close to @xmath4 whereas in order to describe qps one usually needs to go to lower temperatures down to @xmath45 .
( ii ) more importantly , also at @xmath46 tdgl equation remains applicable only in a special limit of gapless superconductors , while it fails in a general situation considered here .
( iii ) tdgl approach does not account for dissipation effects due to quasiparticles inside the qps core ( in certain cases also outside this core ) which are expected to reduce the probability of qps events similarly to the standard problem of quantum tunneling with dissipation @xcite .
( iv ) tdgl approach is not fully adequate to properly describe excitation of electromagnetic modes around the wire during a qps event ( this effect turns out to be particularly important for sufficiently long wires ) .
thus , tdgl - based description of qps effects simply can not be trusted , and a much more elaborate theory is highly desirable in this situation .
a microscopic theory of qps processes in superconducting nanowires was developed @xcite with the aid of the imaginary time effective action technique @xcite .
this theory remains applicable down to @xmath2 and properly accounts for non - equilibrium , dissipative and electromagnetic effects during a qps event .
one of the main conclusions of this theory is that in sufficiently dirty superconducting nanowires with diameters in the 10 nm range qps probability can already be large enough to yield experimentally observable phenomena . also , further interesting effects including quantum phase transitions caused by interactions between quantum phase slips were predicted @xcite . an important parameter of this theory is the qps fugacity @xmath47 where @xmath48 is the dimensionless conductance of the wire segment of length @xmath49 .
provided @xmath48 is very large , typically @xmath50 , the fugacity @xmath51 remains vanishingly small , qps events are very rare and in many cases can be totally neglected . in such case
the standard bcs mean field description should apply and a finite ( though possibly sufficiently long ) wire remains essentially superconducting outside an immediate vicinity of @xmath4 . for smaller @xmath52 qps effects already become important down to @xmath45 .
finally , at even smaller @xmath53 strong fluctuations should wipe out superconductivity everywhere in the wire .
we also point out that in the case of nanowires considered here the parameter @xmath48 is related to the well known ginzburg number as @xmath54 , i.e. the condition @xmath55 also implies that the fluctuation region becomes of order @xmath4 .
another important parameter is the ratio between the ( superconducting ) quantum resistance unit @xmath56 k@xmath57 and the wire impedance @xmath58 : @xmath59 where @xmath60 is the wire capacitance per unit length , @xmath61 is the wire kinetic inductance and @xmath62 is the london penetration depth .
provided this parameter becomes of order one , @xmath63 , superconductivity in sufficiently long wires gets fully suppressed due to intensive fluctuations of the phase @xmath17 of the superconducting order parameter .
we note that both @xmath48 and @xmath35 scale with the wire cross section @xmath32 respectively as @xmath64 and @xmath65 .
it follows immediately that _ with decreasing the cross section below a certain value the wire inevitably looses intrinsic superconducting properties due to strong fluctuation effects_. for generic parameters both conditions @xmath66 and @xmath63 are typically met for wire diameters in the range @xmath67 nm .
a number of recent experimental observations are clearly consistent with the above theoretical conclusions .
perhaps the first unambiguous evidence for qps effects in quasi-1d wires was reported by bezryadin , lau and tinkham @xcite who fabricated sufficiently uniform superconducting wires with thicknesses down to @xmath68 nm and observed that several samples showed no signs of superconductivity even at temperatures well below the bulk critical temperature .
those results were later confirmed and substantially extended by different experimental groups . at present
there exists an overwhelming experimental evidence for qps effects in superconducting nanowires fabricated to be sufficiently uniform and homogeneous .
below we will analyze the main experimental results and compare them with theoretical predictions .
yet another interesting issue is related to persistent currents ( pc ) in superconducting nanorings .
it was demonstrated @xcite that qps effects can significantly modify pc in such systems and even lead to exponential suppression of supercurrent for sufficiently large ring perimeters .
another important factor that can substantially affect pc in isolated superconducting nanorings at low @xmath7 is the electron parity number .
of particular interest is the behavior of rings with odd number of electrons which can develop _ spontaneous _ supercurrent in the ground state without any externally applied magnetic flux @xcite .
the structure of our review is as follows .
a theory of superconducting fluctuations in nanowires will be addressed in chapters 2 - 5 . in chapter 2
we discuss a general derivation of the real time effective action of a superconductor suitable for further investigations of fluctuation effects at temperatures below @xmath4 .
we also formulate the langevin equations and analyze their relation to tdgl - type of equations frequently used in the literature . in chapter 3
we adopt our general formalism to the case of superconducting nanowires and demonstrate the importance of superconducting fluctuations in such structures .
in chapter 4 we will briefly review lamh theory of thermally activated phase slips .
quantum phase slip effects will be analyzed in details in chapter 5 .
chapter 6 is devoted to an elaborate discussion of key experiments in the field and their interpretation in terms of existing theories . in chapter 7
we will analyze persistent currents in superconducting nanorings .
chapter 8 contains a brief summary of our main observations and conclusions .
some technical details are presented in appendix .
the starting point of our analysis is the formal expression for the quantum evolution operator on the keldysh contour or the so - called keldysh `` partition function '' . as usually @xcite
, the kernel of the evolution operator @xmath69 can be expressed in the form of a path integral over quantum fields defined both at the forward ( below denoted by the subscript @xmath70 ) and the backward ( denoted by the subscript @xmath71 ) branches of the keldysh contour .
we have @xmath72-is[\psi _ { b}^{\alpha},\delta_{b},v_{b},\bm{a}_{b}]},\end{aligned}\ ] ] here @xmath73 are electron grasmann fields , @xmath74 is the spin index , @xmath75 are superconducting complex order parameter fields ( which emerge as a result of the standard hubbard - stratonovich decoupling of the bcs coupling term @xcite ) , @xmath76 and @xmath77 are respectively the scalar and vector potentials .
the action @xmath78 is defined as follows @xmath79 \nonumber\\ & + \,\int d^{4}x\left ( \frac{{\bm e}^{2}-{\bm h}^{2}}{8\pi}-\frac{n_{0}|\delta|^{2}}{\lambda}\right ) .\end{aligned}\ ] ] here @xmath80 is the fermi energy , @xmath81 is the disorder potential , @xmath29 is the density of states per unit spin at the fermi level , and @xmath82 is the bcs coupling constant . in our notations
the electron charge is @xmath83 , i.e. we define @xmath84 . here and below we employ the notation @xmath85 .
as the above action is quadratic in the electron fields one can integrate them out exactly .
after that one arrives at the path integral @xmath86},\end{aligned}\ ] ] where the effective action reads @xmath87 the inverse @xmath88 matrix green - keldysh function @xmath89 can be split into @xmath90 sub - blocks ( indicated by a hat ) : @xmath91{cc}\hat{\mathcal{g}}^{-1}_{11 } & -\hat\sigma_{z}\hat\delta\\ -\hat\sigma_{z}\hat\delta & \hat{\mathcal{g}}^{-1}_{22}\end{array } \right ) \label{g-1}\ ] ] where @xmath92 here @xmath93 is the single electron hamiltonian , @xmath94{cc}\nabla\varphi_{f } + \frac{2e}{c}{\bm a}_{f } & 0\\ 0 & \nabla\varphi_{b } + \frac{2e}{c}{\bm a}_{b}\end{array } \right ) , \end{aligned}\ ] ] @xmath95{cc}v_{f } -\frac{\dot\varphi_{f}}{2e } & 0\\ 0 & v_{b } -\frac{\dot\varphi_{b}}{2e}\end{array } \right ) , \end{aligned}\ ] ] define the gauge invariant combiantions of the superconducting order parameter phase @xmath17 and the electromagnetic potentials .
the matrix @xmath96 is constructed analogously .
it reads @xmath97{cc}\delta_{f } & 0\\ 0 & \delta_{b}\end{array } \right ) .\end{aligned}\ ] ] note that both fields @xmath98 and @xmath99 are now real since we have already decoupled the phase factors @xmath100 by the gauge transformation .
although the above expression for the effective action @xmath101 ( [ seff ] ) is exact it remains too complicated for practical calculations . in order to proceed
analogously to the derivation in the matsubara technique @xcite restrict our analysis to quadratic fluctuations .
for this purpose we split the order parameter into the bcs mean field term @xmath102 and the fluctuating part .
performing a shift @xmath103 we re - define the order parameter field in a way to describe fluctuations by the fields @xmath75 .
expanding the effective action ( [ seff ] ) in powers of @xmath96 , @xmath104 and @xmath105 to the second order we obtain the action describing the quadratic fluctuations in the system @xmath106 where @xmath107\end{aligned}\ ] ] defines the first order ( in @xmath108 ) contribution to the action , @xmath109 accounts for fluctuations of the absolute value of the order parameter field , @xmath110 ^{2 } \nonumber\\ & + \hat\sigma_{z}\left ( \frac{i}{2}\{\nabla,\hat{\bm v}_{s}\}+e\hat\phi\right ) \hat f \hat\sigma_{z}\left ( \frac{i}{2}\{\nabla,\hat{\bm v}_{s}\}-e\hat\phi\right ) \hat{\bar f}\nonumber\\ & + \,\left [ \hat\sigma_{z}\left ( \frac{i}{2}\{\nabla,\hat{\bm v}_{s}\}-e\hat\phi\right ) \hat{\bar g}\right ] ^{2 } \bigg\}\label{sem}\ ] ] describes electromagnetic fields and their coupling to the phase of the order parameter field and @xmath111 is responsible for coupling of electromagnetic and phase fluctuations to the absolute value of the order parameter field . all the above contributions to the action are expressed in terms of equilibrium non - perturbed normal and anomalous @xmath90 green - keldysh matrices which enter as sub - blocks @xmath112 into the @xmath88 matrix @xmath113{cc}\hat g & \hat f\\ \hat{\bar f } & \hat{\bar g}\end{array } \right ) , \label{gggg}\ ] ] obtained by inverting the expression ( [ g-1 ] ) at @xmath114 , @xmath115 and @xmath116 .
the formal expressions for the sub - blocks @xmath112 are defined in appendix a.1 .
one can demonstrate that the matrix green - keldysh functions satisfy the ward identities ( [ ward2 ] ) and ( [ ward1 ] ) specified in appendix a.2 . making use of eq .
( [ ward2 ] ) we can rewrite the first order contribution to the action in the form @xmath117 + i\delta_{0 } \,\mathrm{tr}\,\left [ \hat\sigma_{z}\hat \varphi(\hat f-\hat{\bar f } ) \right ] \nonumber\\ & + \,\frac{ie}{2mc}\,\mathrm{tr}\,\left [ \hat\sigma_{z}\{\nabla,\hat { \bm a}\}(\hat g+\hat{\bar g})\right ] + e\,\mathrm{tr}\,\left [ \hat\sigma _ { z}\hat v(\hat g-\hat{\bar g})\right ] .\nonumber\end{aligned}\ ] ] observing that @xmath118 and introducing the current @xmath119 and the charge density @xmath120 in the non - perturbed state we further transform the action @xmath121 and get @xmath122 -\frac{i}{c}\int d^{4}x\,{\bm j}_{0}({\bm a}_{f}-{\bm a}_{b } ) \nonumber\\ & & -\,i\int d^{4}x\ , \rho_{0}(v_{f}-v_{b}).\label{s11}\ ] ] in addition , assuming the non - perturbed system state to be in thermodynamic equlibrium , we set both the current and the charge density equal zero
. then we obtain @xmath123 .\label{s1}\ ] ] finally , we assume that the equilibrium value of the order parameter @xmath30 satisfies the standard bcs gap equation @xmath124 where @xmath125 is the debye frequency . in this case
@xmath121 is cancelled by the first order contribution coming from @xmath126 and the action does not contain the first order terms in @xmath127 any more . the ward identities ( [ ward2 ] ) and ( [ ward1 ] ) also allow one to transform the contribution @xmath128 ( [ sem ] ) and cast it to the form @xmath129 where the terms @xmath130 , @xmath131 and @xmath132 define the terms of a different physical origin which we will identify respectively as josephson , london and drude contributions to the effective action . they read @xmath133 , \label{sj}\ ] ] @xmath134 , \label{sl}\ ] ] @xmath135,\;\;\end{aligned}\ ] ] where @xmath136 at low frequencies and temperatures the josephson contribution @xmath130 can be large , thus suppressing fluctuations of the gauge invariant potential @xmath137 .
in this case one can set @xmath138 and get @xmath139 , which is just the well known josephson relation between the phase and the electric potential .
note that for ultra - thin superconducting wires the josephson relation can be violated , as it will be demonstrated below .
the london contribution @xmath131 is responsible for the screening of the magnetic field penetrating inside the superconductor .
finally , the drude contribution @xmath132 remains non - zero in the normal state where it accounts for ohmic dissipation due to flowing electric currents .
since the correction to the magnetic susceptibility in normal metals is usually small , one can ignore the vector potential in the expression for @xmath140 ( [ u ] ) .
afterwards one can again apply the ward identities ( [ ward1 ] ) and rewrite @xmath132 in the form @xmath141.\;\;\;\;\;\label{sd}\ ] ] at last , let us consider the cross term @xmath142 . again applying the identity ( [ ward2 ] )
we cast this term to the form similar to eq .
( [ s11 ] ) : @xmath143 \nonumber\\ & -\frac{i}{c}\int d^{4}x\,\delta{\bm j}_{0}({\bm a}_{f}-{\bm a}_{b } ) -i\int d^{4}x\ , \delta\rho_{0}(v_{f}-v_{b}),\nonumber\end{aligned}\ ] ] where @xmath144 define the first order corrections to the current and the charge density due to fluctuating order parameter fields .
one can verify that in the absence of both particle - hole asymmetry and charge imbalance these corrections vanish .
likewise , in this case we have @xmath145 .
thus , we conclude that @xmath146 for clarity , we now summarize again the results derived this section .
the complete expression for the effective action describing quadratic fluctuations in a superconductor reads @xmath147 where the terms @xmath121 , @xmath126 , @xmath130 , @xmath131 and @xmath132 are defined respectively in eqs .
( [ s1 ] ) , ( [ sdelta ] ) , ( [ sj ] ) , ( [ sl ] ) and ( [ sd ] ) .
below we will mainly be interested in the limit of so - called dirty superconductors , i.e. we assume that the concentration of defects in the system is sufficiently high and the electron motion is diffusive . in the case of ultra - thin superconducting channels
only this limit appears to be of practical interest , since usually the electron elastic mean free path @xmath148 does not exceed the diameter of the wire . since we will mainly focus our attention on wires with diameters in the 10 nm range , realistic values of @xmath148 should be typically in the same range , i.e. we have @xmath149 @xmath35 m and @xmath150 . in order to account for processes with characteristic length scales exceeding the electron mean free path @xmath148 it will be convenient for us to perform disorder averaging directly in the effective action .
to this end we substitute explicit expressions for the green functions ( [ g ] ) into the effective action derived in the previous section and then apply the standard rules of averaging for the electron wave functions . in the diffusion approximation
we have @xmath151 @xmath152.\label{av}\ ] ] here @xmath153 is the diffusion constant and @xmath154 is the diffuson defined as a solution of the diffusion equation @xmath155 in the following we will mainly consider spatially extended systems in which case one has @xmath156 employing eqs .
( [ av ] ) we arrive at the following expression for the effective action @xmath157 where we introduced `` classical '' @xmath158 and `` quantum '' @xmath159 components of the order parameter field and used analogous definitions for other fluctuating variables @xmath160 , @xmath161 and @xmath162 .
the four kernels @xmath163 ( @xmath164 ) are defined as follows @xmath165}{2}\coth\frac{\omega}{2t},\end{aligned}\ ] ] where we denote @xmath166 .
explicit expressions for the functions @xmath167 , @xmath168 , @xmath169 and @xmath170 are rather cumbersome .
they are presented in appendix a.3 respectively in eqs .
( [ chidelta ] ) , ( [ chij ] ) , ( [ chil ] ) and ( [ chid ] ) . here
we provide simple analytical expressions valid in some limiting cases .
let us first concentrate on the low temperature limit @xmath44 . in this limit at small frequencies and wave vectors
@xmath171 we find @xmath172 while in the limit @xmath173 ( @xmath174 is the euler constant ) one finds @xmath175 , \nonumber\\ \chi_{l } & = \frac{8m^{2}n_{0}d\delta_{0}^{2}(t)}{\omega^{2}+d^{2}q^{4 } } \nonumber\\ & \times\ , \left [ dq^{2}\ln\frac{-i\omega+dq^{2}}{\delta_{0}(0)}+i\omega\ln \frac{-i\omega}{2\delta_{0}(0)}\right ] , \nonumber\\ \chi_{d } & = \frac{\sigma}{-i\omega+dq^{2}},\label{chid1}\ ] ] where @xmath176 is the normal state drude conductivity .
here we explicitly indicated the temperature dependence of the superconducting gap @xmath30 in order to emphasize that these asymptotic expressions are valid at all temperatures rather than only in the limit @xmath177 .
let us now consider higher temperatures @xmath178 .
at @xmath179 our general expression for @xmath167 reduces to the standard result @xmath180 , \label{psi}\ ] ] @xmath170 is again defined by the drude formula ( [ chid1 ] ) , while two other @xmath181-functions vanish identically in this temperature interval , @xmath182 .
the latter observation implies that phase fluctuations remain unrestricted in this case .
hence , no taylor expansion of the action in the phase @xmath17 can be performed . in this case
it is more convenient to undo the gauge transformation restoring the initial dependence of the action on the complex order parameter field and then to expand the action in this field . for simplicity ignoring electromagnetic fields and expanding the action to the second order in @xmath183 we find @xmath184 in the limit of small frequencies and wave vectors @xmath185
one recovers the standard expression @xmath186 which usually serves as a starting point for the derivation of the so - called time dependent ginzburg - landau equation ( tdgl ) which we will address shortly below .
turning to temperatures below the critical one , @xmath187 , and expanding the @xmath181-functions ( [ chidelta])-([chid ] ) in powers of @xmath102 we obtain @xmath188 these results apply for @xmath189 and in the limit of small wave vectors and frequencies , @xmath190 . here
@xmath30 obeys the standard bcs self - consistency gap equation at @xmath191 : @xmath192 for @xmath193 we obtain non - analytic expressions .
for example , @xmath168 in this limit reads @xmath194 \bigg\}.\end{aligned}\ ] ] more accurate expressions for the kernels @xmath167 , @xmath168 and @xmath195 valid at temeperatures close to @xmath4 and at any @xmath196 are given in appendix , see eqs .
( [ chideltaap1]-[chilap ] ) . only in the limit
@xmath197 those expressions match with the well known results for the coefficients of the linearized ( time - independent ) ginzburg - landau equation : @xmath198 at frequencies @xmath199 the functions @xmath167 and @xmath168 turn out to be parametrically different taking much higher values : @xmath200 while @xmath169 is still given by eq .
( [ gl ] ) . the drude susceptibility @xmath170
may be taken in the usual form ( [ chid1 ] ) in both cases .
thus , already at small frequencies ( well below the gap @xmath201 ) microscopic results can strongly deviate from those frequently used within semi - phenomenological tdgl approach . at higher frequencies and/or wave vectors
@xmath202 this difference becomes even more pronounced , cf .
( [ chis ] ) .
one can demonstrate that these kernels are not independent and obey the following exact identity @xmath203 which directly follows from the ward identities ( [ ward1 ] ) .
in addition , in the diffusive limit the kernels @xmath168 and @xmath169 are related to each other as @xmath204 this latter relation applies only for dirty superconductors . for clarity
, it is worthwhile to display the relation between the @xmath205-kernels derived here and some other quantities analyzed in the literature . for example , one can introduce the complex conductivity @xmath206 of a superconductor @xcite @xmath207 where @xmath208 and @xmath209 are the fourier components of respectively the current density and the electric field . in ref .
@xcite the function @xmath210 was analyzed which expresses the current via the vector potential : @xmath211 both @xmath210 and @xmath206 are related to the kernels @xmath169 and @xmath170 as follows @xmath212 .\label{qsigma}\ ] ] let us now rewrite our results in a slightly different manner .
the effective action @xmath213 can be equivalently defined by means of the following formula @xmath214 } \right\rangle _ { \xi_{j}},\label{stoch}\ ] ] where @xmath215,\label{sxi}\ ] ] and averaging is performed over three different stochastic variables @xmath216 , @xmath217 , @xmath218 defined by the pair correlators @xmath219,\nonumber\\ \langle\xi_{j}(t,{\bm r})\xi_{j}(0,0)\rangle & = \int\frac{d^{4}q}{(2\pi)^{4}}\,e^{-i\omega t+i{\bm qr}}\ , \coth\frac{\omega}{2t}\ , \nonumber\\ & \times\ , \mathrm{im}\,\big[\chi_{j}(q)+q^{2}\chi_{d}(q)\big ] , \nonumber\\ \langle\xi_{l}^{\alpha}(t,{\bm r})\xi_{l}^{\beta}(0,0)\rangle & = \delta_{\alpha\beta}\int\frac{d^{4}q}{(2\pi)^{4}}\,e^{-i\omega t+i{\bm qr}}\ , \coth\frac{\omega}{2t}\,\nonumber\\ & \times\ , \mathrm{im}\,\left [ -\chi_{l}(q)+\frac{m^{2}\omega^{2}}{e^{2}}\chi_{d}(q)\right ] , \nonumber\\ \langle{\bm\xi}_{l}(t,{\bm r})\xi_{j}(0,0)\rangle & = -\int\frac{d^{4}q}{(2\pi)^{4}}\,e^{-i\omega t+i{\bm qr}}\ , \coth\frac{\omega}{2t}\ , \nonumber\\ & \times\ , \frac{m\omega{\bm q}}{e } \ , \mathrm{im}\,\left [ \chi_{d}(q)\right ] .\label{correl}\ ] ] all other cross correlators of the above stochastic variables are equal to zero .
the representation ( [ stoch ] ) is just the result of the standard hubbard - stratonovich decoupling transformation in the effective action ( [ action ] ) .
we have also used the identity @xmath220 let us now find the least action paths for @xmath221 . setting the variational derivatives of the action ( [ sxi ] ) with respect to quantum fields
@xmath222 , @xmath223 , @xmath224 and @xmath225 equal to zero we arrive at four different equations for the fields @xmath226 , @xmath227 , @xmath228 and @xmath229 which provide the minimum for the action @xmath221 ( [ sxi ] ) .
the first equation describes fluctuations of the absolute value of the order parameter .
it reads @xmath230 the second one is the continuity equation for the supercurrent .
we obtain @xmath231 where we introduced the superconducting density @xmath232 and the superconducting current density @xmath233 @xmath234 the remaining two saddle point equations take the form @xmath235 and @xmath236 here @xmath237 is the normal quasiparticle current .
( [ eqdelta ] ) , ( [ eqjs ] ) , ( [ 3rd ] ) and ( [ maks ] ) together with noise correlators ( [ correl ] ) represent the set of _ langevin equations _ fully describing quantum dynamics of the order parameter and electromagnetic fields for dirty superconductors within the gaussian approximation . as it is clear from our derivation , these equations remain valid provided the electron distribution function is not driven far from equilibrium .
generalization of our approach to non - equilibrium situations is also possible but will not be discussed here .
now let us establish the relation between our results and the approach based on the so - called time dependent ginzburg - landau equation ( tdgl ) which is widely used to model various non - stationary effects in superconductors at temperatures close to @xmath4 .
for example , above the critical temperature this tdgl approach allows to correctly evaluate the so - called aslamazov - larkin fluctuation correction to the conductivity of the system .
below @xmath4 it enables one to describe formation of phase slip centers and the resistive state of current biased superconducting wires .
relative simplicity of the tdgl approach makes it possible to apply powerful numerical methods thus making this technique particularly appealing . in many cases
the tdgl - based analysis was employed even far beyond its applicability range , e.g. , in order to describe quantum phase slips in superconducting nanowires at @xmath238 .
the tdgl equation is usually written in the following simple form @xcite @xmath239 is the so - called ginzburg - landau time and @xmath240 although this form can be justified for gapless superconductors at high concentration of magnetic impurities , in general no consistent microscopic derivation of eq .
( [ tdgl1 ] ) was ever performed .
nevertheless , it is usually believed that eq .
( [ tdgl1 ] ) is microscopically justified at temperatures above @xmath4 where the average value of the bcs order parameter is zero , @xmath241 .
unfortunately , our present microscopic derivation ( as well as earlier imaginary time analysis @xcite ) does not fully support this statement .
an independent analysis based on the real - time non - linear @xmath242-model was recently performed by levchenko and kamenev @xcite who also noticed that even at @xmath243 eq .
( [ tdgl1 ] ) is not quite correct .
these authors formulated a more accurate real - time tdgl equation in the form convenient for a comparison with eq .
( [ tdgl1 ] ) : @xmath244 \delta(x ) - \frac{7\zeta(3)}{\pi^{3}t } \delta(x ) \int d^{4}x^{\prime}{\mathcal{d}}(x , x^{\prime } ) \nonumber\\ & \times\ , \left ( \delta^{*}(x^{\prime})\frac{\partial\delta(x^{\prime})}{\partial t^{\prime } } -2ei |\delta(x^{\prime2 } \frac{\partial{\mathcal{k}}(x^{\prime})}{\partial t^{\prime}}\right ) = \tilde\xi_{\delta}(x),\label{tdgl2}\ ] ] where @xmath245 $ ] .
it is obvious that eq .
( [ tdgl2 ] ) does not in general coincide with eq .
( [ tdgl1 ] ) . at the same time
, it is satisfactory to observe that eq .
( [ tdgl2 ] ) agrees with our results up to terms @xmath246 . in order to demonstrate this fact it is necessary to identify @xmath247 and consider terms linear in @xmath226 and @xmath227 . in this way
one arrives at our langevin eqs .
( [ eqdelta])-([eqjs ] ) with @xmath248const , where the functions @xmath167 , @xmath168 , @xmath169 coincide with those given by eq .
( [ chis ] ) in the leading order in @xmath102 .
thus , we conclude that in the limit @xmath249 our langevin equations are equivalent to tdgl - type of equation @xcite within the order @xmath250 .
some differences , however , arise for higher order terms , namely for terms @xmath251 originating from the functions @xmath168 and @xmath169 as well as for terms @xmath250 emerging from @xmath167 .
this observation indicates that eq .
( [ tdgl2 ] ) is still not fully justified at temperatures @xmath187 .
in fact , it is not quite clear to us whether it would be of any practical importance to pursue the gl expansion up to terms @xmath252 in the tdgl - type equations .
of course , a regular expansion in powers of @xmath253 can be performed in the initial effective action ( [ seff ] ) . in the order @xmath254
this expansion generates many complicated non - local ( both in space and in time ) terms containing the quantum field @xmath255 with @xmath14 ranging from 1 to 4 . in order to recover terms @xmath252 in the tdgl equation one should disregard all terms in the action @xmath256 with @xmath257 . in certain situations
this approximation might be difficult to justify .
in addition , the remaining terms @xmath252 are hardly tractable except in the zero frequency limit . finally , the whole approach remains restricted to temperatures @xmath258 . in view of all these problems
it appears more appealing to perform the expansion of the effective action in superconducting fluctuations around the mean field value @xmath102 .
this strategy was pursued in the bulk of this chapter . restricting these expansion to second order terms in @xmath183 we arrive at the langevin equations ( [ eqdelta])-([eqjs ] ) with @xmath167 , @xmath168 and @xmath169 defined in eqs . ( [ chidelta]-[chil ] ) .
this approach remains applicable at all temperatures down to @xmath2 and is sufficient for practical calculations in a large number of situations .
we now turn to the specific case of sufficiently long and very thin superconducting wires which will be of particular interest for us here . for such systems
the terms describing the action of free electro - magnetic field can be rewritten in the form @xmath259 here we have defined the coordinate along the wire @xmath260 , the capacitance per unit length of the wire @xmath60 and the inductance times unit length @xmath261 .
@xmath262 stands for the component of the vector potential parallel to the wire . for a cylindric wire with radius @xmath263 embedded in a dielectric environment with susceptibility
@xmath264 , the capacitance @xmath60 and inductance @xmath261 are @xmath265 where @xmath266 is the distance from the center of the wire and the bulk metallic electrode . in order to transform other terms one should apply a simple rule @xmath267 where @xmath32 is the wire cross section . in the low temperature limit @xmath44
all @xmath181-functions ( [ chit0 ] ) are real and , hence , the noise terms in all four langevin equations ( [ eqdelta]-[maks ] ) vanish .
this enables propagation of electromagnetic modes along a quasi-1d superconducting wire .
the equations of motion for such a wire take the form @xmath268 for dirty metallic wires with the diameter of the order of superconducting coherence length @xmath269 one finds @xmath270 and @xmath271 . in this case
the last equation gives @xmath272 , while the second and third equations describe the propagation of the plasmon mooij - schn mode @xcite with dispersion @xmath273 where the velocity of this mode @xmath274 is @xmath275 and @xmath276 is the kinetic inductance of a superconducting wire .
the effective action ( [ action ] ) fully accounts for gaussian fluctuations in diffusive superconducting structures .
for instance , from eqs .
( [ eqdelta]-[maks ] ) one readily establishes the correlation functions for all fluctuating variables in our problem . for the order parameter fields
@xmath183 in a quasi-1d wire we have @xmath277 \coth\frac{\omega}{2t}.\label{corr}\ ] ] correlation functions @xmath278 , @xmath161 and @xmath162 are defined analogously via the corresponding kernels @xmath168 , @xmath169 and @xmath170 .
consider gaussian fluctuations of the order parameter in thin one - dimensional wires .
the simplest possible average @xmath279 is divergent since the function @xmath280 grows very slowly at large @xmath281 and @xmath282 .
let us define and analyze a slightly different object @xmath283 one can verify that , for example , the non - local kernel @xmath284 significantly decays as long as @xmath285 exceeds @xmath286 and @xmath287 becomes bigger than the coherence length @xmath49 .
therefore , the parameter @xmath288 provides a qualitative measure of the ratio between the fluctuation correction to the current and its mean field value . in order to estimate the parameter @xmath288 we note that at low temperatures @xmath44 the kernel @xmath167 ( [ chidelta ] ) can be expressed in the form @xmath289 where @xmath290 is a certain dimensionless function
. then we obtain @xmath291 where @xmath292 is a numerical prefactor .
below we will demonstrate that the probability for quantum tunneling of the order parameter field @xmath293 in superconducting nanowires is proportional to @xmath294 $ ] , where @xmath295 is the action of single quantum phase slips ( qps ) .
hence , for @xmath296 gaussian fluctuations are small and qps events are rare , which are important pre - conditions for the bcs mean field theory . on the other hand , at @xmath297 one enters the regime of strong non - gaussian fluctuations which fully suppress the mean field order parameter thus driving the wire to a normal state .
the concept of qps also becomes ill - defined in this regime of strong quantum fluctuations .
the same conclusions can be extracted from the so - called ginzburg - levanyuk criterion .
let us consider the ginzburg number @xmath298 defined as the value @xmath299 at which the fluctuation correction to the specific heat becomes equal to the specific heat jump at the phase transition point . in the case of quasi-1d wires
this number reads @xcite : @xmath300 where @xmath301 is the elastic mean free time . typically in thick wires
one finds @xmath302 and fluctuations become strong only in a very narrow region close to @xmath4 , i.e. at @xmath303 .
one can also rewrite eq .
( [ gi ] ) as @xmath304 or simply @xmath305 , where @xmath48 is the dimensionless conductance of the wire segment of length @xmath49 .
thus , for @xmath306 the width of the fluctuation region @xmath307 is comparable to @xmath4 and the bcs mean field approach becomes obsolete down to @xmath2 . to complete our analysis
we will briefly address the imaginary time ( matsubara ) version of the effective action .
technically it is more convenient to deal with this form of the action provided one needs to account for quantum tunneling processes .
this is precisely what we will do below when we describe quantum phase slips in superconducting nanowires .
the calculation is described in details in refs .
@xcite and is completely analogous to one carried out above in real time .
our starting point is the path integral representation of the grand partition function @xmath308 where @xmath309 is the euclidean version of the effective action , the fluctuating order parameter field @xmath293 as well as scalar and vector potentials @xmath19 and @xmath310 depend on coordinate @xmath260 along the wire and imaginary time @xmath311 restricted to the interval @xmath312 .
as before , assuming that deviations of the amplitude of the order parameter field from its equilibrium value @xmath102 are relatively small we expand the effective action in powers of @xmath313 and in the electromagnetic fields up to the second order terms .
the next step is to average over the random potential of impurities .
after that the effective action becomes translationally invariant both in space and in time .
performing the fourier transformation we obtain @xcite @xmath314 the functions @xmath315 , @xmath316 , @xmath317 and @xmath318 , which depend both on the frequencies and the wave vectors , are expressed in terms of the averaged products of the matsubara green functions @xcite .
these functions represent the imaginary time version of the analogous real time functions @xmath319 already encountered above . in order to recover the expressions for @xmath320
one just needs to substitute @xmath321 in eqs .
( [ chidelta])-([chid ] ) for @xmath322 .
the action @xmath309 ( [ a105 ] ) represents the imaginary time analogue of the real time effective action ( [ action ] ) .
note that the action @xmath309 ( [ a105 ] ) is quadratic both in the voltage @xmath19 and the vector potential @xmath262 .
hence , these variables can be integrated out exactly . performing this integration
one arrives at the effective action which only depends on @xmath17 and @xmath323 .
we obtain @xmath324 the general expression for @xmath325 and the saddle point relations between the electromagnetic potentials and the fluctuating phase @xmath17 are presented in appendix a.4 ( eqs .
( [ calf])-([aphi ] ) ) for completeness .
as we already discussed , usually the wire geometric inductance remains unimportant .
therefore here and below we put @xmath326 .
( [ calf])-([aphi ] ) get simplified and read @xmath327 and @xmath328 note that according to eq .
( [ noj ] ) the josephson relation @xmath329 is in general _ not _ satisfied .
this relation may approximately hold only in the limit @xmath330 . making use of the results presented in appendix one
easily observes that in the important limit of small elastic mean free paths @xmath148 the latter condition is obeyed only at low frequencies and wave vectors @xmath331 and @xmath332 .
let us now perform yet one more approximation and expand the action in powers of @xmath281 and @xmath333 . keeping the terms of the order @xmath334 and @xmath335 we find @xmath336 the term @xmath337 turns out to be equal to zero .
at even smaller wave vectors , @xmath338 we get @xmath339 here , as before , we have assumed @xmath340 .
the first two terms in this action correspond to the effective hamiltonian of the form @xmath341\ ] ] which again defines the mooij - schn plasma modes propagating along the wire with the velocity @xmath274 ( [ c0 ] ) .
the effective action ( [ a116 ] ) allows to directly evaluate the fluctuation correction to the order parameter in superconducting nanowires . performing gaussian integration over both @xmath17 and @xmath323 we arrive at the wire free energy @xmath342 , \end{aligned}\ ] ] where @xmath343 is the standard bcs free energy
the order parameter is defined by the saddle point equation @xmath344 and can be written in the form @xmath345 , where @xmath102 is the solution of the bcs self - consistency equation @xmath346 ( [ bcs ] ) and the fluctuation correction @xmath347 has the form @xmath348 .\label{ddelta}\ ] ] making use of the above expressions for the functions @xmath325 and @xmath349 and having in mind that for a wire of length @xmath350 one has @xmath351 , at @xmath352 we obtain @xmath353 in eq .
( [ estim22 ] ) fluctuations of both the phase and the absolute value of the order parameter give contributions of the same order .
the estimate ( [ estim22 ] ) again demonstrates that at low temperatures suppression of the order parameter in superconducting nanowires due to gaussian fluctuations remains weak as long as @xmath354 and it becomes important only for extremely thin wires with @xmath355 .
finally let us return to the action ( [ a105 ] ) which we will use to illustrate a deficiency of the tdgl approach in the imaginary time .
considering the superconducting part of the action only and assuming that temperature is close to @xmath4 we can identify @xmath323 with @xmath293 and set @xmath241 in all the @xmath356-kernels .
exactly as in the real - time approach one then has @xmath357 and the phase fluctuations become unrestricted . for this reason one should again undo the gauge transformation and return to the complex order parameter field . for simplicity
let us ignore both the scalar and the vector potentials .
the tdgl action for the wire is then usually written in the form @xmath358 this form can be obtained from the action ( [ a105 ] ) by formally expanding the kernel @xmath315 in matsubara frequencies and wave vectors @xmath359 , cf .
( [ psi ] ) and ( [ limt ] ) .
note , however , that since the validity of the gl expansion is restricted to temperatures @xmath258 , the matsubara frequencies @xmath360 are never really smaller than @xmath361 .
hence , the expansion @xmath362 which yields ( [ ssgl ] ) is never correct except in the stationary case @xmath363 .
already these simple arguments illustrate the failure of the tdgl action ( [ ssgl ] ) in the matsubara technique .
further problems with this tdgl approach arise in the presence of the electromagnetic potentials @xmath19 and @xmath262 .
we refer the reader to the paper @xcite for the corresponding analysis .
as we already discussed , sufficiently thin superconducting wires can acquire non - zero resistance even below @xmath4 due to fluctuations of the superconducting order parameter . in this section
we will address thermal fluctuations which are particularly important in the immediate vicinity of @xmath4 .
the theory of thermally activated phase slips ( taps ) was developed by langer and ambegaokar @xcite and then completed by mccumber and halperin @xcite . here
we will briefly review this lamh theory with minor modifications related to the fact that the tdgl - based approach is not sufficiently accurate to correctly determine the pre - exponent in the expression for the taps rate .
this rate @xmath364 is defined by the standard activation dependence @xmath365 where @xmath366 is the free energy difference which determines an effective potential barrier which the system should overcome in order to create a phase slip . in order to evaluate @xmath366
we make use of the general expression for the ginzburg - landau free energy functional @xmath367&= sn_{0}\int dx \bigg ( \frac{\pi d}{8t}\left| \frac { \partial\delta}{\partial x}\right| ^{2 } \nonumber\\ & + \ , \frac{t - t_{c}}{t_{c } } |\delta|^{2 } + \frac{7\zeta(3)}{16\pi^{2}t^{2}}|\delta|^{4 } \bigg ) .\label{fgl}\ ] ] the saddle point paths for this functional are determined by the standard gl equation @xmath368 in the case of a long quasi-1d wire this equation has two solutions , a trivial one @xmath369 providing the minimum for the free energy and a metastable one @xmath370 the potential barrier @xmath366 in eq .
( [ gamma111 ] ) is set by the difference @xmath371-f[\delta_{0}]=\frac{16\pi^{2}}{21\zeta(3 ) } sn_{0}\sqrt{\pi d}(t_{c}-t)^{3/2}.\label{df}\ ] ] note that this result applies as long as the transport current @xmath25 across the wire is sufficiently small . with increasing @xmath25 the height of the potential barrier @xmath366 decreases and finally vanishes as @xmath25 approaches the critical ( depairing ) current of the wire .
the corresponding expression for @xmath372 can be found in refs .
recently a microscopic calculation of @xmath366 in the case of a clean single channel superconducting wire was reported in ref .
@xcite .
now let us turn to the pre - exponent @xmath373 in the expression for the taps rate ( [ gamma111 ] ) . in order to evaluate
@xmath373 one should go beyond the stationary free energy functional ( [ fgl ] ) and include time - dependent fluctuations of the order parameter field @xmath374 . in ref .
@xcite this task was accomplished within the framework of a tdgl - based analysis .
employing tdgl equation it is possible to re - formulate the problem in terms of the corresponding fokker - planck equation @xcite which can be conveniently solved for the problem in question .
since the important time scale within the tdgl approach is the ginzburg - landau time @xmath375 ( [ taugl ] ) , this time also enters the expression for the pre - exponent @xmath373 derived in @xcite .
unfortunately , as it was demonstrated , e.g. , in chapters 2 and 3 , the tdgl approach fails below @xmath4 .
hence , one should employ a more accurate effective action analysis .
the microscopic effective action for superconducting wires is rather complicated and it can not be easily reduced to any fokker - planck - type of equation . for this reason , below we will take a somewhat different route @xcite and combine our effective action approach with the well known general formula for the decay rate of a metastable state ( see , e.g. , @xcite ) @xmath376
as our effective action does not contain the parameter @xmath375 we expect that our final result for the pre - exponent will not contain this parameter either .
following the standard procedure @xcite we expand the general expression for the effective action around both saddle point solutions ( [ delta0 ] ) and ( [ deltam ] ) up to quadratic terms in both the phase @xmath17 and @xmath323 . neglecting the contributions from fluctuating electromagnetic fields
we obtain @xmath377 + \delta^{2 } s_{0/m},\ ] ] where @xmath378,\nonumber\\ \delta^{2 } s_{m } & = \frac{st}{2 } \sum_{\omega_{n}}\int dxdx^{\prime } \ ; \nonumber\\ & \times\ , \big[\delta\delta(\omega_{n},x)\tilde\chi_{\delta}^{(m)}(|\omega _ { n}|;x , x^{\prime } ) \delta\delta(\omega_{n},x^{\prime})\nonumber\\ & + \ , \varphi(\omega_{n},x)k_{\varphi}^{(m)}(|\omega_{n}|;x , x^{\prime } ) \varphi(\omega_{n},x^{\prime } ) \big].\end{aligned}\ ] ] here @xmath379 are bose matsubara frequencies .
the functions @xmath380 and @xmath381 are expressed in terms of the kernels @xmath315 , @xmath316 and @xmath317 as follows : @xmath382 the functions @xmath383 and @xmath384 describe the fluctuations around the coordinate dependent metastable state @xmath385 , and , therefore , can not be straitforwardly related to @xmath315 , @xmath316 and @xmath317 .
fortunately , the explicit form of @xmath383 and @xmath384 is not important for us here .
the pre - exponent @xmath373 in eq .
( [ gamma111 ] ) is obtained by integrating over fluctuations @xmath323 in the expression for the grand partition .
one arrives at a formally diverging expression which signals decay of a metastable state .
after a proper analytic continuation one arrives at the decay rate in the form ( [ gamma111 ] ) with @xmath386 here it is necessary to take an imaginary part since one of the eigenvalues of the operator @xmath387 is negative .
the key point is to observe that at @xmath388 all matsubara frequencies @xmath389 except for one with @xmath390 strongly exceed the order parameter , @xmath391 .
hence , for all such values the function @xmath392 approaches the asymptotic form ( [ psi ] ) which is not sensitive to superconductivity and we have @xmath393 and @xmath394 .
the corresponding determinants in eq .
( [ b1 ] ) cancel out and only the contribution from @xmath363 remains .
it yields @xmath395 the ratio of these determinants can be evaluated with the aid of the gl free energy functional ( [ fgl ] ) with the result @xcite @xmath396 where @xmath366 is the free energy barrier ( [ df ] ) , @xmath350 is the wire length and @xmath397 is the superconducting coherence length in the vicinity of @xmath4 . combining all the above results we arrive at the taps rate @xmath398 .\label{gamma222}\ ] ] as we expected , this result ( [ gamma222 ] ) does not contain the ginzburg - landau time @xmath375 and exceeds the corresponding expression @xcite by the factor @xmath399 .
( [ gamma222 ] ) is applicable at @xmath400 and as long as @xmath401 .
combining these two inequalities with eq .
( [ df ] ) we arrive at the condition @xmath402 where the ginzburg number @xmath403 is defined in eq .
( [ gi ] ) . the double inequality ( [ abe ] )
is standard for the gl theory .
obviously , it also restricts the applicability range of the lamh theory .
every phase slip event implies changing of the superconducting phase in time in such a way that the total phase difference values along the wire before and after this event differ by @xmath404 .
since the average voltage is linked to the time derivative of the phase by means of the josephson relation , @xmath405 , for the net voltage drop across the wire we obtain @xmath406 , \ ] ] where @xmath407 are the taps rates corresponding to the phase changes by @xmath404 . in the absence of any bias current @xmath408 both rates are equal @xmath409 and the net voltage drop @xmath19 vanishes . in the presence of a non - zero bias current the symmetry between these two rates
is lifted since in complete analogy to the case of josephson junctions ( cf .
, e.g. , @xcite ) the potential barrier for these two processes differ .
as long as the bias current @xmath25 is sufficiently small , we obtain @xmath410 thus , at such values of @xmath25 and at temperatures slightly below @xmath4 the @xmath411 curve for quasi-1d superconducting wires takes the form @xmath412 with @xmath364 defined in eq .
( [ gamma222 ] ) .
this important result @xcite implies that thermal fluctuations effectively destroy long range phase coherence in the system and the wire acquires non - vanishing resistance @xmath23 even below @xmath4 .
this resistance demonstrates strong ( exponential ) dependence on temperature and the wire cross section @xmath413 leading to effective fluctuation - induced broadening of the superconducting phase transition which can be detected experimentally .
the corresponding discussion is presented in chapter 6 . to complete our description of thermal fluctuations in superconducting wires
we point out that in addition to non - zero resistance ( [ taps r(t ) ] ) taps also cause the voltage noise below @xmath4 . treating taps as independent events
one immediately concludes that they should obey poissonian statistics .
hence , the voltage noise power @xmath414 is proportional to the taps rate @xmath364 .
the contributions from taps changing the phase by @xmath404 add up and we obtain @xmath415 similarly to the wire resistance the voltage noise rapidly decreases as one lowers the temperature away from @xmath4 . only in the vicinity of the critical temperature this noise remains appreciable and can be detected in experiments .
as temperature goes down thermal fluctuations decrease and , hence , taps become progressively less important and eventually die out in the limit @xmath238 . at low enough temperatures _ quantum fluctuations _ of the order parameter field @xmath293
take over and essentially determine the behavior of ultra - thin superconducting wires .
as we have already discussed , the most important quantum fluctuations in such wires are _ quantum phase slips _ ( qps ) .
each qps event involves suppression of the order parameter in the phase slip core and a winding of the superconducting phase around this core .
this configuration describes quantum tunneling of the order parameter field through an effective potential barrier and can be conveniently described within the imaginary time formalism . below we will elaborate on the microscopic theory of quantum phase slips in superconducting nanowires .
in doing so , to a large extent we will follow the papers @xcite .
let us denote the typical size of the qps core as @xmath416 and the typical ( imaginary time ) duration of the qps event as @xmath417 . at this stage
both these parameters are not yet known and remain to be determined from our subsequent analysis .
it is instructive to separate the total action of a single qps @xmath418 into a core part @xmath419 around the phase slip center for which the condensation energy and dissipation by normal currents are important ( scales @xmath420 , @xmath421 ) , and a hydrodynamic part outside the core @xmath422 which depends on the hydrodynamics of the electromagnetic fields , i.e. @xmath423 let us first evaluate the hydrodynamic part @xmath422 .
this task is simplified by the fact that outside the core the absolute value of the order parameter field remains equal to its mean field value @xmath102 , and only its phase @xmath424 changes in space and time . without loss of generality
we can assume that the absolute value of the order parameter is equal to zero at @xmath425 and @xmath10 .
for sufficiently long wires and outside the qps core the saddle point solution corresponding to a single qps event should satisfy the identity @xmath426 which follows from the fact that after a wind around the qps center the phase should change by @xmath427 . in a way qps is just a vortex in space - time with the phase distribution @xmath424 described by the saddle point solution
@xmath428 substitutung the solution ( [ qpsphi ] ) into the action ( [ action11 ] ) we obtain @xmath429\ ; , \label{hydr}\ ] ] where the parameter @xmath430 sets the scale for the the hydrodynamic contribution to the qps action . here and below @xmath431
is the fine structure constant .
we also note that at @xmath432 the contribution @xmath422 ( [ hydr ] ) diverges logarithmically for infinitely long wires thus making _ single _ qps events unlikely in this limit .
let us now turn to the core contribution to the action of a single qps . in order to exactly evaluate this contribution
it is necessary to explicitly find the qps saddle point of the full non - linear effective action .
this is a formidable task which can hardly be accomplished in practice . on the other hand ,
this task is greatly simplified if one is aiming at estimating the term @xmath419 up to a numerical prefactor of order one .
below we will recover the full microscopic expression for the core contribution @xmath419 leaving only this numerical prefactor undetermined . in this way we fully capture all essential physics of qps .
the dimensionless prefactor can be regarded as a fit parameter which can be extracted , e.g. , from the comparison with available experimental data .
the above strategy allows us to approximate the complex order parameter field inside the qps core by two simple functions which should satisfy several requirements .
the absolute value of the order parameter @xmath433 should vanish at @xmath10 and @xmath425 and coincide with the mean field value @xmath102 outside the qps core .
the phase @xmath424 should flip at @xmath10 and @xmath425 in a way to provide the change of the net phase difference across the wire by @xmath427 . on top of that , in a short wire and outside the qps core the phase @xmath17 should not depend on the spatial coordinate in the zero bias limit .
all sufficiently smooth functions obeying these requirements can be used to estimate @xmath434 . for concreteness , in what follows we will choose @xmath435 for the amplitude of the order parameter field and @xmath436 for its phase .
rewriting the action ( [ action11 ] ) in the space - time domain @xmath437 ( where we dropped unimportant terms @xmath438 ) and substituting the trial functions ( [ deltaqps ] ) , ( [ phiqps2 ] ) into the action ( [ action21 ] ) one finds @xmath439 \frac{x_{0}}{\tau_{0}}+ a_{3 } sn_{0}d\delta_{0}\frac{\tau_{0}}{x_{0}}\nonumber\\ & + \
, a_{4}\frac{s\sigma}{2e^{2}\delta_{0}}\frac{1}{x_{0}\tau_{0 } } + a_{5 } sn_{0}\delta^{2}_{0 } x_{0}\tau_{0}+ a_{6 } \frac{\tilde c}{e^{2}\tau_{0}},\label{action1111}\ ] ] where @xmath440 are numerical factors of order one which depend on the precise form of the trial functions , @xmath441 is the total capacitance of the wire and @xmath350 is the wire length .
note that fictitious divergencies emerging from a singular behavior of the function ( [ phiqps2 ] ) at @xmath442 and @xmath443 are eliminated since the order parameter vanishes in this space - time point .
let us first disregard capacitive effects neglecting the last term in eq .
( [ action1111 ] ) . minimizing the remaining action with respect to the core parameters @xmath416 , @xmath417 and making use of the inequality @xmath444
, we obtain @xmath445 these values provide the minimum for the qps action , and we find @xmath446 substituting the drude expression for the normal conductance of our wire @xmath176 into eqs .
( [ xtau0 ] ) and ( [ s321 ] ) we arrive at the final results for the core parameters @xmath447 and for the core action @xmath448 here @xmath449 is the numerical prefactor @xmath450 is the total normal state wire resistance , @xmath451 k@xmath57 is the `` superconducting '' resistance quantum , @xmath269 is the superconducting coherence length and @xmath452 is the dimensionless normal conductance of a wire segment of length @xmath49 .
as it was already pointed out , the results ( [ xtau ] ) and ( [ otvet ] ) hold provided the capacitive effects are small .
this is the case for relatively short wires @xmath453 in the opposite limit the same minimization procedure of the action ( [ action1111 ] ) yields @xmath454 the qps core action then takes a somewhat more complicated form @xmath455 where @xmath456 is again a numerical prefactor . eqs .
( [ qpsac ] ) , ( [ hydr ] ) , ( [ otvet ] ) and ( [ otvet2 ] ) provide complete information about the action for single qps in diffusive superconducting nanowires .
let us analyze the above expressions .
introducing the number of conducting channels in the wire @xmath457 , setting @xmath458 and making use of the condition @xmath459 satisfied for typical metals , one can rewrite the inequality ( [ short ] ) in a very simple form @xmath460 to give an idea about the relevant length scales , for typical values @xmath461 nm and @xmath462 according to eq .
( [ short2 ] ) the wire can be considered short provided its length does not exceed @xmath463 @xmath35 m .
this condition is satisfied in a number of experiments , e.g. , in refs . @xcite . on the other hand , in experiments
@xcite much longer wires with lengths up to @xmath464 @xmath35 m were studied .
apparently such samples are effectively in the long wire regime @xmath465 . in the case of short wires
we observe a clear separation between different fluctuation effects contributing to the qps action : fluctuations of the order parameter field and dissipative currents determine the core part ( [ otvet ] ) while electromagnetic fluctuations are responsible for the hydrodynamic term ( [ hydr ] ) . in the case of longer wires
capacitive effects also contribute to the core part ( [ otvet2 ] ) .
we also observe that in the short wire limit different contributions to the qps action depend differently on the wire thickness ( or the number of conducting channels @xmath466 ) : the core part ( [ otvet ] ) decreases linearly with the wire cross section , @xmath467 , whereas the hydrodynamic contribution shows a weaker dependence @xmath468 . in the long wire limit
the dependence of the core part ( [ otvet2 ] ) on @xmath466 also becomes weaker , @xmath469 , due to capacitive effects .
yet another important observation is that in the interesting range of wires thicknesses @xmath470 nm the core part @xmath419 usually exceeds the hydrodynamic term @xmath471 .
e.g. for @xmath472 we obtain @xmath473 .
setting @xmath474 nm and estimating @xmath35 and @xmath419 for typical system parameters @xmath475 nm , @xmath476 nm , @xmath477 nm , @xmath478 nm , @xmath479 m / s and @xmath480 k we find @xmath481 and @xmath482 .
the latter inequality becomes even stronger for thicker wires .
note that the condition @xmath482 allowed us to ignore the hydrodynamic part of the qps action while minimizing the core part with respect to @xmath416 and @xmath417 . at the first sight
the result ( [ otvet ] ) derived in the short wire limit could create an illusion that our microscopic description would not be needed in order to recover the correct form of the core part @xmath419 .
indeed , the same form could be guessed , e.g. , from an oversimplified tdgl - based approach or just from the `` condensation energy '' term ( proportional to @xmath483 ) without taking into account dissipative effects .
for instance , minimization of the contribution @xmath484 ( the last three terms in eq .
( [ action21 ] ) ) is formally sufficient to arrive at the correct estimate @xmath485 .
the same equation demonstrates , however , that not only the amplitude but also the phase fluctuations of the order parameter field provide important contributions to the qps action .
if the latter fluctuations were taken into account _ without _ including dissipative effects ( this would correspond to formally setting @xmath486 in eq . ( [ action21 ] ) ) minimization of the core action would immediately yield the meaningless result @xmath487 , @xmath488 ( cf . , eq .
( [ xtau0 ] ) ) implying that the hydrodynamic contribution @xmath422 could not be neglected in that case .
minimization of the total action @xmath418 would then yield the estimate for the action parametrically different from that of eq .
( [ otvet ] ) . on the other hand , in the strong damping limit ( approached by formally setting @xmath489 in eq .
( [ action21 ] ) ) the core size would become very large and the action @xmath419 ( [ s321 ] ) would diverge meaning that no qps would be possible at all
. these observations clearly illustrate crucial importance of dissipative effects . under the condition @xmath490 ( usually well satisfied in metallic nanowires ) dissipation
plays a dominant role during the phase slip event , and the correct qps core action _ can not _ be obtained without an adequate microscopic description of dissipative currents flowing inside the wire . only employing the drude formula for the wire conductivity
@xmath176 enables one to recover the correct result ( [ otvet ] ) whereas for some other models of dissipation different results for the core action would follow , cf .
, e.g. , ref .
@xcite . to complete our discussion of the qps action
let us recall that in the course of our derivation we employed two approximations : ( i ) we expanded the action up to the second order in @xmath491 and ( ii ) in eq .
( [ action11 ] ) we expanded the action ( [ a116 ] ) in powers of @xmath492 and @xmath493 . the approximation ( i )
is sufficient everywhere except inside the qps core where @xmath374 is small . in these space- and time - restricted regions
one can expand already in @xmath374 and again arrive at eq .
( [ a116 ] ) with @xmath494 and with all the @xmath181-functions defined in appendix a3 with @xmath495 .
both expansions match smoothly at the scale of the core size @xmath496 , @xmath497 .
hence , the approximation ( i ) is sufficient to derive the correct qps action up to a numerical prefactor of order one .
the approximation ( ii ) is sufficient within the same accuracy .
one can actually avoid this approximation and substitute the trial functions ( [ deltaqps ] ) , ( [ phiqps2 ] ) directly into the action ( [ a116 ] ) .
neglecting capacitive effects in the limit ( [ short ] ) one can rewrite the qps action as a function of the dimensionless parameters @xmath498 and @xmath499 only . making use of the general results for the @xmath205-functions collected in appendix a3 and minimizing the qps action with respect to @xmath416 and @xmath417 one again arrives at the result ( [ otvet ] ) with @xmath500 .
if the inequality ( [ short ] ) is violated , the accuracy of our expansion in powers of @xmath501 can only become better ( cf .
( [ cap ] ) ) .
we now proceed further and evaluate the qps rate @xmath502 .
provided the qps action is sufficiently large @xmath503 this rate can be expressed in the form @xmath504 the results for the qps action derived above allow to determine the rate @xmath502 with the exponential accuracy . here
we evaluate the pre - exponential factor @xmath373 in eq .
( [ gamma ] ) . for this purpose
we will make use of the standard instanton technique @xcite .
consider the grand partition function of the wire @xmath505 . as we already discussed this function
can be expressed via the path integral @xmath506 which will be evaluated the saddle point approximation .
the least action paths @xmath507 determine all possible qps configurations . integrating over small fluctuations around all qps trajectories
one represents the grand partition function @xmath505 in terms of infinite series where each term describes the contribution of one particular qps saddle point .
provided interaction between different quantum phase slips is sufficiently weak one can perform a summation of these series in a straightforward manner with the result @xmath508 where @xmath509 defines the wire free energy @xmath510)}{\int{\mathcal{d}}\delta y\exp(-\delta^{2}s_{0}[\delta y ] ) } \exp ( -s_{qps } ) \nonumber\\ & \equiv & f_{0 } -\frac{\gamma_{qps}}{2}.\label{freeinst}\ ] ] here @xmath511 is the free energy without quantum phase slips , @xmath512 describe fluctuations of relevant coordinates ( fields ) , @xmath513 $ ] are the quadratic in @xmath514 parts of the action , and the subscripts `` 0 '' and `` 1 '' denote the action respectively without and with one qps . the integrals over fluctuations in eq .
( [ freeinst ] ) can be evaluated exactly only in simple cases . technically such a calculation can be quite complicated even if explicit analytical expressions for the saddle point trajectories are avalilable . in our case
such expressions for the qps trajectories are not even known .
hence , an exact evaluation of the path integrals in eq .
( [ freeinst ] ) is not possible .
furthermore , any attempt to find an explicit value for such a prefactor would make little sense simply because the numerical value of @xmath262 in eq .
( [ otvet ] ) is not known exactly .
what to do in this situation ? below we will present a simple approach which allows to establish the correct general expression for the pre - exponent @xmath373 up to an unimportant numerical prefactor
. our approach may be useful not only in the case of superconducting wires but for various other situations since numerical prefactors in the pre - exponent are usually of little importance . in order to evaluate the ratio of the path integrals in eq .
( [ freeinst ] ) let us introduce the basis in the functional space @xmath515 in which the second variation of the action around the instanton @xmath516 $ ] is diagonal . here
the basis functions depend on a general vector coordinate @xmath517 which is simply @xmath518 in our case .
the first @xmath519 functions @xmath520 are the so - called `` zero modes '' reflecting the instanton action invariance under shifts in certain directions in the functional space . in our case
the problem has two zero modes corresponding to shifts of the qps position along the wire and in imaginary time , i.e. @xmath521 . obviously such shifts do not cause any changes in the instanton energy .
the eigenfunctions corresponding to these zero modes are : @xmath522 , where @xmath523 the instanton ( or qps ) trajectory , @xmath524 and the number of zero modes @xmath519 coincides with the dimension of the vector @xmath517 .
an arbitrary fluctuation @xmath525 can be represented in terms of the fourier expansion @xmath526 then we get @xmath527&=&\frac{1}{2}\sum\limits_{k , n=1}^{\infty } a_{kn}u_{k}u_{n}^ { * } , \nonumber\\ \delta^{2}s_{1}[\delta y]&=&\frac{1}{2}\sum \limits_{k = n+1}^{\infty } \lambda_{k } |u_{k}|^{2},\label{des}\end{aligned}\ ] ] where for @xmath528 the fourier coefficients @xmath529 are just the shifts of the instanton position along the @xmath530th axis and @xmath531 are the eigenvalues of @xmath532 $ ] . integrating over the fourier coefficients
one arrives at the standard formula for the ratio of determinants with excluded zero modes @xcite @xmath533 ) } { \int{\mathcal{d}}\delta y\exp(-\delta^{2}s_{0}[\delta y ] ) } = \int\limits_{0}^{l_{1 } } d\delta x_{1 } .. \int\limits_{0}^{l_{n } } d\delta x_{n } \nonumber\\ & & \times\ , \sqrt{\frac{\det a_{kn}}{(2\pi)^{n}\prod\limits_{k = n+1}^{\infty}\lambda_{k } } } , \label{detex}\end{aligned}\ ] ] where @xmath534 is the system size in the @xmath530th dimension .
now we will argue that with a sufficient accuracy in the latter formula one can keep the contribution of only first @xmath519 eigenvalues . indeed , the contribution of the `` fast '' eigenmodes ( corresponding to frequencies and wave vectors much larger than the inverse instanton size in the corresponding dimension ) is insensitive to the presence of an instanton . hence , the corresponding eigenvalues are the same for both @xmath535 and @xmath536 and just cancel out from eq .
( [ detex ] ) .
in addition to the fast modes there are several eigenmodes with frequencies ( wave vectors ) of order of the inverse instanton size . the ratio between the product of all such modes for @xmath536 and the product of eigenvalues for @xmath535 with the same numbers is dimensionless and may only affect a numerical prefactor which is not interesting for us here .
dropping the contribution of all such eigenvalues one gets @xmath533 ) } { \int{\mathcal{d}}\delta y\exp(-\delta^{2}s_{0}[\delta y ] ) } \approx \int\limits_{0}^{l_{1 } } d\delta x_{1 } .. \int\limits_{0}^{l_{n } } d\delta x_{n } \nonumber\\ & & \times\ , \sqrt{\frac{\det a_{kn}|_{k , n\leq n}}{(2\pi)^{n}}}.\label{appz}\end{aligned}\ ] ] what remains is to estimate the parameters @xmath537 for @xmath524 . for this purpose
let us observe that the second variation of the action becomes approximately equal to the instanton action , @xmath538 , when the shift in the @xmath530th direction becomes equal to the instanton size in the same direction @xmath539 .
then we find @xmath540 and @xmath541 finally , combining eqs .
( [ gamma ] ) , ( [ freeinst ] ) , ( [ appz ] ) and ( [ prod22 ] ) we obtain @xmath542 here @xmath543 is an unimportant numerical prefactor .
this formula demonstrates that the functional dependence of the pre - exponent can be figured out practically without any calculation .
it is sufficient to know just the instanton action , the number of the zero modes @xmath519 and the instanton effective size @xmath544 for each of these modes .
in fact , a similar observation has already been made @xcite in the case of local lagrangians equal to the sum of kinetic and potential energies .
here we demonstrated that the result ( [ ffinal ] ) can be applied to even more general effective actions , including nonlocal ones .
turning to the interesting for us case of qps in superconducting nanowires we set @xmath545 , @xmath546 .
( [ ffinal ] ) yields @xmath547 this equation provides an accurate expression for the pre - exponent @xmath373 up to a numerical factor @xmath543 of order one .
this result is parametrically different from the resuls derived within the tdgl - type of analysis @xcite or suggested phenomenologically in ref .
the inequality @xmath548 allows to substitute @xmath419 ( [ otvet ] ) instead of @xmath418 in eq .
( [ b ] ) . substituting also @xmath549 and @xmath550 , for the qps rate we finally obtain @xmath551 this result concludes our calculation of the tunneling rate for single quantum phase slips .
finally we would like to emphasize that the method employed here works successfully in various other problems described both by local and nonlocal in time lagrangians .
several examples of such problems are discussed in ref .
@xcite . here
we mention only one such example .
it is the well known problem quantum tunneling with dissipation @xcite . in the limit of strong dissipation quantum decay of a metastable state
was treated by larkin and ovchinnikov @xcite who found the exact eigenvalues and , evaluating the ratio of the determinants , obtained the prefactor in expression for the decay rate @xmath552 , where @xmath553 is an effective friction constant and @xmath554 is the particle mass .
this result would imply that the pre - exponential factor in the decay rate @xcite should be very large and may even diverge if one formally sets @xmath555 .
later it was realized @xcite that this divergence should be regularized by means of proper renormalization of the bare parameters in the effective action .
after that the high frequency contribution to the pre - exponent is eliminated and one arrives at the result @xcite @xmath556 which does not contain the particle mass @xmath554 at all .
this result allowed to fully resolve a discrepancy between theory and experiments @xcite .
note that the result @xcite is trivially reproduced from eq .
( [ ffinal ] ) : the pre - exponent @xmath373 can also be expressed in the form @xmath557 , where @xmath558 is the instanton ( bounce ) action and @xmath559 is its typical size . it is remarkable that in this particular case our approach correctly reproduces even the numerical prefactor .
although typically the hydrodynamic part of the qps action ( [ hydr ] ) can be smaller than its core part ( [ otvet ] ) the former also plays an important role since it determines interactions between different quantum phase slips .
consider two such phase slips ( two vortices in space - time ) with the corresponding core coordinates @xmath560 and @xmath561 .
provided the cores do not overlap , i.e. provided @xmath562 and @xmath563 , the core contributions are independent and simply add up . in order to evaluate the hydrodynamic part we substitute the superposition of two solutions @xmath564 satisfying the identities @xmath565 ( where @xmath566 are topological charges of two qps fixing the phase change after a wind around the qps center to be @xmath404 ) into the action ( [ action11 ] ) and obtain @xmath567 , \label{sint}\ ] ] i.e. different quantum phase slips interact logarithmically in space - time .
qpss with opposite ( equal ) topological charges attract ( repel ) each other .
the next step is to consider a gas of @xmath14 quantum phase slips . again assuming that the qps cores do not overlap we can substitute a simple superposition of the saddle point solutions for @xmath14 quantum phase slips @xmath568 into the action and find @xmath569 where @xmath570 here @xmath571 defines the distance between the i - th and j - th qps in the @xmath572 plane , @xmath573 are the qps topological charges and @xmath574 is the flux quantum . in eq .
( [ mul ] ) we also included an additional term which keeps track of the applied current @xmath25 flowing through the wire .
this term trivially follows from the standard contribution to the action @xcite @xmath575 the grand partition function of the wire is represented as a sum over all possible configurations of quantum phase slips ( topological charges ) : @xmath576 where an effective fugacity @xmath51 of these charges is related to the qps rate as @xmath577 we also note that only neutral qps configurations with @xmath578 ( and hence @xmath14 even ) contribute to the partition function ( [ partit ] ) .
this fact is a direct consequence of the boundary condition @xmath579 in the path integral for the partition function @xcite .
it is easy to observe that for @xmath580 eqs .
( [ mul ] ) , ( [ partit ] ) define the standard model for a 2d gas of logarithmically interacting charges @xmath581 .
the only specific feature of our present model as compared to the standard situation is that here the space and time coordinates are not equivalent and one can consider different limiting cases of `` long '' and `` short '' wires .
let us first consider the limit of very long wires and assume that @xmath582 .
following the standard analysis of logarithmically interacting 2d coulomb gas @xcite which is based on the renormalization group ( rg ) equations both for the interaction parameter @xmath35 and the charge fugacity @xmath51 . defining the scaling parameter @xmath583 we have @xcite @xmath584 following the standard line of reasoning
we immediately conclude that a quantum phase transition for phase slips occurs in a long superconducting wire at @xmath238 and @xmath585 this is essentially a berezinskii - kosterlitz - thouless ( bkt ) phase transtion @xcite for charges @xmath581 in space - time .
the difference from the standard bkt transition in 2d superconducting films is only that in our case the transition is driven by the wire thickness @xmath586 ( which enters into @xmath35 ) and not by temperature . in other words , for thicker wires with @xmath587
quantum phase slips with opposite topological charges are bound in pairs ( dipols ) and the @xmath588 resistance of a superconducting wire is strongly suppressed and @xmath7-dependent . this resistance tends to vanish in the limit @xmath238 .
thus , we arrive at an important conclusion : at @xmath2 a long quasi-1d superconducting wire remains in a superconducting state , with vanishing linear resistance , provided its thickness is sufficiently large and , hence , the electromagnetic interaction between phase slips is sufficiently strong , i.e. @xmath587 . on the other hand , for @xmath589 the density of free ( unbound ) quantum phase slips in the wire always remains finite , such fluctuations destroy the phase coherence ( and , hence , superconductivity ) and bring the wire into the normal state with non - vanishing resistance even at @xmath2 .
thus , another important conclusion is that superconductivity in sufficiently thin wires is _ always _ destroyed by quantum fluctuations .
the above analysis is valid for sufficiently long wires . for typical experimental parameters , however , @xmath590 ( or even @xmath591 ) , and the finite wire size needs to be accounted for . for this purpose we modify the above rg treatment in the following manner .
starting from small scales @xmath592 we increase the scaling parameter @xmath593 and renormalize both @xmath35 and @xmath51 according to rg equations ( [ btkrg ] ) . solving these equations up to @xmath594 we
obtain the renormalized fugacity @xmath595 . for larger scales
@xmath596 only the time coordinate matters and we arrive at the partition function fully equivalent to one for a ( 0d ) superconducting weak link ( josephson junction ) in the presence of quantum fluctuations of the phase .
these systems are described in details in ref .
@xcite , therefore an extended discussion of this issue can be avoided here .
we only point out that our renormalized fugacity @xmath597 is equivalent to the tunneling amplitude of the josephson phase ( normalized by the josephson plasma frequency ) , i.e. @xmath598 where @xmath599 is the action of an instanton ( kink ) in a josephson junction , @xmath600 and @xmath601 are respectively the josephson and the charging energies .
the subsequent analysis essentially depends on the presence ( or absence ) of additional dissipation in our system . in the absence of dissipation at @xmath352 the josephson junction _ always _ remains in the normal ( i.e. non - superconducting ) state since superconductivity is suppressed by quantum fluctuations of the phase and tunneling of cooper pairs is prohibited @xcite . due to eq .
( [ equival ] ) exactly the same conclusion applies for sufficiently short superconducting wires studied here . in the presence of additional dissipation , however , quantum fluctuations of the phase can be suppressed and , hence , superconductivity can be restored @xcite . of particular importance is the case of ohmic dissipation which is realized either provided the wire is shunted by a normal resistor @xmath602 , or the wire itself has a non - vanishing normal conductivity @xmath603 even far from the qps core .
the latter situation can occur , e.g. , at finite ( and not too low ) temperatures due to the presence of a sufficient number of quasiparticles above the gap or possibly also due to nonequilibrium effects .
bearing in mind that in a number of experiments with ultra - thin wires ( to be discussed below ) qps effects were observed already at sufficiently high temperatures @xmath604 it is worth to briefly address the model with ohmic dissipation here . in this case
our consideration should be modified employing a two stage scaling procedure @xcite .
at it was already explained , we first proceed with 2d rg equations ( [ btkrg ] ) up to the scale @xmath594 . for simplicity
we assume that ohmic dissipation does not significantly affect rg eqs .
( [ btkrg ] ) at such small scales .
eventually we arrive at the renormalized fugacity @xmath595 . at larger scales
the space coordinate is irrelevant and the problem reduces to that of a 1d coulomb gas with logarithmic interaction .
therefore , ( for @xmath605 ) further scaling is defined by @xcite @xmath606 where the interaction parameter @xmath607 depends on the dissipation strength being equal to either ( i ) @xmath608 or to ( ii ) @xmath609 . for @xmath610
the fugacity scales down to zero , which again corresponds to a superconducting phase , whereas for @xmath611 it increases indicating a resistive phase in complete analogy to a single josephson junction with ohmic dissipation . in the case ( ii )
the phase transition point again depends on the wire cross section @xmath32 as well as on its total length @xmath350 and the value @xmath612 .
thus , two different quantum phase transitions can occur in superconducting nanowires .
one of them is the bkt - like phase transition which is controlled by the strength @xmath35 of inter - qps electromagnetic interactions and eventually by the wire thickness .
another one is the schmid phase transition occurring in the presence of ohmic dissipation @xcite provided , e.g. , by an external shunt resistance @xmath602 .
note that this situation is somewhat reminiscent of that occurring in chains of resistively shunted josephson junctions or granular arrays @xcite where , however , schmid - like qpt is driven by local ( rather than global ) shunt resistance .
let us now turn to the calculation of the wire resistance in the presence of quantum phase slips .
we first consider the limit of long wires . at any nonzero @xmath7
such wires have a nonzero resistance @xmath613 even in the ordered phase @xmath614 .
in order to evaluate @xmath3 in this regime we proceed perturbatively in the qps fugacity @xmath51 .
since for @xmath614 quantum phase slips form closed pairs ( dipols ) and , hence , interactions between different dipols can be neglected .
for this reason it suffices to evaluate the correction @xmath366 to the wire free energy due to one bound pair of quantum phase slips with opposite topological charges .
this procedure is completely analogous to that described in details in ref .
@xcite ( see chapter 5.3 of that paper for an extended discussion )
. taking into account only logarithmic interactions within bound pairs of quantum phase slips we can easily sum up the series in eq . ( [ partit ] ) and arrive at the result @xmath615}\;,\label{fren}\ ] ] where @xmath616 .
it is convenient to first integrate over the spatial coordinate @xmath260 and take the wire length @xmath617 .
for nonzero @xmath25 the expression in eq .
( [ fren ] ) is formally divergent for @xmath618 and acquires an imaginary part i m @xmath619 after analytic continuation of the integral over the temporal coordinate @xmath311 @xcite .
this indicates a qps - induced instability of the superconducting state of the wire : the state with a zero phase difference @xmath620 decays into a lower energy state with @xmath621 . the corresponding decay rate @xmath622 is defined by eq .
( [ imf ] ) .
the rate for the opposite transition from @xmath427 to 0 ( or from 0 to @xmath623 ) which is nonzero as long as @xmath624 is defined analogously with @xmath625 .
the average voltage drop across the wire is then given by the difference between these rates @xmath626\label{vvv}\ ] ] evaluating the imaginary part of the free energy @xmath627 from ( [ imf ] ) , ( [ vvv ] ) we finally obtain @xmath628^{2\mu-2}\;,\label{volt}\ ] ] where @xmath629 is the euler gamma - function .
for the wire resistance @xmath630 this expression yields @xmath631 for @xmath632 and @xmath633 for @xmath634 .
thus , for sufficiently thick wires with @xmath614 , we expect a strong temperature dependence of the linear resistance which eventually vanishes in the limit @xmath432 indicating the superconducting behavior of the wire .
unfortunately for thinner wires with @xmath589 and at low temperatures the above simple analysis becomes insufficient because of the presence of unbound qps and the necessity to account for many - body effects in the gas of quantum phase slips .
we expect that the temperature dependence of the wire resistivity should become linear at the transition to the disordered ( i.e. non - superconducting ) phase . at @xmath635
we expect a strongly nonlinear @xmath411 characteristics @xmath636 in thick wires , and the universal power @xmath637 in thin wires at the transition into the resistive state with @xmath638
. note that in contrast to the bkt transition in 2d superconducting films , in the case of wires the jump is expected to be from @xmath639 to 1 rather than from @xmath640 to 1 .
in the case of short wires @xmath591 we again proceed in two steps . a 2d scaling analysis yields the `` global '' fugacity @xmath597
. in analogy with resistively shunted josephson junctions @xcite , the voltage drop from the imaginary part of the free energy reads @xmath641^{2\gamma-1}\;,\label{volt2}\ ] ] where @xmath642 .
this equation gives @xmath643 and @xmath644 respectively at high and low @xmath7 .
the above result is valid for @xmath610 and also for smaller @xmath607 at not very small @xmath7 @xcite . in the low temperature limit in the metallic phase
the linear resistance becomes @xcite @xmath645 or @xmath646 , i.e. @xmath26 is just equal either to the quasiparticle resistance of the wire or to the shunt resistance .
the physics of this result is exactly the same as in the case of resistively shunted josephson junctions : the superconducting channel turns out to be completely blocked by quantum fluctuations and the current can only flow through the normal resistance .
it is worthwhile to emphasize that the above results were derived for sufficiently _ uniform _ nanowires with ( almost ) constant cross - section @xmath32 .
as we already mentioned above , the very existence of quantum phase slips and/or the possibility to experimentally observe qps effects in such wires were debated at an early stage of the research . at present there already exist a large number of independent experiments by different groups providing an overwhelming evidence that qps effects essentially determine low temperature properties of ultra - thin superconducting wires .
a detailed comparison between these experimental results and the above theoretical predictions will be carried out in the next chapter .
it is quite obvious that our qps theory does not require strict uniformity of the system , it will also apply provided there exist relatively small diameter variations along the wire which are practically unavoidable in any realistic experimental situation .
the probability for qps to occur will be higher in thinner parts of the wire than in thicker ones , otherwise all the physics remains essentially the same .
this situation should be contrasted to that of chains of josephson junctions and 1d granular wires in which case quantum phase slips occur inside the junctions ( tunnel barriers ) and do not require any suppression of the superconducting order parameter .
in other words , such quantum phase slips have no core at all .
accordingly , no condensation energy is lost during such phase slip events and also dissipative effects remain insignificant ( unless josephson junctions are shunted by external normal resistors ) . note that , although the qps core physics for uniform and granular quasi-1d wires is entirely different , the behavior of these systems at larger scales can be qualitatively similar .
for instance , it is well known that the bkt - like quantum phase transition ( driven by the ratio @xmath647 ) occurs in 1d superconducting granular arrays @xcite similarly to the case of uniform wires @xcite .
also , schmid - like quantum dissipative phase transitions in arrays and chains of resistively shunted josephson junctions were studied in a great detail , see @xcite and further references therein .
one of the main conclusions reached there was that dissipative phase transitions yield _ local _ phase coherence in a long chain of junctions provided each of such junctions is shunted by an effective ohmic resistance below some value of order the quantum resistance unit @xmath648 . in contrast , at @xmath45 bkt - like quantum phase transition yields global phase coherence in josephson chains provided @xmath649 . as a result ,
one arrives at a non - trivial phase diagram with all possible combinations of local and global order @xcite .
let us also mention that more recently the models of resistively shunted superconducting granular chains were studied @xcite aiming to interpret some recent experimental data @xcite seemingly indicating that the superconducting phase transition can be driven by the global wire resistance . within this approach dissipation
is introduced phenomenologically employing an effective two - fluid model .
it remains unclear , however , what could be the physical origin of ohmic dissipation in sufficiently long and nominally uniform superconducting wires at temperatures well below the gap @xmath102 .
an alternative proposal aimed to explain recent experimental observations in relatively short nanowires @xcite is discussed below in sec . 6.3 we should also stress that even if the diameter remains strictly constant the wire can be considered uniform only at length scales exceeding the elastic electron mean free path @xmath148 whereas at the scale @xmath650 this uniformity is naturally broken by the presence of impurities and defects .
scattering of electrons at such impurities and defects plays a very important role since it breaks galilei invariance thus preventing from momentum conservation for electrons propagating along the wire @xcite . momentum non - conservation is in turn necessary for qps to occur since quantum phase slips `` unwind '' the supercurrent thus generating momentum transfer . at @xmath2 ( i.e. in the absence of phonons ) this extra momentum can not be absorbed in perfectly clean and uniform wires in which case qps would be strictly prohibited . on a formal level , this implies that the qps action @xmath418 should become large in the clean limit @xcite .
in fact , this effect can be observed within our analysis too , one just needs to formally take the limit @xmath489 in eq .
( [ s321 ] ) . for completeness ,
let us mention about several proposals concerning the nature of qpt in long superconducting nanowires with impurities .
khlebnikov and pryadko @xcite argued that this transition should occur at @xmath651 as a dissipative qpt @xcite rather than the bkt - like qpt discussed here .
accordingly , for the wire resistance they suggested @xmath652 instead of @xmath653 , cf .
( [ 2mu-3 ] ) .
@xcite , on the contrary , found qpt of the bkt universality class but claimed it to be superconductor - insulator transition ( sit ) rather than superconductor - metal transition ( smt ) as originally suggested in ref .
@xcite .
in their analysis khlebnikov and pryadko @xcite used a simplified gross - pitaevski action which is in many respects different from the microscopic effective action for superconducting wires derived in chapter 2 . within this phenomenological approach
the authors @xcite first evaluated the qps rate for a given disorder configuration and only then performed averaging over disorder .
in contrast , within our analysis averaging over disorder is performed in the effective action before introducing qps .
this latter way is appropriate for diffusive superconducting wires , e.g. , because the qps core size @xmath654 greatly exceeds the electron mean free path @xmath148 .
the condition @xmath655 allows to treat the wire as effectively homogeneous for qps and the problem acquires an extra zero mode corresponding arbitrary shifts of the qps core along the wire .
this zero mode is lacking within the analysis @xcite . and
cross section @xmath656.,width=226 ] although for homogeneous ( at scales @xmath657 ) and uniform wires qpt is of the bkt universality class and the temperature dependence of the resistance in the ordered ( superconducting ) phase is given by eq .
( [ 2mu-3 ] ) , there exist other situations in which the dependence ( [ 2mu-2 ] ) may apply at not very low @xmath7 .
consider , e.g. , a long superconducting wire with cross - section @xmath32 and length @xmath350 which is uniform everywhere except a small part ( constriction ) of length @xmath658 and cross - section @xmath659 , as shown in fig .
2 . qps would then occur easier inside the constriction than in the rest of the wire .
assuming for simplicity that @xmath660 is of order of ( or just slightly larger than ) @xmath49 , for the corresponding qps fugacity inside the constriction area we obtain @xmath661 where @xmath419 is the qps core action ( [ otvet ] ) for the wire .
as before , we can express the partition function @xmath505 in terms of series in powers of the qps fugacity ( cf .
, eq . ( [ z ] ) ) which is now equal to @xmath662 inside the constriction and to @xmath51 otherwise . in other words
, we should now deal with two different types of logarithmically interacting qps .
quantum phase slips in the wire are treated exactly as before , and we again arrive at the bkt - like qpt at @xmath663 .
qps inside the constriction interact logarithmically only in time , thus giving rise to the schmid - like qpt at @xmath651 . in this case
the quasi-1d superconducting wire with mooij - schn plasmons as elementary excitations plays the role of an effective _ dissipative _ environment for the constriction .
such qpt was discussed by hekking and glazman @xcite in the case of a narrow superconducting ring interrupted by the josephson junction . in the situation considered here
this qpt appears to be of little importance since @xmath664 , i.e. at @xmath432 bkt - like qpt occurs when quantum phase slips inside the constriction are still bound in pairs . on the other hand , at not very low temperatures and
provided @xmath665 quantum phase slips inside the constriction may give an additional contribution to the wire resistance with the temperature dependence ( [ 2mu-2 ] ) . in order to demonstrate that we modify eq .
( [ fren ] ) by taking into account the difference between qps fugacities in the wire and in the constriction .
this is trivially handled by splitting the space integrals into those over the wire and constriction areas . as a result ,
the linear resistance of the system becomes @xmath666 where @xmath667 is determined by qps inside the wire ( [ 2mu-3 ] ) , i.e. @xmath668 , while the additional contribution @xmath669 comes from qps inside the constriction .
this contribution has the form @xmath670 .
hence , at sufficiently high temperatures the constriction contribution @xmath669 can dominate over @xmath667 , in which case the dependence ( [ 2mu-2 ] ) applies .
however , at lower temperatures @xmath669 becomes irrelevant and @xmath3 again is determined by eq .
( [ 2mu-3 ] ) .
obviously , the same arguments hold in the case of several constrictions or , more generally , for some wires with strongly fluctuating cross - section . a similar scenario was recently discussed in ref .
@xcite where gaussian fluctuations of the wire thickness were considered .
finally , let us briefly address the non - superconducting phase @xmath671
. in our opinion , frequently used oversimplified approaches , like tdgl or effective luttinger liquid models , are by far insufficient in order to judge whether this disordered phase is actually metallic or insulating , i.e. whether we are dealing with smt or sit at @xmath2 and @xmath672 . also rg equations ( [ btkrg ] )
_ can not _ unambiguously resolve this issue . to illustrate this point
let us compare the behavior of josephson chains and homogeneous wires .
the qps fugacities @xmath51 of both systems obey rg equations ( [ btkrg ] ) and , hence , for @xmath673 these fugacities grow upon renormalization in both cases . in the case of josephson chains
it only implies strong fluctuations of the phases across barriers with _ no suppression _ of superconductivity inside metallic grains . in this case inter - grain tunneling of cooper pairs is prohibited by electron - electron interactions and the system is in the insulating ( coulomb blockade ) phase . on the contrary , large renormalized qps fugacity @xmath51 in the case of homogeneous wires
implies that the qps gas becomes dense , the qps cores overlap and , hence , for @xmath673 and @xmath238 superconductivity gets eventually destroyed everywhere in the wire . in other words , in this regime
quantum fluctuations drive homogeneous ultra - thin superconducting wires _ normal_. this important difference between homogeneous and granular wires was emphasized in ref .
what remains is to figure out whether such normal diffusive wires stay metallic or turn insulating at low enough @xmath7 .
few years after the work @xcite it was realized that , similarly to highly conducting tunnel barriers @xcite disordered metallic wires with large conductances also demonstrate the feature of _ weak coulomb blockade _ @xcite caused by non - trivial interplay between scattering and electron - electron interactions .
the presence of an ( exponentially small ) coulomb gap implies that disordered wires most likely turn insulating at @xmath2 .
however , in generic metallic wires with many conducting channels such a state can only be observed at exponentially low @xmath674 , while for any experimentally attainable temperature one expects a metallic behavior with a weak trend for the wire resistance to grow with decreasing @xmath7 due to coulomb blockade corrections @xcite .
interestingly , such a trend was indeed observed in recent experiments @xcite and the measured dependence @xmath3 was found to closely follow theoretical predictions @xcite .
further theoretical analysis of the role of electron - electron interactions in quasi-1d structures @xcite would be highly desirable @xcite .
it follows from our theoretical analysis that both taps and in particular qps can be observed in sufficiently thin superconducting wires , i.e. experiments aimed to observe qps - related phenomena should be performed on samples with smallest possible cross section values @xmath32 . on top of that , it is highly desirable to deal with structurally ( and chemically ) homogeneous samples in order to eliminate spatial variations of @xmath4 along the wire and rule out the effect of ( possibly existing ) constrictions and tunnel barriers .
it has been shown that structural imperfections of real quasi-1d structures ( non - uniform cross section , existence of probes and finite length effects ) might effectively mask the phenomena related to thermal or quantum fluctuations @xcite .
in addition to the above requirements , in order to provide optimal conditions for observation of fluctuation effects it is crucial to properly choose the sample material . this is obvious from the fact that , for instance , the qps core action ( [ otvet ] ) at low enough @xmath7 can be represented in the form @xmath675 where @xmath676 is the wire resistivity in the normal state .
hence , it is desirable to select superconducting materials with smaller values of @xmath4 ( or @xmath677 ) and perform experiments on sufficiently dirty nanowires with higher values of @xmath678 .
the latter requirement is in line with the well known general observation that fluctuation effects are more pronounced in dirtier systems .
this requirement also implies that the electron elastic mean free path should obey the dirty limit condition @xmath679 where @xmath680 is the bcs clean coherence length .
the parameters for various conventional superconductors are listed in table i. judging from these numbers the most suitable materials for experimental investigations of qps effects are those with higher resistivity values , i.e. @xmath681 and @xmath682:ino . on the other hand , a certain disadvantage of these materials
is that theu can be strongly inhomogeneous . in this respect , zr or ti
can be advantageous for experimental studies of quantum fluctuations . though measuring the temperature dependence of the system resistance @xmath3 in the vicinity of the transition point might seem a routine experimental task , in the case of nanostructures more care is required .
typically , @xmath3 dependencies are measured in the current - biased regime .
a standard requirement is to keep the bias current @xmath25 much smaller than the critical ( depairing ) current @xmath683 in order to avoid hysteresis effects due to overheating .
additionally , in order to stay within the linear response regime , the measuring current @xmath25 should remain smaller than the charactristic scale @xmath684 @xcite equal to few tens of na for the majority of materials ( see table i ) .
the characteristic normal state resistance of a typical metallic nanowire with diameter @xmath685 10 nm and length @xmath350 @xmath686 1 @xmath35 m is of order few k@xmath57 .
hence , in the normal state the expected voltage is at least few @xmath35__v _ _ , which is certainly not a problem to measure .
however , with decreasing temperature only slightly below @xmath4 the resistance @xmath3 drops exponentially and the measured signal quickly reduces to the _ nv _ range . employing room - temperature electronics ,
it is preferable to use ac lock - in technique in order to increase the signal - to - noise ratio .
this can be associated with some hidden problems .
one of them is that even a tiny fraction of dc component ( e.g. from the ground loop ) adds a parasitic signal @xmath687 to the valuable one @xmath688 . for this reason it is advisable to decouple the ac current source from the sample using a low - noise transformer .
[ c]|c|c|c|c|c|c|c|c|material & @xmath689 & @xmath690 & @xmath691 & @xmath692 & @xmath693 & @xmath694 & @xmath678 + & @xmath695 & @xmath696 & @xmath697 & @xmath698 & @xmath699 & @xmath700 @xmath701 & @xmath702 + w & 0.015 & - & 165 * & 0.1 & 0.002 * & 1.8 * & 5.7 + ir & 0.11 & - & 22.5 * & 1.6 & 0.017 * & 1.8 * & 5.1 + hf & 0.13 & - & 19.0 * & 1.3 & 0.020 * & 1.8 * & 35 + ti & 0.40 & - & 6.2 * & 5.5 & 0.061 * & 1.8 * & 42.0 + zr & 0.61 & - & 4.0 * & 4.7 & 0.093 * & 1.8 * & 44.0 + zn & 0.86 & 28 & 1.8 & 5.5 & 0.120 & 1.82 & 5.9 + al & 1.19 & 16 & 1.6 & 10 & 0.175 & 2.02 & 2.7 + @xmath682:ino & 2.8 & - & 0.330 & - & 0.430 * & - & 3000 + in & 3.41 & 65 & 0.360 & 28 & 0.520 & 1.74 & 8.4 + sn & 3.73 & 50 & 0.230 & 30.5 & 0.557 & 1.88 & 11.0 + moge ( film ) & 5.5 & 720 & 0.005 & 66 & 1.1 & - & 200 + pb & 7.2 & 40 & 0.090 & 80.3 & 1.365 & 1.82 & 20.7 + nb & 9.25 & 85 & 0.030 & 206 & 1.520 & 1.37 & 12.5 + table - i . material parameters of several superconductors .
the data are taken from refs .
@xcite ; for @xmath681 from refs .
@xcite and for @xmath682:ino from refs .
the numbers marked with * are estimated values .
another problem might originate from the presence of rf filters which are mandatory to protect nano - sized samples from noisy electromagnetic environments .
very often such filters are just _ rlc _ circuits shunting the rf component to the ground through a capacitor as high as few @xmath703
. such configuration might provide good results with high - ohmic systems ( e.g. tunnel structures ) . in a superconducting nanowire measured in 4-probe configuration each electrode contacting the body of the sample is typically made of the same material .
hence , the resistance of these probes also varies from few k@xmath57 down to zero over the same temperature range within the transition .
depending on particular configuration , at a certain @xmath7 the ac sample impedance might become comparable to that of the current leads through the ground , causing the current re - arrangement throughout the sample . even at rather low measuring frequencies
@xmath704 10 _ hz _ the parasitic effect might manifest itself as a non - monotonous @xmath3 dependence with unusual bumps or foot at the bottom of the transition .
additional complication might arize from the non - negligible dependence of the gain of the nanovolt pre - amplifier on the total impedance of the load . concluding this part
, an experimentalist should be extremely cautious in designing the measuring set - up for unambiguous interpretation of the data on superconducting nanowires .
structural and geometrical homogeneity of quasi-1d samples is the central question in interpretation of experimental data related to contribution of superconducting fluctuations .
theoretically one usually assumes that ( i ) the critical temperature @xmath4 of quasi-1d wires under consideration remains spatially constant , i.e. it does not vary along the wire , ( ii ) the cross section @xmath32 does not vary along the wire either and ( iii ) the measuring probes are non - invasive .
unfortunately , in realistic nanowires none of these conditions is usually well satisfied .
to which extent can these imperfections be neglected while interpreting the experimental data using models developed under the assumptions ( i)-(iii ) ?
if _ structural _ imperfections ( e.g. non - uniform chemical composition ) can alter the critical temperature , then the shape of the experimentally observed @xmath3 dependence is determined by the sequence of transitions of various parts of the wire with different local critical temperature @xmath705 .
if the degree of such inhomogeneity is not too strong , a step - like @xmath3 transition may not be observed as the variations of the critical temperature are averaged on the scale of the coherence length @xmath49 resulting in a relatively wide smooth - looking
@xmath3 dependence .
one might naively expect that working with superconducting samples fabricated of initially pure material can help to eliminate the problem .
unfortunately this is not the case .
first , properties of low - dimensional superconductors are known depend on the fabrication process details , such as thin film deposition rate , residual pressure in the vacuum chamber , material of the substrate etc .
second , even if to make an effort to keep the fabrication parameters constant , it is hard to get rid of size - dependent effects .
it is a well - known experimental fact that the critical temperature @xmath4 of thin superconducting films frequently differs from one for bulk samples .
the same tendency is observed in metallic nanostructures . in indium , aluminum and
zink @xmath4 increases with decreasing characteristic dimension @xcite . on the contrary , in lead , niobium and
moge an opposite tendency is observed @xcite .
no noticeable variations of @xmath4 were detected in tin nanowires @xcite .
the origin of this phenomenon is not clear .
there exist models predicting both suppression @xcite and enhancement @xcite of the mean field critical temperature for low dimensional superconductors .
one can take the empirical fact of variation of @xmath4 with effective diameter of a superconducting nanowire as granted .
using @xmath706 lift - off fabricated nanowires as a representative example ( figs .
[ f3 tc(size ) al ] and [ f4 r(t ) al inhjomogeneous ] ) it has been shown that size - dependent effects are extremely important for interpretation of fluctuation phenomena in quasi-1d systems @xcite . the conclusion of a quantitative analysis @xcite is rather disappointing : even chemically pure nanostructures can not be considered as sufficiently homogeneous as soon as the size dependence of the critical temperature @xmath4 comes into play .
one should study atomically homogeneous systems as single crystalline whiskers @xcite .
unfortunately , modern nanotechnology does not enable growth of high - quality quasi-1d single crystalls of arbitrary diameter made of all materials of interest .
the lithographic processes results in much lower quality samples .
the only exception are materials , such as @xmath706 , @xmath707 and @xmath708 , where the critical temperature decreases with reduction of the effective dimension ( e.g. , nanowire diameter )
. under certain conditions , experiments on lift - off fabricated nanowires made of these materials can be interpreted using fluctuation models developed for perfect 1d channels .
provided broadening of the @xmath3 dependencies is detected below the bulk critical temperature @xmath709 , the size effects can not account for the phenomenon , and presumably real physics is observed . on cross section @xmath710 for aluminum nanowires .
line is a guide for the eye .
inset : spm measured distribution of cross sections for a typical lift - off fabricated nanowire : length @xmath350 = 10 @xmath35 m and effective diameter @xmath711 75 nm @xcite . ,
width=302 ] ) correspond to simulated @xmath3 transition without taking into consideration the nodes .
diamonds ( @xmath712 ) represent the results of similar calculations with contribution of the node regions .
open circles ( @xmath713 ) denote the experimentally measured @xmath3 dependence .
simulations and experimental data are for the same wire as in the inset from the previous figure @xcite.,width=302 ] as we have already discussed in chapter 4 the taps mechanism provides non - zero voltage drop across the superconducting wire at temperatures below @xmath714 . for reference purposes let us rewrite again the expression for the @xmath411 curve derived in the presence of taps at sufficiently low bias currents @xmath25 . combining eqs .
( [ gamma222 ] ) and ( [ ivtaps ] ) and restoring some fundamental constants ( set equal to unity in our theory analysis ) we obtain @xmath715 \sinh\left[\frac{\pi \hbar i}{2ek_bt}\right ] , \ ] ] where @xmath33 is the temperature - dependent coherence length of a dirty wire , @xmath716 is the boltzmann constant , @xmath717 is the the potential barrier for taps and @xmath718 . very quickly after development of the lamh theory @xcite two experimental groups reported experiments aimed at verification of the taps model . at 70-s microfabrication technique
was not much developed . in these early studies of 1d superconductivity @xcite metallic whiskers ( fig .
[ f 5 whisker ] ) with characteristic diameter @xmath704 1 @xmath35 m and lengths up to 1 mm were utilized .
growth and basic properties of these highly anisotropic objects have been widely described in the literature @xcite . in the particular case of tin whiskers
squeeze method was typically applied for their growing @xcite .
then the crystals were literally hand picked from the ingots and positioned on substrates .
the electrodes were made either by conducting paste or epoxy @xcite , soldered by wood s metal @xcite or sn - pb alloy @xcite , or squeezed by a soft metal ( e.g. indium ) @xcite .
an obvious disadvantage of the procedure is inevitable damage of the crystal within the locus of the electric probes .
an attempt to overcome these difficulties was made by combining planarization technique with electron beam lithography @xcite ( figs .
[ f 5 whisker ] and [ f6 sog structure ] ) .
the technique appears promising , but extremely time consuming and resulting in a rather low yield of working samples .
dependence for a tin whisker @xcite .
solid symbols are the experimental data , dashed line represents the fit to the taps model @xcite.,width=340 ] an example of experimental @xmath3 dependence measured on tin whisker is shown in fig .
[ fig7 r(t ) in whisker tinkham ] .
one observes a very good quantitative agreement between the experiment @xcite and the taps model @xcite .
the superconducting transition @xmath3 is very steep : resistance of the sample drops five orders of magnitude within the temperature range @xmath307 @xmath704 1 mk below @xmath4 .
this is a consequence of rather large effective diameter values @xmath719 @xmath35 m and extremely high homogeneity of single crystals . in the case of thinner single - crystalline structures
the width of the @xmath3-curve would be larger , cf .
( [ taps r(t ) ] ) .
unfortunately , the dimensions of whiskers made of superconducting materials do not vary much .
manual manipulation of sub-1 @xmath720 is extremely time consuming and results in a low yield of suitable samples , while the dimensions of whiskers are still too large to use scanning probe ( spm ) technique well developed for manipulation of nano - sized objects , such as carbon nanotubes .
an alternative to whiskers is superconducting microcylinders encapsulated in a dielectric substrate ( e.g. glass ) fabricated using taylor - ulitovski method @xcite .
the quality of these wires is not as high as that of atomically - pure whiskers .
however , an important advantage of this approach is the possibility to fabricate very long samples .
fabrication of a wire with metal core of about few @xmath35 m and length up to a km ( ! ) is a routine work , while fine adjustment of parameters enables fabrication of sub-100 nm filaments with length up to few cm @xcite .
a remarkable feature of wires obtained by this method is that typically they are single crystals .
a number of superconducting microcylinders ( sn and in ) fabricated using taylor - ulitovski method were studied @xcite .
bigger length values of the filaments ( @xmath721 1 cm ) as compared to those for previously studied whiskers ( @xmath721 0.5 mm ) enabled observation of interesting features related to interaction of current - induced phase slips . the cross section values for microcylinders did not differ much from those for whiskers , while the homogeneity was clearly worse : no qualitatively new phenomena related to fluctuation - governed 1d superconductivity has been observed .
there were several reports on experimental studies of 1d superconductivity in lift - off - fabricated superconducting nanostructures ( wires and loops ) with the line - width much smaller than of whiskers or microcylinders @xcite .
the advantage of the method is the ability to fabricate samples with a complicated shape in a reproducible way .
typically , the metal is deposited through pmma mask using thermal or e - beam evaporation , or sputtering .
the quality of the structures is far from perfect : the majority of superconducting thin - film structures fabricated at room temperatures are polycrystalline with the grain size of about few tens of nm .
deposition of metal ( in particular low melting ones , such as tin or indium ) on a cryogenically cooled substrate might reduce the grain size .
in contrast to whiskers or microcylinders , the lift - off - fabricated superconducting nanostructures studied so far were all in the dirty limit @xmath722 .
attempts to _ quantitatively _ describe the shape of superconducting transition of these quasi-1d systems using the taps model failed : experimental curves for @xmath3 were always significantly broader than theoretical predictions .
nevertheless , it was believed that with a certain adjustment of fit parameters ( e.g. , reduction of the effective size of the phase slip center @xcite ) a reasonable agreement between the experiments and the taps model predictions could be achieved .
later it was shown that inevitable inhomogeneity of lift - off - fabricated nanostructures , the presence of node regions and finite size effects can dramatically broaden the experimentally observed dependencies @xmath3 making any comparison with the taps theory inconclusive or even impossible ( fig .
[ f4 r(t ) al inhjomogeneous ] ) @xcite .
very thin superconducting nanowires have been studied using the template decoration technique : deposition of a metal film on top of a suspended insulating molecule used as a template @xcite ( fig .
[ f8 bezryadin method ] ) .
only pseudo - four probe configuration is available with the deposited metal film from both sides of the trench serving as electrodes .
the approach enabled fabrication of superconducting nanowires ( moge , nb and @xmath682:ino ) with effective diameters down to @xmath7043 nm and lengths up to @xmath7041 @xmath35 m .
tem study revealed amorphous structure for moge and @xmath682:ino .
obtained from @xmath682:ino nanowire with width 100 nm together with the data from a similarly prepared 500 @xmath35 m wide film .
inset : a scanning electron micrograph ( sem ) of a typical device @xcite.,width=302 ] the @xmath3 dependencies in @xmath682:ino appeared to be significantly wider than it is predicted by the taps model , though no claims about the impact of the qps mechanism were made @xcite .
possibly , broadening of experimental curves @xmath3 in @xmath682:ino can be associated with inevitable inhomogeneity of the samples introduced during their fabrication .
it is well known that physical properties of @xmath682:ino low dimensional systems strongly vary by annealing @xcite .
interpretation of some experiments on _ moge _ nanowires remains under debates . in earlier reports @xcite very broad @xmath3 dependencies
were observed and associated with quantum phase slips .
however , in later works @xcite broadening of a superconducting transition in similar _ moge _ structures was interpreted within the taps model with no qps contribution ( fig .
[ f10 moge short with taps ] ) . on the other hand , later it was argued @xcite that fits were actually produced outside the applicability range of the lamh theory ( [ abe ] ) .
moreover , for the experimental parameters @xcite the strong inequalities ( [ abe ] ) are not satisfied , i.e. this theory has almost no applicability range at all , see also fig .
[ f11 taps range moge ] .
thus , employing the lamh model in order to fit the resistance curves @xmath3 for these samples is problematic .
in addition , even if one performs such fits one is bound to use fit values for the electron mean free path much larger than the wire diameter @xcite which is rather unrealistic for such structures ( fig .
[ f10 moge short with taps ] ) . at this point
let us recall that estimates for the coherence length @xmath723 both in _ moge _ thin films @xcite and wires @xcite typically yield values @xmath724 to 7 nm which translates into mean free path values @xmath148 of only few nm , i.e. much shorter than used in the fits @xcite .
we will return to possible interpretation of these experiments below in connection with qps effects .
transmission electron microscopy ( tem ) study of _ nb _ nanowires , fabricated using template decoration method , revealed polycrystalline structure with an average grain size about few nm . in order to fit the data obtained in these _ nb _ structures it has been proposed @xcite to represent the measured wire conductance @xmath725 as a sum of contributions from normal electrons above the gap and from taps , @xmath726 .
the experimental curves @xmath3 could be reasonably well fitted @xcite to this phenomenological model ( fig .
[ f12 r(t ) nb wires ] ) . still , the ability to detect taps contribution in _ nb _ is rather unexpected since the energy gap @xmath102 is very large in this material ( see table i ) .
it is also quite surprising that the inhomogeneities in granular _ nb _ nanowires do not seem to broaden the @xmath3 curves , which can be fitted by the taps model developed for homogeneous quasi-1d systems .
are 70.0 , 19.0 , 11.5 , 9.4 , 5.6 , and 6.7 nm , and lengths @xmath261 are 177 , 43 , 63 , 93 , 187 , 99 nm for samples 16 , respectively .
double - step shape of @xmath3 transitions in the top and bottom plots comes from the superconducting transition of the contact regions contributing to the 2-probe measurement configuration .
, width=302 ] = 6.7 nm , and the critical temperature , @xmath4 = 4.7 k , marked by an arrow .
the lamh theory can only be applied within the strip 4.23 k @xmath727 k @xcite ( shown in grey ) which is clearly outside the region where @xmath26 strongly depends on temperature.,width=302 ] = 5.8 , 5.6 , 2.7 , 2.5 , and 1.9 k ; lengths ( nm ) are @xmath350 = 137 , 120 , 172 , 177 and 113 ; normal state resistance ( @xmath57 ) are @xmath450 = 470 , 650 , 1600 , 4250 and 9500 ; coherence lengths ( nm ) are @xmath728= 8.5 , 8.1 , 18 , 16 , and 16.5 , respectively .
solid lines show fits to the phenomenological taps model @xcite . the dashed and dotted lines are some theoretical curves that include qps contributions.,width=302 ] let us now turn to the experiments detecting qps effects in superconducting nanowires .
mooij and coworkers @xcite discussed the possibility to observe quantum phase slips experimentally and attempted to do so as early as in 1987 . according to our theory , the crossover between taps and qps regimes can be expected at temperatures @xmath729 . for superconductors with typical material parameters and
@xmath730 1k ( see table i ) this condition implies that qps effects may become important already at @xmath731 100 mk .
as we already discussed , in order to be able to observe qps it is necessary to fabricate nanowires with effective diameters in the 10 nm range , @xmath732 10 nm .
the wires should be sufficiently uniform and homogeneous in order not to override fluctuation effects by trivial broadening of @xmath3 dependencies due to wire imperfections and inhomogeneities @xcite .
thus , proper fabrication technology is vitally important for experimental studies of qps in quasi-1d superconductors .
perhaps the first experimental indication of the effect of quantum fluctuations was obtained in amorphous moge nanowires with effective diameter down to @xmath685 30 nm @xcite . the samples in 4-probe configuration with length @xmath350 up to 20
@xmath35 m were fabricated using e - beam lithography .
though the paper @xcite was mainly focused on the effect of disorder in low - dimensional superconductors , it was clearly stated that for the narrowest samples significantly broader curves @xmath3 were observed than it is predicted by the taps model ( see fig .
[ f13 r(t ) moge graybeal ] ) .
the effect of quantum fluctuations was pointed out as a possible reason for this disagreement . upon a single film of thickness @xmath733 nm @xcite.,width=302 ] detailed experimental studies of transport properties of superconducting nanowires
have been carried out by giordano and co - workers @xcite . in these experiments _ in _ and _ in - pb _ wires with triangular cross - section were fabricated using step - edge lithographic technique @xcite .
this method utilizes the shadow effect produced by extremely shallow steps in substrates and the corresponding tilted ion beam milling of the deposited material ( fig .
[ f14 giordano step technique ] ) .
polycrystalline wires were fabricated with effective diameters @xmath586 in the range 40 - 100 nm with the grain size from 10 to 20 nm .
the uniformity of the samples was controlled with sem and with the accuracy of @xmath1 nm . ) , 50.5 nm ( @xmath734 ) and 72 nm ( @xmath713 ) .
the solid curves are fits to the taps model , while the dashed curves indicate fits to the phenomenological model @xcite .
the sample lengths are 80 , 150 and 150 @xmath35 m , and the normal state resistance values are 5.7 , 7.1 and 1.2 k@xmath57 , respectively @xcite.,width=302 ] for wider structures a reasonable agreement with the taps model was observed . on the other hand , for thinner wires with @xmath735 50 nm
clear deviations from taps predictions were demonstrated ( see fig . [
f15 giordano r(t ) in ] ) .
the discrepancy was interpreted as a manifestation of quantum phase slippage . in order to explain their observations giordano and co - workers proposed phenomenological description based on the caldeira - leggett model for macroscopic quantum tunneling with dissipation @xcite . though qualitative agreement with this simple phenomenological model has been obtained ( fig .
[ f15 giordano r(t ) in ] ) , quantitative interpretation of the data @xcite is problematic due to poor uniformity of the samples : @xmath736 10 nm for wires with @xmath735 50 nm @xcite .
recently , pai _ et al . _
@xcite performed a fit of the same experimental data @xcite to the power - law dependence @xmath737 derived from the result ( [ 2mu-3 ] ) assuming gaussian fluctuations of the wire thickness .
unfortunately the temperature dependence of the qps fugacity ( [ fug ] ) was not taken into account in ref .
this temperature dependence ( predominantly determined by that of the gap , @xmath738 ) should , however , still be significant at temperatures not very far from @xmath4 and will alter the fits .
it appears that strong granularity of the wires was most likely a very important factor in the experiments by giordano and co - workers .
for instance , estimating @xmath419 for the thinnest wires @xcite with diameters 16 and 25 nm , for the experimental parameters we obtain respectively @xmath739 and @xmath740 .
for uniform wires such huge values of the qps action would totally prohibit any signature of quantum phase slips .
since qps effects were very clearly observed , it appears inevitable that these samples contained constrictions with diameters @xmath741 times smaller than the average thickness values presented in @xcite .
another set of experiments was performed by dynes and co - workers @xcite .
suspended stencil technique has been developed ( fig .
[ f16 dynes stencil technique ] ) enabling fabrication of quench - condensed granular nanowires with cross - section area @xmath32 down to 15 nm@xmath742 and length @xmath350 ranging from 1 to 2 @xmath35 m from various materials including several superconductors , such as _ pb , sn , pb - bi_. the samples edge roughness was claimed to be about 3 nm .
a remarkable feature of the method is the ability to vary the wire thickness @xmath743 and , hence , its cross section @xmath32 at a constant width @xmath744 _ in situ _ inside the cryostat in - between the sessions of truly 4-probe @xmath3 measurements .
dependencies for the same _ pb _ nanowire with width @xmath745 22 nm and various thicknesses .
solid lines at the top of the transitions are fits to the aslamazov - larkin theory . dashed lines are fits to the taps model , which clearly yields steeper @xmath3 dependencies compared to the experimental data for the thinnest wires @xcite.,width=302 ] .,width=302 ] the experiments @xcite clearly indicated systematic deviations of the experimental data points @xmath3 from the taps model predictions ( fig . [
f17 dynes r(t ) pb ] ) .
this discrepancy increases as the wires become narrower .
the width of the superconducting transition was found to scale with the normal state resistance @xmath450 ( fig .
[ f18 dynes r(t)width pb ] ) .
it should be noted that in lead nanowires as narrow as 15 nm and as thin as 10 nm no low temperature resistance tails were observed . instead , a less dramatic but systematic broadening of the superconducting transition beyond the taps limit was noted @xcite . on the other hand ,
long resistance tails were always present in tin structures fabricated and measured using similar technique @xcite .
very probably , the discrepancy comes from the difference in the heights of the potential barrier @xmath366 between these two materials ( see table i ) : lead wires should be significantly narrower than tin ones in order to obtain a similar magnitude of fluctuation effects .
in addition , we point out that the experiments @xcite demonstrated clear superonducting transitions reaching experimental `` zero '' in nanowires with normal state resistance @xmath746 10 k@xmath57 .
this observation is not in line with the conjecture @xcite that superconductivity in _ moge _ nanowires could only be possible for wire resistances below the quantum resistance unit @xmath747 6.45 k@xmath57 .
a clear manifestation of a crucial role of qps effects in superconducting nanowires was provided in the experiments by tinkham and co - workers @xcite .
these authors developed a novel technique which allowed them to fabricate sufficiently uniform superconducting wires considerably thinner than 10 nm with lengths ranging between @xmath748 nm and 1 @xmath35 m .
this was achieved by sputtering a superconducting alloy of amorphous @xmath749 over a free - standing carbon nanotube or bundle of tubes laid down over a narrow and deep slit etched in the substrate ( fig .
[ f8 bezryadin method ] ) .
three out of eight samples studied in the experiments @xcite demonstrated no sign of superconductivity even well below the bulk critical temperature @xmath4 .
furthermore , in the low temperature limit the resistance of these samples was found to show a slight upturn with decreasing @xmath7 . in view of
that one can conjecture that these samples may actually become insulating at @xmath432 .
the resistance of other five samples @xcite decreased with decreasing @xmath7 . also for these five samples no sharp superconducting phase transition was observed .
all three non - superconducting wires ( i1,i2 and i3 ) @xcite had the normal state resistance above the quantum unit @xmath648 , while @xmath750 for the remaining five `` superconducting '' samples was lower than @xmath648 .
this observation allowed the authors @xcite to suggest that a dramatic difference in the behavior of these two groups of samples ( otherwise having similar parameters ) can be due to the dissipative quantum phase transition ( qpt ) @xcite ( see also chapter 5 ) analogous to schmid phase transition @xcite observed earlier in josephson junctions @xcite . already at this stage
we would like to emphasize that an important pre - requisite for this qpt is the existence of a source for linear ohmic dissipation _ at low energies _ which typically requires the presence of some normal shunt resistance .
it appears that no such condition was fulfilled in the experiments @xcite .
hence , the interpretation of the data @xcite in terms of a dissipative qpt is problematic . and
indeed , this interpretation was not confirmed in the later experiments of the same group @xcite who observed superconducting behavior in samples with normal resistances as high as 40 k@xmath57 @xmath751 .
the authors @xcite concluded that the relevant parameter controlling the superconducting transition is not the ratio of @xmath752 , but appears to be resistance per unit length , or equivalently , the cross - sectional area of a wire ( fig.[f19 lau r(l ) ] ) . .
( a ) linear plot .
the dotted line is a guide to the eye .
( b ) semilog plot with an exponential fit .
slope of the fitted line is 0.39 k@xmath57/ nm @xcite.,width=302 ] this conclusion clearly favors interpretation of the data @xcite either in terms of a bkt - like qpt @xcite at @xmath753 or just as a sharp crossover between the regimes of vanishingly small and sufficiently high qps rates @xmath502 which will correspond respectively to superconducting and normal behavior of the nanowires . in both cases
the crucial parameter is the wire cross section @xmath32 .
both these qpt and crossover are expected to occur for wire diameters @xmath754 nm .
let us estimate the qps core action @xmath419 defined by eq .
( [ otvet ] ) for the eight samples studied in the experiments @xcite . with the density of states
@xmath755 s / m@xmath756 , the superconducting critical temperature @xmath757 k and the measured resistivity @xmath758 @xmath759 m we obtain the coherence length @xmath760 nm in agreement with the estimate @xcite .
our estimates for the action @xmath419 are summarized in the following table ii : [ c]|c|c|c|sample & @xmath761 , k@xmath57/nm & @xmath419 + i1 & 0.122 & 7.8 @xmath262 + i2 & 0.110 & 8.7 @xmath262 + i3 & 0.079 & 12.7 @xmath262 + s1 & 0.038 & 25.1 @xmath262 + s2 & 0.028 & 33.7 @xmath262 + s3 & 0.039 & 22.6 @xmath262 + ss1 & 0.054 & 15.4 @xmath262 + ss2 & 0.044 & 19.6 @xmath262 + we observe that for @xmath500 the qps rate @xmath762 is expected to be much higher in the normal samples i1 , i2 and i3 than in five remaining wires which demonstrated the superconducting behavior in the low temperature limit .
this observation is consistent with the interpretation in terms of the crossover between the regimes with low and high qps rates .
further support for this interpretation comes from the data @xcite obtained for more than 20 different nanowires .
the resistance of these wires measured at @xmath763 k is presented in fig .
[ f19 lau r(l ) ] versus the inverse normal resistance per unit length @xmath764 . in fig .
[ f19 lau r(l ) ] ( a ) one observes a sharp crossover between normal and superconducting behavior at wire diameters @xmath765 nm .
this crossover could be interpreted as an indication to the qps - binding - unbinding qpt @xcite controlled by the parameter @xmath65 .
note , however , that all wires studied in @xcite were quite short .
hence , this qpt should inevitably be significantly broadened by finite size effects .
re - plotting the same data on a semilog scale ( fig .
[ f19 lau r(l ) ] ( b ) ) we indeed observe a rather broad distribution of measured resistances which despite some scatter can be fitted to the linear dependence on the wire cross section @xmath32 .
this fit is highly suggestive of the crossover scenario although it can not yet rule out the ( broadened by size effects ) qpt either . in order to finally discriminate between these two options
it is necessary to analyze the temperature dependence of the resistance @xmath3 . according to our theory , in the linear regime
the ( caused by qps ) wire resistance is defined by eq .
( [ 2mu-3 ] ) .
the dependence @xmath766 comes from pairs of qps events where the fugacity ( [ fug ] ) depends on temperature via @xmath738 .
an additional ( weak ) power - law dependence @xmath767 enters because of interactions between quantum phase slips in space - time . in order to fit the data
@xcite we will ignore this power - law dependence and use the simplified formula @xmath768 with @xmath419 defined in ( [ otvet ] ) and @xmath769 being an unimportant constant factor . as this formula
can only be applied to samples with vanishing low @xmath7 resistances @xmath770 , we select the data @xcite for the samples 3 to 8 seemingly demonstrating such a behavior
. the results of the fits to eq .
( [ rtt ] ) are presented in fig .
[ f20 lau moge our fits ] .
we observe a very good agreement between our qps theory and experiment which was obtained using the same value @xmath771 for all six samples .
we also note that the fits remain essentially unchanged if we take into account an additional power law dependence @xmath767 ( with @xmath35 estimated from the wire parameters ) .
this observation proves that inter - qps interaction is indeed insignificant for the interpretation of the data @xcite .
thus , similarly to lau _
@xcite , we conclude that the temperature dependence of the resistance of their ultra - thin @xmath681 wires is determined by qps effects and the observed sharp transition between normal and superconducting behavior is most likely a thickness - governed crossover between the regimes of respectively large and small qps rates @xmath772 .
dependences comes from the superconducting transition of the contact regions contributing to the 2-probe measurement configuration .
the samples normal state resistances and lengths are 1 : 14.8 k@xmath57 , 135 nm ; 2 : 10.7 k@xmath57 , 135 nm ; 3 : 47 k@xmath57 , 745 nm ; 4 : 17.3 k@xmath57 , 310 nm ; 5 : 32 k@xmath57 , 730 nm ; 6 : 40 k@xmath57 , 1050 nm ; 7 : 10 k@xmath57 , 310 nm ; 8 : 4.5 k@xmath57 , 165 nm .
symbols stand for calculations using eq .
[ rtt ] with the single numerical coefficient @xmath262=0.7 .
the critical temperature @xmath4 and the dirty - limit coherence length @xmath773 used as fitting parameters for samples 3 - 8 are 3 : 5.0 k , 8 nm ; 4 : 6.4 k , 8.5 nm ; 5 : 4.6 k , 8.9 nm ; 6 : 4.8 k , 8.9 nm ; 7 : 5.6 k , 11.9 nm ; 8 : 4.8 k , 8.5 nm .
data for samples 1 and 2 can not be fitted by any reasonable set of parameters.,width=302 ] a bulk of recent experimental data @xcite accumulated on _ moge _ nanowires fabricated using the molecular template method was recently reviewed in detail by bezryadin @xcite .
the main observations can be summarized as follows : ( @xmath774 ) `` shorter '' nanowires ( @xmath775 nm ) demonstrate either weakly insulating behavior with clear features of weak coulomb blockade @xcite , or relatively steep superconducting transition @xmath3 with virtually no samples showing an intermediate regime , ( @xmath776 ) `` longer '' samples ( 200 nm @xmath777 1 @xmath35 m ) typically showed the behavior which similarly to the data @xcite can reasonably well be interpreted in terms of a crossover between the regimes of small and large qps rates ( corresponding to respectively superconducting and normal behavior ) .
the @xmath3 curves of longer wires in the regime ( @xmath776 ) showed a decrease of the resistance with cooling no matter whether their normal state resistance @xmath450 was smaller or bigger than @xmath648 .
the crossover between normal and superconducting behavior of the samples was controlled by the wire cross section @xmath32 or , equivalently , by the ratio @xmath778 . as in ref .
@xcite , the overall picture is consistent with the qps scenario @xcite .
turning to the regime of shorter wires ( @xmath774 ) , it was argued @xcite that the transition between superconducting and normal behavior for these samples is most likely controlled by the _ total _
normal wire resistance @xmath450 rather than by the resistance per unit length @xmath778 , see fig .
this observation implies that a significant fraction of short wires with lengths @xmath775 nm showed superconducting - like behavior even though the corresponding estimated values of the qps core action could be as high as @xmath779 ( in which case for longer wires one expects strong proliferation of qps leading to total destruction of superconductivity ) .
one should , therefore , explain why these samples stay superconducting instead of being normal or even insulating in the limit of low enough @xmath7 .
an empirical fact that short samples @xcite stayed normal for @xmath780 and turned superconducting as soon as @xmath781 allowed the authors to assume the presence of some kind of qpt that could , for instance , be similar to schmid dissipative phase transition in josephson junctions @xcite .
thus , in some sense the authors @xcite revived an earlier idea @xcite , now in application to short wires only .
supporting this idea meidan _ et al .
_ @xcite suggested a phenomenological scheme essentially employing the well known rg analysis developed for resistively shunted josephson junctions @xcite .
unfortunately , as was also noticed in refs .
@xcite , the experimental parameters correspond to large fugacity values @xmath782 , i.e. the rg approach @xcite was employed in ref . @xcite
well beyond its applicability range .
another even more significant problem is that in the structures @xcite no source for linear ohmic dissipation seems to exist at low energies .
et al . _ phenomenologically argued that such dissipation could occur due to proliferation of qps inside the wire ( which would make the wire to a large extent normal ) .
however , even _ intrinsically normal _ wires in contact with superconducting electrodes develop a _ proximity - induced gap _ in its energy spectrum @xcite of order thouless energy @xmath783 .
hence , ohmic dissipation can hardly occur in such sns structures at energies @xmath784 .
since for short samples @xcite with @xmath785 nm the energy @xmath786 can easily reach few k , it is hard to expect that a dissipative qpt can occur in such systems .
the same arguments suggest that the proximity effect may actually play an important role in the interpretation of the experimental data @xcite for short samples .
even if the wire is intrinsically normal , supercurrent can flow through such a weak link due to the proximity effect .
accordingly , superconductivity in the short samples with @xmath781 can simply be associated with the onset of dc josephson current in an sns junction .
the corresponding scenario is as follows . in the interesting limit @xmath787 the josephson energy @xmath788 of an sns junction @xcite
is exponentially small at high temperatures @xmath789 .
hence , at such @xmath7 we have @xmath790 and the weak josephson current is fully suppressed by thermal fluctuations . upon decreasing temperature
one eventually reaches the regime @xmath791 , in which case @xmath600 becomes @xcite @xmath792 thus , for @xmath781 at @xmath793 we have @xmath794 , i.e. thermal fluctuations are negligible and the sns junction becomes superconducting . in the vicinity of @xmath795
this argument also allows to predict exponential decrease of the wire resistance @xmath796 in a qualitative agreement with exponential dependencies observed in @xcite .
on the other hand , for @xmath780 at @xmath795 we still have @xmath797 , i.e. supercurrent is disrupted by thermal fluctuations down to lower @xmath7 and significantly broader @xmath3 curves are expected . at lower @xmath7 thermal fluctuations become unimportant but , on the other hand , quantum fluctuations take over destroying the josephson current for @xmath798 , where @xmath799 is an effective charging energy of an sns junction . thus , in the low temperature limit one would expect to see the crossover at @xmath800 or , equivalently , at @xmath801 now , similarly to the case of josephson junctions @xcite , one can demonstrate that the capacitance @xmath60 is given by the expression @xmath802 , where @xmath803 is the junction geometric capacitance and @xmath804 is the renormalization term @xmath805 .
since in our case geometric capacitance is most likely very small @xmath806 , one has @xmath807 . substituting this estimate into eq .
( [ crov ] ) , one immediately obtains the crossover condition @xmath808 in agreement with experimental findings .
we believe that these simple arguments are sufficient to understand a superconducting - to - normal crossover observed in short wires @xcite .
these arguments also demonstrate that for such samples josephson physics ( both classical and quantum ) appears to be more important than that of taps and qps . hence , although short wires can also demonstrate interesting phenomena , longer wires appear to be more suitable for experimental investigations of qps effects .
now let us turn to other experiments where qps in superconducting nanowires have been observed . the original method of forcing the molten alloy into a porous media @xcite or capillary @xcite enables fabrication of metallic wires with diameters @xmath8091 @xmath35 m .
at these early stages of research no studies of the shape of superconducting transition were performed in such structures .
chemical or electrochemical methods developed later allowed filling of very narrow pores with a wide variety of materials including superconductors : @xmath810 @xcite , @xmath811 @xcite and @xmath707 @xcite .
the most widely used media are tracketched membranes and alumina nanoporous films @xcite .
the diameter of pores in these materials can be rather uniform ranging from @xmath70415 nm to few @xmath35 m .
one side of the membrane is coated with a thick metal film serving as a cathode .
the `` sandwich '' is immersed into a corresponding electrolyte .
standard three - terminal reduction of ions in the solution leads to growth of nanowires from the cathode .
typically the nanowires are polycrystalline . however , with properly adjusted electrodeposition conditions nanowires can be made crystalline @xcite .
the method enables fabrication of large amounts of almost vertically aligned nanowires of desired morphology and diameters .
electric contacts to a bundle of nanowires can be made easily connecting the thick bottom cathode ( from which the filaments start to grow ) and the top metal plate on top of the nanopore membrane .
the number of nanowires contacting metallic electrodes depends on squeezing pressure of the top electrode and the filling factor of the membrane @xcite . for unambiguous interpretation of the data it is desirable to perform electric measurements on a single nanowire .
if the density of the nanopores is sufficiently high , a possible solution is to dissolve the hosting membrane , isolate a single nanowire and fabricate contacts using conventional lithographic technique @xcite .
a nanoindentation - based method has been proposed enabling electric measurements on a single nanowire inside the hosting membrane @xcite .
dependence for four pb nanowires ( a ) - ( d ) with diameters 40 , 55 , 55 and 70 nm , correspondingly .
the transition is broader for narrower wires @xcite.,width=340 ] curves for 20 , 40 , 60 , 70 , and 100 nm wide and 6 @xmath35 m long sn nanowire arrays containing respectively 18 , 1 , 8 , 15 and 53 wires in the bundle .
the solid lines for 20 , 40 and 60 nm wires are the results based on the taps model near @xmath4 and qps model below @xmath4 @xcite.,width=302 ] experiments with _ zn _ nanowires @xcite lead to observations of unexpected `` antiproximity '' effect .
the shape of the supeconducting transition @xmath3 of a bundle of _ zn _ nanowires with length up to few @xmath35 m depends on the material of the electrodes .
it was found that superconductivity is completely or partially suppressed in samples with superconducting electrodes , while clear superconducting transition has been observed in systems with normal electrodes .
the origin of the phenomenon is not yet clear , while a model has been proposed @xcite .
experiments with individual _ pb _ nanowires grown using nanopore electrochemical method show clear superconducting transitions , which get broadened with reduction of the sample diameter ( fig .
[ f21 piraux r(t ) pb ] ) .
unfortunately no theory fits was provided by the authors , while the linear scale of the reported @xmath3 dependencies complicates an independent quantitative comparison @xcite .
experiments on bundles of several _ sn _ nanowires show clear broadening of the @xmath3 transitions in the narrowest samples ( fig .
[ f22 tian r(t ) sn ] ) , which was associated with qps effects @xcite .
as the number of wires in the measured bundle is not known precisely , quantitative comparison with theoretical predictions appears complicated .
in addition , the experiments were presumably performed outside the linear regime : only rather high dc excitation current densities @xmath812 10@xmath813 a / cm@xmath742 were used being just slightly below the experimentally measured critical current density at low @xmath7 , @xmath815 10@xmath816 a / cm@xmath742
. an advanced method of nanowire fabrication uses mbe grown _ inp _ layer on a cleaved _ _ in__@xmath817__ga__@xmath818_as / inp _ substrate as a support for thermally evaporated metal ( fig .
[ f23 altomare method ] ) .
the applicability of the approach was demonstrated on _ aupd _ and _ al _ @xcite .
the method allows fabrication of very long ( up to 100 @xmath35 m ) superconducting ( _ al _ ) nanowires with effective diameters down to 7 nm .
the approach enables pseudo-4-terminal measurements ( fig .
[ f24 altomare r(t ) al ] ) .
the observed broad @xmath3 dependencies were associated with qps .
though rather scarce experimental data make quantitative conclusions on the qps mechanism difficult , the experiments on long _ al _ nanowires @xcite have a clear message : no correlations between the total normal state wire resistance @xmath450 ( compared to the quantum resistance unit @xmath819 k@xmath57 ) and superconductivity in such wires was found .
0 ) crystallographic plane of the two pieces and a pattern is written using standard ebl and then developed , ( c ) thermal evaporation is used to deposit a film of ti / au ( the portion of the film deposited on the pmma is not shown ) and then the two halves are separated , ( d ) after lift - off and oxygen plasma etching , wet etching is used to define the inp ridge , ( e ) the wire is formed through the final evaporation appropriately masking the substrate of the sample ( top : side view of the evaporation arrangement , bottom : final result ) @xcite.,width=642 ] dependencies for al nanowires . at temperatures @xmath8201 k ,
samples s2 100 @xmath821 m long , @xmath450 = 86 k@xmath57 ( @xmath822 ) and s1 : 10 @xmath821 m long , @xmath450 = 8.3 k@xmath57 ( @xmath823 ) became superconducting .
the insets show sem images and schematic of a typical device containing a 10 @xmath821 m long , nominally 8 nm wide al nanowire ( similar to samples s1 and s2 ) : the al layer does not entirely cover the au / ti pads which were measured in series with the superconducting wire ( in the main panel this series resistance has been subtracted ) @xcite.,width=302 ] a convenient material to study the phenomena associated with phase slips is aluminum .
its bulk critical temperature @xmath824 k is relatively low ( see table i ) , hence , the qps rate should be comparatively high enabling pronounced manifestation of fluctuation effects .
an additional useful feature of aluminum is its peculiar size dependence of @xmath4 .
although the origin of this effect remains unclear , an increase of @xmath4 with reduction of the characteristic dimension of aluminum structures ( wire diameter or film thickness ) is a well - known experimental fact and can be taken as granted @xcite .
this effect does not allow to interpret broadening of @xmath3 dependencies at temperatures @xmath825 in terms of sample inhomogeneities , such as constrictions .
aluminium was chosen for investigations of 1d superconductivity in refs .
it was demonstrated that low energy @xmath826 ion sputtering can progressively and non - destructively reduce dimensions of various nanostructures including nanowires @xcite .
the penetration depth of @xmath826 ions into _ al _ matrix at acceleration voltages of @xmath704 500 ev is about 2 nm and is comparable to the thickness of naturally formed oxide .
the accuracy of the effective diameter determination from the normal state resistance by sem and spm measurements is about @xmath736 2 nm .
only those samples which showed no obvious geometrical imperfections were used for further experiments . to a large extend the method allows one to study the evolution of the size phenomenon , eliminating artifacts related to uniqueness of samples fabricated in independent processing runs .
the ion beam treatment polishes the surface of the samples removing inevitable roughness just after fabrication @xcite ( fig .
[ f25 al nanowire sputtering ] ) .
if there were no detectable geometrical imperfections in the original ( thick ) wires , they could not be introduced in the course of diameter reduction by low energy ion sputtering . after a sequence of sputterings ( alternated with @xmath3 measurements ) the wire diameter was reduced from @xmath827 nm down to @xmath586 @xmath828 10 nm .
experiments were performed on several sets of aluminum nanowires with length @xmath350 equal to 1 , 5 and 10 @xmath35 m . for larger diameters
@xmath829 20 nm the shape of the @xmath3 dependence is rather `` sharp '' and can be qualitatively described by the taps mechanism .
note that the abovementioned size dependent variation of @xmath4 in aluminum nanowires results in broadening of the @xmath3 transition and significantly reduces applicability of the taps model @xcite ( fig .
[ f4 r(t ) al inhjomogeneous ] ) .
when the wire diameter is further reduced , deviations from the taps behavior become obvious ( fig .
[ f26 r(t ) al zgirski prb 2008 ] ) .
fits to the taps model fail to provide any reasonable quantitative agreement with experiment for diameter values below @xmath830 20 nm even if one hypothetically assumes the existence of unrealistically narrow constrictions not observed by spm . and
, as we already discussed , broadening of the @xmath3 dependencies in aluminum nanowires at @xmath825 can hardly be ascribed to geometrical imperfections , such as constrictions .
we conclude that the most natural interpretation of our observations is associated with quantum fluctuations .
curves for the thinnest samples obtained by progressive diameter reduction for the same aluminium nanowire with length @xmath350 = 10 @xmath697 .
the taps model fitting is shown with dashed lines for 11 and 15 nm samples with the best fit mean free path @xmath831 3 and 10 nm , correspondingly , @xmath832 1.46 k and critical magnetic field @xmath833 10 mt
. fits to eq .
( [ rtt ] ) are shown by solid lines . for 11 , 12 , 13 and 15 nm wires the fit parameters are : @xmath834 ; @xmath689 : 1.64 k , 1.52 k , 1.47 k , and 1.47 k ; mean free path @xmath148 : 7.5 nm , 8.2 nm , 9.5 nm and 9.5 nm ; normal state resistance @xmath450 : 7200 k@xmath57 , 5300 k@xmath57 , 4200 k@xmath57 and 2700 k@xmath57 @xcite.,width=302 ] let us now perform a detailed comparison of the data with theoretical predictions @xcite discussed in chapter 5 . to begin with
, we should select a proper formula for the current - voltage characteristics of the wire which can be applied to the samples @xcite .
one can employ the general formula ( [ volt ] ) derived for sufficiently long wires .
it remains not completely clear if this assumption is well satisfied for the wires used in experiments @xcite , though it appears that at least the longest wires with @xmath835 @xmath35 m are already in this regime ( @xmath836 in this case ) .
in addition , one can recall that in the experimental temperature range not far from @xmath4 there can still be sufficiently many quasiparticles above the gap which can produce additional dissipation . although it would not be fully justified to use eq .
( [ volt2 ] ) in this case , still dissipation can result in additional inter - qps interaction which can slightly modify the power - law dependencies ( [ 2mu-3 ] ) , ( [ 2mui ] ) .
fortunately , this effect , even if exists , does not change our fitting procedure . in order to describe the qps contribution to the wire resistance
we will use the formula @xmath837 while for the non - linear ( current - dependent ) wire resistance we will use @xmath838 here @xmath839 is an unimportant constant which remains the same for all samples .
let us again remind the reader that the dependence @xmath840 comes from pairs of qps events . as we already discussed , the fugacity ( [ fug ] ) depends on temperature via @xmath841 .
an additional power - law dependence enters because of inter - qps interaction .
the parameter @xmath842 can be equal to @xmath843 ( as in eqs .
( [ 2mu-3 ] ) , ( [ 2mui ] ) ) or take somewhat different values due to dissipative contribution . below we will just use it as a fit parameter instead of the parameter @xmath35 .
we first determine this parameter by fitting the @xmath411 curves at a given temperature to the dependence ( [ b2 ] ) , see fig .
[ f27 i - v al prb 2008 ] . from these fits
we obtain the values for parameter @xmath842 ( and , correspondingly , for @xmath35 ) for each sample .
now we perform the fits of the resistance data @xmath3 taken in the linear regime @xmath844 to the dependence ( [ b1 ] ) .
there are four parameters to specify for our fits : critical temperature @xmath4 , normal state resistance @xmath450 , electron elastic mean free path @xmath148 ( used to re - calculate the dirty limit coherence length @xmath845 ) , and the numerical factor @xmath262 of order unity .
the critical temperature and the normal state resistance can be trivially deduced from the experimental data for @xmath3 .
the electron mean free path @xmath148 can be roughly estimated from the normal state resistivity @xmath846 as the product @xmath847m@xmath848 is a well - tabulated value for dirty aluminum . since the cross section for ultra - narrow nanowires is known from spm measurements within @xmath736 2 nm accuracy , there remains a certain room to choose particular values of the mean free path . as a rule of thumb , for all
_ al _ nanowires with effective diameter @xmath735 20 nm the best - fitted mean free path ( at low temperatures ) was found to be roughly equal to one - half of the diameter ( fig .
[ f26 r(t ) al zgirski prb 2008 ] , caption ) .
this estimate appears quite reasonable taking into consideration that at these scales and temperatures electron scattering occurs merely at the sample boundaries .
our experimental data and the fits to the theoretical dependence are presented in fig .
[ f26 r(t ) al zgirski prb 2008 ] .
we observe that as the sample diameter decreases the @xmath3 curves become progressively broader , exactly as it is predicted by the qps theory .
our fits demonstrate a good agreement between theory and experiment , thus confirming the important role of qps effects in our thinnest wires .
for all aluminum nanowires the best - fit value for the parameter @xmath262 was determined to be @xmath834 .
this value is smaller than that extracted from the fits to the experimental data by lau _
we believe that this difference can be attributed to different geometry and degree of inhomogeneity of samples used in these two experiments .
we also point out that the temperature dependence of all fitted curves @xmath3 is merely determined by that of the qps action @xmath849 which enters the exponent in the expression for the qps fugacity ( [ fug ] ) whereas an additional power - law dependence @xmath850 turns out to be insignificant : the fits remain essentially unchanged if we set @xmath851 .
this is quite natural since @xmath852 and , on top of that , the temperature dependence of @xmath853 dominates over that emerging from qps interaction in the temperature interval under consideration .
dependencies of the same samples as in fig .
26 taken at close temperatures stabilized with accuracy @xmath736 0.1 mk .
solid lines correspond to the dependence @xmath854 with @xmath855 0.58 , 1.31 and 1.49 ( from top to bottom ) @xcite.,width=302 ] to complete our discussion of qps experiments we point out that very recently an additional experimental evidence for quantum phase slip effects in short moge wires was reported @xcite . in this experiment
the superconducting wires were biased by a large current ( just slightly below the critical one ) in order to decrease an effective potential barrier for qps . similarly to macroscopic quantum tunneling experiments ( see , e.g. , ref . @xcite and further references therein ) the contribution of the qps mechanism was identified in ref .
@xcite by observing low temperature broadening of the distribution for switching currents at which the nanowire is driven to the normal state by quantum fluctuations of the order parameter .
further quantitative investigations of this effect as well as the analysis of the role of possibly existing external noise and some other extrinsic factors would be highly desirable . summarizing our analysis of the experimental data on superconducting nanowires , we can conclude that homogeneity of these wires is one of the central issues which can influence the data interpretation .
the methods enabling experiments on the same nanowire while increasing @xcite or decreasing @xcite its characteristic dimension to a large extent eliminate artifacts related to the uniqueness of samples fabricated in independent processing runs . on the other hand ,
series of experiments on @xmath681 wires @xcite provide large statistics of independent samples .
another conclusion is that longer nanowires appear to be more suitable for experimental investigations of qps effects while the behavior of very short wires could be to a large extent determined by the josephson physics ( classical or quantum ) which can essentially mask the qps one .
the whole scope of the analyzed experimental data obtained in different groups within different fabrication methods employing different materials clearly demonstrates that at sufficiently low temperatures quantum phase slips provide an important mechanism which may cause non - zero resistance of superconducting nanowires .
a vast majority of the data on wires with diameters in the 10 nm range confirms that the taps scenario can not explain the broad @xmath3 dependencies observed in these structures and favors qps @xcite as the only possible scenario .
although in general this scenario can include a number of rather sophisticated physical effects , such as quantum phase transitions ( see chapter 5 ) , it appears that in many cases the rapid change from superconducting to normal behavior can simply be interpreted as a crossover between the regimes of low and high qps rates @xcite controlled by the wire cross section @xmath32 or , equivalently , by the the wire normal state resistance per unit length @xmath778 .
also the dominating contribution to the temperature dependence observed , e.g. , in sub-15-nm @xmath681 @xcite and aluminium @xcite nanowires is well described by that of the qps rate @xcite , see figs .
[ f20 lau moge our fits ] and [ f26 r(t ) al zgirski prb 2008 ] .
although more experiments would be highly desirable , already at this stage we can conclude that the existing experimental data unambiguously confirm our understanding of basic features of qps physics in superconducting nanowires .
application of a weak perpendicular magnetic field to thin superconducting strips governed by fluctuations revealed an unusual effect : negative magnetoresistance ( nmr ) ( figs .
[ f28 nmr dynes ] and [ f29 nmr zgirski ] ) .
the phenomenon has been observed in lead @xcite and aluminium @xcite nanowires .
the effect is seen only in very narrow quasi-1d superconducting channels in the resistive state sufficiently far from the critical temperature @xmath4 .
a trivial explanation related to the kondo mechanism is problematic , as the presence of magnetic impurities in those experiments is believed to be negligible and in any case has not been independently proven . even if we were to assume the presence of kondo impurities , several features of nmr would still remain unclear .
first , the corresponding magnetic fields are too small to polarize magnetic moments of any impurities at such values of @xmath7 . in addition
, it is known that aluminum is immune to creation of localized magnetic moments with concentration up to few at % @xcite .
second , there is a pronounced diameter dependence , making nmr observable only in thinnest wires .
third , the onset of superconductivity is not affected by weak magnetic fields : nmr is observed only at the bottom part of the @xmath3 transition .
perhaps we should also mention another effect
enhancement of the critical current by magnetic field which was observed in _
moge _ and _ nb _ nanowires @xcite and attributed to interplay between spin - exchange scattering from residual magnetic impurities and the orbital and zeeman effects @xcite .
however , it remains unclear if this effect is related to nmr in any way . on top of that , presently there exists no evidence for the presence of magnetic impurities in experiments with aluminium @xcite and lead @xcite . 1 h ) recorded dependence @xmath856 for the 11-nm sample from fig .
[ f26 r(t ) al zgirski prb 2008 ] .
the perpendicular magnetic field @xmath85719.6 mt was switched on and off several times while sweeping the temperature .
the top branch corresponds to zero field , while the lower one to the field `` on '' .
inset : resistance versus perpendicular magnetic field @xmath858 for the same sample measured at constant temperature and small ac current @xcite.,width=302 ] nmr has been observed in tin microbridges @xcite and aluminium nanostructures demonstrating the resistive transition anomaly ( see below ) @xcite .
however , the dimensions of those structures were significantly larger as compared to ultrathin nanowires @xcite .
presumably the origin of nmr in these earlier reports can be ascribed to charge imbalance @xcite or to sample inhomogeneity @xcite .
nmr has been predicted in disordered superconducting wires @xcite .
however , the contribution responsible for nmr within this model is exponentially small as compared to the effective resistance produced by taps .
this is clearly not the case in experiments @xcite .
a plausible explanation of the nmr effect could be related to the reduction of the energy gap @xmath102 in the magnetic field .
it was argued @xcite that this reduction leads to the following trade - off .
on one hand , the core action @xmath859 decreases with increasing magnetic field and , hence , the qps rate becomes bigger . on the other hand ,
the quasiparticle resistance @xmath860 decreases too implying increasing dissipation which , in turn , suppresses the qps rate .
provided the latter trend dominates over the former one , the contribution of quantum fluctuations to the wire resistance gets reduced with increasing magnetic field , thus leading to nmr @xcite .
yet another idea employs possible formation of a charge imbalance region accompanying each phase slip event @xcite .
this non - equilibrium region , if it exists , would provide dissipation outside the core of a phase slip .
the corresponding ohmic contribution can be effectively suppressed by the magnetic field , resulting in nmr .
however , so far the validity of the charge imbalance concept was only demonstrated at temperatures sufficiently close to @xmath4 and its applicability to qps is by no means obvious .
this mechanism still requires a solid theoretical justification .
rather unusual step - like shape of @xmath3 transition has been reported in tin whiskers @xcite . in relatively long structures @xmath861 mm with `` manually '' fixed electrodes using conducting epoxy and/or wood metal @xcite the steps
were observed at relatively high measuring current densities @xmath862 a / cm@xmath742 ( fig .
[ f30 meyer r(t ) steps])*. * in hybrid whisker - based structures with lithographically - fabricated electrodes @xcite ( fig .
[ f6 sog structure ] ) the steps on @xmath3 dependencies were observed between closely located probes ( @xmath863 10 @xmath35 m ) and at current densities as low as @xmath864 a / cm@xmath742 * * ( fig .
[ f31 step r(t ) sog whisker])*. * for comparison , the experimental critical current density in pure tin whiskers at sufficiently small temperatures reaches its theoretical value @xmath865 a / cm@xmath742 . the origin of the phenomenon is not clear .
the step - like @xmath3 dependencies were not reported in similar experiments using tin whiskers @xcite .
note that the steps on @xmath3 transition should not be confused with step - like @xmath411 characteristics observed in 1d superconducting wires very close to the critical temperature and at bias currents exceeding the temperature - dependent critical value @xmath866 .
the origin of the @xmath411 steps is usually associated with current - induced non - equilibrium effects @xcite .
dependence in tin whisker at various excitation currents @xcite.,width=302 ] dependence in a hybrid whisker - based structure similar to that of fig .
[ f6 sog structure ] .
arrows indicate the direction of the temperature sweep .
small hysteresis is a consequence of the temperature measurement finite response time .
the inset shows an enlarged view of several @xmath3 steps measured at two different currents .
the curves are shifted slightly due to zero offset drift of the front - end battery powered preamplifier @xcite.,width=302 ] intensive studies of various lift - off fabricated quasi-1d nanostructures revealed an unusual `` resistive transition anomaly '' : resistance increase above the normal state value @xmath450 at the top of the superconducting transition ( fig .
[ f32 santhanam rt anomaly ] ) @xcite .
no special correlation has been observed between the absolute value of the sample length @xmath350 ( distance between the probes ) and the magnitude of the anomaly .
pronounced `` bumps '' were observed only in relatively short structures with width @xmath744 @xmath867 .
no dependence upon the film thickness within the range 3090 nm has been noticed either .
the same statement holds for the sample topology : both single - connected and non - single - connected structures displayed the bumps .
for the same multiterminal structure , the anomaly could be clearly observed for a pair of voltage contacts , while the neighboring segment showed no signs of the effect . for a given set of contacts ,
the magnitude of the resistive anomaly sometimes depends on the cooling history .
heating up to @xmath70450 k could eliminate the effect .
variation of the measuring current or application of the magnetic field can modify the magnitude of the bump ( fig .
[ f33 r(t ) anomaly current - field ] ) . for aluminium multiterminal nanostructure at different measuring ac currents ( top ) and at various external magnetic fields normal to the sample surface ( bottom ) @xcite.,width=302 ]
let us point out that experimentalists frequently use lock - in technique in order to increase the signal - to - noise ratio .
should a dc component occasionally occur ( e.g. , due to improper grounding ) , experimental dependencies could become proportional to @xmath868 rather than to the desired @xmath869 resulting in `` veird''-looking @xmath3 curves .
utilization of rf filters , such as widely used commercial @xmath870-filters , can also contribute to non - monotonous @xmath3 transitions due to re - arrangement of the measuring ac current through the ground loop even at rather low frequencies @xmath70410 hz . various physical reasons for the resistive transition anomaly in superconducting nanostructures have been proposed , such as the fluctuation - governed resistive state @xcite nonequilibrium quasiparticle charge imbalance within the locus of n / s boundary @xcite or phase - slip centers @xcite .
it should be noted that the resistive transition anomaly has only been reported in relatively inhomogeneous lift - off fabricated nanostructures with the width of the @xmath3 transition strongly exceeding that predicted by the fluctuation models .
it has been shown that the phenomenon can be qualitatively explained by simple geometric considerations concerning the shape of the n / s phase boundary of an inhomogeneous wire in the resistive state @xcite .
the approach has been extended @xcite demonstrating that the combination of the two effects provides reasonable agreement with experiments : ( i ) a geometry effect of the formation of non - perpendicular ( to the wire axis ) n / s boundaries and the corresponding rearrangement of current across the wire , and ( ii ) the existence of nonzero and strongly anisotropic effective resistance of the nonequilibrium superconducting region close to the n / s interface .
considered separately , neither of these two contributions could provide quantitative agreement with the experimental data . to summarize , it is possible that the resistive transition anomaly originates from `` dirty physics '' related to inhomogeneity of the finite length lift - off fabricated wires .
this is supported by the fact that no signs of the anomaly were reported in pure 1d systems @xcite . since early days of low - dimensional superconductivity
it is known that the critical temperature of thin films @xmath871 and wires @xmath872 differ from the corresponding bulk value @xmath873 . in indium , aluminum and zink nanowires @xmath4 increases with decreasing the characteristic dimension @xcite . in lead , niobium and moge an opposite tendency has been observed @xcite .
no noticeable variations of @xmath4 have been reported in tin nanowires @xcite .
the early models considered size - dependent modification of the electron - phonon spectrum and found the corresponding variation of the critical temperature of thin films @xcite .
however , later it was noticed that the phonon spectrum renormalization can significantly contribute to variations of @xmath4 only in ultra - thin structures , while experimentally the effect is observed also in rather thick films ( e.g. up to @xmath874 nm in the case of aluminum ) , where the phonon spectrum is almost identical to that of bulk samples .
additionally , it was found that the critical temperature @xmath871 of ultra - thin quench - condensed films depends not only on thickness , but also on annealing made _ in situ _ in the measuring chamber at cryogenic temperatures @xcite , indicating that the morphology of the film is also important .
another significant effect is an interplay between disorder and electron - electron interactions which can lead to substantial suppression of @xmath4 in reduced dimensions @xcite . on the other hand
, the so - called shape resonance effect @xcite can be responsible for the enhancement of @xmath4 . to conclude , after years of intensive experimental and theoretical studies ,
the origin of the size dependence of the critical temperature @xmath4 remains not fully understood and in many cases it can be determined by a number of different factors .
it is well known that superconducting rings pierced by external magnetic flux @xmath875 develop circulating persistent currents ( pc ) which never vanish .
this phenomenon is a fundamental consequence of macroscopic phase coherence of cooper pair wave functions . in the case of bulk metallic rings fluctuations of the phase @xmath17 of the order parameter
can be neglected , i.e. @xmath17 can be considered as a purely classical variable . in this case
the total phase difference @xmath876 accumulated along the ring circumference @xmath877 is linked to the external flux @xmath875 inside the ring by the well known relation @xmath878 where @xmath879 is an integer number , @xmath880 and , as before , @xmath881 is the superconducting flux quantum .
this relation implies the existence of a phase gradient along the ring which , in turn , means the presence of discrete set of current and energy states labelled by the number @xmath879 . at sufficiently low temperatures @xmath44 quasiparticles are practically irrelevant and the grand partition function of the ring takes the form @xmath882 where @xmath883 defines flux - dependent energy states of a diffusive superconducting ring with radius @xmath26 and cross - section @xmath32 . the ground state energy @xmath884min@xmath885 is a periodic function of the flux @xmath875 with the period @xmath881 .
the derivative of the ground state energy with respect to the flux defines the persistent current @xmath886 combining eqs .
( [ energ ] ) and ( [ defper ] ) one finds @xmath887 where @xmath554 is the electron mass and @xmath888 is the superconducting velocity .
both the current @xmath25 and the velocity @xmath889 are periodic functions of the magnetic flux @xmath875 showing the familiar sawtooth behavior , see also fig .
[ f34 energy and curent vs flux ] .
this picture remains applicable as long as fluctuations of the superconducting phase @xmath17 can be neglected , i.e. as long as the ring remains sufficiently thick . upon decreasing the thickness of a superconducting wire @xmath586 down to values @xmath754 nm range
one eventually reaches the new regime in which quantum fluctuations of the phase gain importance and , as it will be demonstrated below , essentially modify the low temperature behavior of superconducting nanorings .
as we have already learned , at low @xmath7 the most important fluctuations in superconducting nanowires are quantum phase slips .
each phase slip event implies transfer of one flux quantum @xmath881 through the wire out of or into the ring and , hence , yields the change of the total flux inside the ring by exactly the same amount . in other words ,
each qps yields a jump between two neighboring energy states @xmath890 ( [ energ ] ) ( i.e. between two neighboring parabolas in fig
. 34 ) . as a result of such jumps the flux inside the ring fluctuates , its average value @xmath891 decreases and so does the persistent current @xmath25 .
one can also anticipate that the magnitude of this effect should increase with the ring perimeter @xmath350 .
this is because the qps rate @xmath502 increases linearly with @xmath350 , i.e. the bigger the ring perimeter the higher the probability for qps to occur anywhere along the ring .
a quantitative theory of qps effects in nanorings was proposed by matveev _
these authors employed a model of a closed chain of josephson junctions in which case qps can occur only across such junctions . here
will extend our theory of qps ( chapter 5 ) and analyze the effect of quantum fluctuations on persistent currents in _ uniform _ superconducting nanorings .
the key conclusions remain the same for both models . before turning to technical details
it is instructive to point out a formal equivalence between the phenomenon discussed here and charging effects in ultrasmall josephson junctions ( or cooper pair boxes ) @xcite .
indeed , the energy states of our ring ( [ energ ] ) driven by the ( normalized ) external magnetic flux @xmath892 ( see fig .
[ f34 energy and curent vs flux ] ) are fully analogous to such states of a capacitor in a cooper pair box driven by the ( normalized ) gate charge @xmath893 .
furthermore , we will demonstrate that there exists a direct analogy between qps events changing the flux inside the ring by @xmath881 and tunneling events of single cooper pairs changing the capacitor charge by @xmath894 .
this equivalence is reminiscent of the well known duality between phase and charge representations @xcite and will be exploited in our consideration .
we start from the case of sufficiently thick superconducting rings where qps effects can be neglected . in this case by virtue of poisson s resummation theorem ( see , e.g. , sec .
3.3.2 in ref .
@xcite ) one can identically transform the ring partition function ( [ z0 ] ) to the following expression @xmath895),\label{z1}\ ] ] where @xmath896=\int_{0}^{\beta}d\tau\frac{1}{2e_{r}}\left ( \frac{\partial \theta}{\partial\tau}\right ) ^{2}\label{ss0}\ ] ] and @xmath897 plays the role of a
mass for a particle with the coordinate @xmath898 .
this coordinate represents an effective angle .
it is formally analogous to the josephson phase variable in which case @xmath899 just coincides with the charging energy .
let us now consider thinner rings where qps effects gain importance .
in this case we should include all transitions between different energy states ( [ energ ] ) labelled by the number @xmath879 . as we already discussed these transitions are just qps events with the rate @xmath502 defined in eq .
( [ gammaf ] ) .
let us fix @xmath900 and take into account virtual transitions to all other energy states and back .
performing summation over all such contributions we arrive at the series in powers of the rate @xmath502 ( or fugacity @xmath51 ) similar to eq .
( [ partit ] ) . assuming that both the ring perimeter @xmath350 and its thickness are not too large ( the latter condition restricts the parameter @xmath35 ) one can neglect weak logarithmic interaction ( [ sint ] ) between different phase slips .
then spatial qps coordinates @xmath901 become exact zero modes and can be trivially integrated out . as a result
, we arrive at the following contribution to the partition function @xmath902 @xmath903 where , as before , the summation is carried out over neutral charge configurations ( [ nutot ] ) , @xmath904 and @xmath905 is the theta - function .
we note that the term in the exponent describes virtual energy changes occurring due to qps events .
what remains is to add up similar contributions from all other parabolas with @xmath906 . in this way
we arrive at the final expression for the grand partition function of the nanoring @xmath907 together with eq .
( [ tz ] ) this result fully accounts for all qps events in our system . in the absence of qps ( i.e. for @xmath908 ) eqs .
( [ z2 ] ) , ( [ tz ] ) obviously reduce to ( [ z0 ] ) . for non - zero @xmath502
one can rewrite the partition function ( [ tz ] ) , ( [ z2 ] ) in the equivalent form of the following path integral @xmath909),\label{z5}\ ] ] where @xmath910=s_{0}[\theta ] -\gamma_{qps}\int_{0}^{\beta}d\tau\cos\theta(\tau ) .\label{sss}\ ] ] in order to demonstrate that the partition function ( [ z5 ] ) , ( [ sss ] ) is identical to that defined in eqs .
( [ tz ] ) , ( [ z2 ] ) it suffices to perform the hubbard - stratonovich transformation of the kinetic term ( [ ss0 ] ) ( which amounts to introducing additional path integral over @xmath911 ) , to formally expand @xmath912)$ ] in powers of @xmath502 and then to integrate out the @xmath898-variable in all terms of these series .
after these straightforward steps we arrive back at eqs .
( [ tz ] ) , ( [ z2 ] ) .
we also note that this procedure is described in sec .
3.3.5 of ref .
@xcite , therefore we can avoid further details here .
( [ z5 ] ) , ( [ sss ] ) define the grand partition function for a quantum particle in a cosine periodic potential which is , in turn , equivalent to that for a josephson junction in the presence of charging effects @xcite .
this partition function can be identically rewritten as ( cf .
( 3.94 ) in ref .
@xcite ) @xmath913 where @xmath914 are the energy bands of the problem defined by the solutions of the well - known mathieu equation . for @xmath915 one
has @xmath916 , \ ] ] i.e. the energy bands remain nearly parabolic @xmath917 except in the vicinity of the crossing points where gaps open due to level repulsion ( see fig .
[ f34 energy and curent vs flux ] ) . in this
limit the value of the gap between the lowest and the first excited energy bands just coincides with the qps rate @xmath918 . for larger @xmath502 the bandwidth shrinks while the gaps become bigger . in the limit @xmath919
the lowest band coincides with @xmath920 the gap between the two lowest bands is @xmath921 .
these results are sufficient to evaluate pc in superconducting nanorings in the presence of quantum phase slips .
as before , taking the derivative of the ground state energy with respect to the flux @xmath875 and making use of eq .
( [ gammaf ] ) we find that for @xmath915 and outside immediate vicinity of the points @xmath922 pc is again defined by eqs .
( [ evcur ] ) , ( [ st ] ) . in the opposite limit @xmath919
we find@xmath923 we observe that in the latter limit pc is exponentially suppressed as @xmath924 where neglecting a numerical prefactor we estimate @xmath925 this simple formula sets the size scale beyond which one would expect pc to be exponentially suppressed in superconducting nanorings with @xmath926 nm .
e.g. for @xmath49 of order 100 nm and @xmath927 eq .
( [ rc ] ) yields @xmath928 of order of a micron .
we would like to emphasize that during our analysis we employed only one approximation neglecting logarithmic inter - qps interaction effects .
if needed , such effects can also be included into our consideration and may only lead to unimportant modifications in our results for rings with very large perimeters @xmath350 .
the above results demonstrate practically the same qualitative features as those found within a different approach for the model of nanorings formed by josephson chains @xcite . in particular , both for granular and for homogeneous rings the dependence of pc on @xmath892 gradually changes from sawtooth to sinusoidal as the ring perimeter @xmath350 increases .
this change is accompanied by suppression of pc amplitude which eventually becomes an exponentially decaying function of the ring radius @xmath26 ( [ lsc ] ) in the limit @xmath929 .
in other words , eq .
( [ lsc ] ) sets the length scale @xmath930 beyond which persistent currents in superconducting nanorings should be essentially washed out by quantum fluctuations .
in some sense the scale @xmath931 plays the role of a dephasing length in our problem demonstrating that zero - point fluctuations can destroy macroscopic phase coherence down to @xmath45 even in superconducting systems .
this conclusion is qualitatively consistent with the results derived for normal metallic conductors @xcite and single channel rings coupled to dissipative baths @xcite .
in fact , a close similarity between the partition function ( [ z5 ] ) , ( [ sss ] ) and those studied in refs .
@xcite exists on a formal level as well : these problems are described by similar path integrals except in the case of normal rings coupled to dissipative environments one should include the corresponding non - local in time term into the action instead of the last term in eq .
( [ sss ] ) . at sufficiently large @xmath350 these path integrals can also be handled in exactly the same manner : in both problems the dominating contribution comes from instantons ( kinks ) describing quantum tunneling of the angle variable @xmath898 between different topological sectors of the problem . in all cases this procedure yields low temperature pc in the form ( [ esm ] ) where one finds @xmath932 in the case of caldeira - leggett environments @xcite and @xmath933 in the case of so - called dirty electron gas environments @xcite .
an important qualitative difference between our present problem and those of normal nanorings with dissipation lies in the fact that in the latter case dissipation explicitly _ violates time - reversal symmetry _ ( thus causing genuine decoherence of electrons ) , while no such symmetry is violated in eqs .
( [ z5 ] ) , ( [ sss ] ) . hence , in our case in contrast to @xcite exponential suppression of pc ( [ esm ] ) can also be interpreted as a non - trivial coordinate - dependent renormalization effect .
this difference is just the same as that between _ dissipativeless _
cooper pair tunneling ( which , just like qps , opens inter - band gaps in the spectrum of josephson junctions ) and _ dissipative _ single electron tunneling ( which opens no such gaps ) . for more details on the latter
subject we refer the reader to the review @xcite .
finally we should mention that one can also consider a slightly modified situation of rings consisting of thicker and thinner parts , as it is shown in fig .
[ f35 ring constriction ] . assuming that qps effects are negligible in a thicker part of the ring and they can only occur in its thinner part ( of length @xmath934 ) we arrive at exactly the same results as above in which one should only replace the ring perimeter @xmath350 by the length @xmath934 .
in particular , it follows from our analysis that such rings also exhibit the property of an exact duality to mesoscopic josephson junctions if we identify @xmath899 with the junction charging energy @xmath601 and the qps rate @xmath502 with the josephson energy @xmath600 . for this reason mooij and
nazarov @xcite suggested to call systems depicted in fig .
[ f35 ring constriction ] quantum phase slip junctionsand argued that any known result on electron transport in circuits containing josephson junctions can be exactly mapped onto a dual result for a qps junction in a dual circuit .
this observation can be used in metrology , e.g. , for practical implementation of the electric current standard in the above structures .
mooij and harmans @xcite proposed to use rings with qps junctions for experimental realization of quantum phase slip flux qubits .
it is also worth pointing out that previously a similar exact duality between phase and charge variables in josephson junctions was discussed @xcite , see , e.g. , eq .
( 5.76 ) in ref .
@xcite . according to this property the josephson junction itself
can also be a qps junction if we interchange the canonically conjugate phase and ( quasi)charge variables @xmath935 . in this sense
one can identify the angle variable @xmath898 with the ( quasi)charge @xmath936 passing through the qps junction .
in our previous analysis of qps effects in superconducting nanorings we followed the standard technique developed for grand canonical ensembles , i.e. we implicitly assumed that the total number of electrons in the system @xmath519 may fluctuate and the chemical potential @xmath937 is fixed .
obviously , this assumption is not correct for rings which are disconnected from any external circuit . in that case electrons can not enter or leave the ring and , hence , the number @xmath519 is strictly fixed , but the chemical potential @xmath937 , on the contrary , fluctuates .
it turns out that novel effects emerge in this physical situation .
these effects will be discussed below in this section . to a large extent
we will follow the analysis developed in refs .
@xcite .
it is well known that thermodynamic properties of isolated superconducting systems are sensitive to the parity of the total number of electrons @xcite even though this number is macroscopically large .
this parity effect is a fundamental property of a superconducting ground state described by the condensate of cooper pairs .
the number of electrons in the condensate is necessarily even , hence , for odd @xmath519 at least one electron remains unpaired having an extra energy equal to the superconducting energy gap @xmath102 .
this effect makes thermodynamic properties of the ground state with even and odd @xmath519 different .
clear evidence for such parity effect was demonstrated experimentally in small superconducting islands @xcite . at the first sight , this parity effect should have little impact on the supercurrent because of the fundamental uncertainty relation @xmath938 .
should the electron number @xmath519 be fixed , fluctuations of the superconducting phase @xmath17 become large disrupting the supercurrent . on the other hand , in transport experiments with fluctuations of @xmath17
being suppressed the parity effect can not be observed because of large fluctuations of @xmath519 .
consider now isolated superconducting rings pierced by the magnetic flux @xmath875 . in accordance with the number - phase uncertainty relation the _ global _ superconducting phase of the ring fluctuates strongly in this case , however these fluctuations are decoupled from the supercurrent and therefore can not influence the latter .
in this situation the parity effect may substantially modify pc in superconducting nanorings at sufficiently low temperatures .
in particular , we will show that changing the electron parity number from even to odd results in spontaneous supercurrent in the ground state of such rings without any externally applied magnetic flux .
in other words , our fundamental conclusion will be that the _ bcs ground state of a canonical ensemble with odd number of electrons is the state with spontaneous supercurrent_. in order to systematically investigate interplay between the parity effect and persistent currents in superconducting nanorings we will employ the parity projection formalism @xcite which we will briefly outline here .
the grand canonical partition function @xmath939 is linked to the canonical one @xmath940 as @xmath941 here and below @xmath942 is the system hamiltonian .
inverting this relation and defining the canonical partition functions @xmath943 and @xmath944 respectively for even ( @xmath945 ) and odd ( @xmath946 ) ensembles , one gets @xmath947 where @xmath948e^{-\beta({\mathcal{h}}-\tilde{\mu}n)}\right\ } \nonumber\\ & = & { \frac{1}{2}}\left ( { \mathcal{z}}(t,\tilde{\mu})\pm{\mathcal{z } } ( t,\tilde{\mu}+i\pi t)\right ) \label{pm1}\ ] ] are the parity projected grand canonical partition functions .
for @xmath949 it is sufficient to evaluate the integral in ( [ invrel ] ) within the saddle point approximation @xmath950 where @xmath951 are the parity projected thermodynamic potentials , @xmath952\ ] ] and @xmath953 $ ] .
chemical potentials @xmath954 are defined by the saddle point condition @xmath955 .
the main advantage of the above formalism is that it allows to express the canonical partition functions and thermodynamical potentials in terms of the parity projected grand canonical ones thereby enormously simplifying the whole calculation .
we further note that @xmath956 is just the standard grand canonical thermodynamic potential and @xmath957 represents the corresponding potential linked to the partition function @xmath958 .
it is easy to see @xcite that in order to recover this function one can evaluate the true grand canonical partition function @xmath959 , express the result as a sum over the fermi matsubara frequencies @xmath960 and then substitute the bose matsubara frequencies @xmath961 ( @xmath962 ) instead of @xmath963 .
this procedure automatically yields the correct expression for @xmath958 and , hence , for @xmath957 .
having found the thermodynamic potentials for the even and odd ensembles one can easily determine the equilibrium current @xmath964 .
consider , as before , isolated superconducting rings pierced by the magnetic flux @xmath875 .
making use of the above expressions one finds pc circulating inside the ring : @xmath965 where the upper / lower sign corresponds to the even / odd ensemble and we have defined @xmath966 eqs .
( [ ie / o ] ) , ( [ cureo ] ) represent a direct generalization of the grand canonical formula ( [ defper ] ) to canonical ensembles .
let us first consider homogeneous nanorings with cross section @xmath32 and perimeter @xmath877 .
as before , rings will be assumed sufficiently thin , @xmath967 . on the other hand , below we will neglect qps effects , i.e. describe superconducting properties of such rings within the parity projected mean field bcs theory .
as we have already learned , this description is justified provided the condition @xmath968 is satisfied .
hence , the ring should not be too thin and the total number of conducting channels should remain large @xmath969 .
in addition , the perimeter @xmath350 should not exceed the scale @xmath970 ( [ lsc ] ) .
finally , we will neglect the difference between the mean field values of the bcs order parameter for the even and odd ensembles @xcite .
this is legitimate provided the ring volume is large enough , @xmath971 , where , as before , @xmath29 is the density of states at the fermi level . evaluating thermodynamic potentials @xmath972 and expressing the result in terms of the excitation energies @xmath973 and the order parameter @xmath102 one finds @xcite @xmath974 where @xmath975 , @xmath976 is the single - particle energy operator , @xmath977 , where @xmath978 is the quasiparticle momentum , @xmath979 and the superconducting velocity @xmath980 is defined in eq .
( [ st ] ) .
the above equations allow to fully determine pc in superconducting nanorings with even and odd number of electrons @xmath964 .
the parity effect becomes observable at sufficiently low temperatures @xcite @xmath981 .
here we consider the most interesting limit @xmath982 . from eqs .
( [ ie / o])-([omb ] ) we find that in this limit the current @xmath983 exactly coincides with the standard grand canonical result @xmath984 with @xmath985 , while the current @xmath986 for the odd ensemble reads @xcite @xmath987 here we introduced the electron density in the case of even / odd total number of electrons @xmath988 .
we observe that in the case of odd ensembles there exists an additional term which modifies the flux dependence of pc and , as we will demonstrate below , leads to a number of fundamentally important effects .
unfortunately , in the case of homogeneous rings the difference between pc in even and odd ensembles turns out to be hardly observable .
indeed , these difference is inversely proportional to the total number of conducting channels , @xmath989 . for this reason the parity effect remains vanishingly small in generic metallic rings with @xmath990 . on the other hand , for ultra - thin nanorings with @xmath991 pc
is essentially wiped out due to proliferation of quantum phase slips .
estimating @xmath992 , we conclude that for @xmath53 ( i.e. when the qps fugacity is already large and , hence , quantum suppression of pc becomes very strong ) the number of conducting channels yet remains parametrically large @xmath993 .
this estimate demonstrates that the parity effect on pc is never important in the case of _ homogeneous _ superconducting nanorings and appears to be practically unobservable in such systems .
@xcite.,width=302 ] we now turn to the situations in which parity effect gains importance and can be directly probed in modern experiments .
let us slightly modify our system and consider a relatively thick superconducting ring with large @xmath968 interrupted by a thin wire of length @xmath934 with only few conducting channels @xmath994 thus forming a weak link inside the superconducting ring , see fig .
[ f36 sns ring ] . without loss of generality
this wire can be considered normal no matter if it is made of a normal or a superconducting material . in the latter case
quantum fluctuations would fully suppress the order parameter inside such a wire bringing it into the normal state .
in contrast , qps effects in superconducting parts of the ring can be neglected thus making the mean field bcs description applicable .
a clear advantage of these structures in comparison to homogeneous rings is that in the former case the effect of the electron parity number on pc can be large due to the condition ( [ n1 ] ) . in order to evaluate pc in such structures
we again employ the parity projection formalism . according to eqs .
( [ ie / o ] ) , ( [ cureo ] ) we need to evaluate both the currents @xmath995 and the difference between the `` fermi '' and bose thermodynamic potentials @xmath996 .
the currents @xmath995 can be conveniently expressed via the phase difference across the weak link @xmath997 by means of the general formula @xcite @xmath998 where the sum runs over conducting channels of the normal wire and the function @xmath999 was evaluated in ref .
the difference of thermodynamic potentials @xmath1000 is defined as a sum of the contributions from superconducting ( @xmath1001 ) and normal ( @xmath1002 ) parts of the ring .
the former is evaluated with the aid of eqs .
( [ omf ] ) , ( [ omb ] ) which yield the standard result @xcite @xmath1003 while the latter is found by integrating @xmath1004 ( [ fbfb ] ) over the phase @xmath17 .
we now consider several important limiting cases .
the first limit is that of a very short normal wire @xmath1005 .
this is essentially the limit of a quantum point contact . in practice
this limit is realized for @xmath1006 .
even smaller values of @xmath934 are required provided the contact transmission is small . in the limit of a quantum point contact one
finds @xcite @xmath1007 where @xmath1008 define transmissions of the conducting channels . substituting this function into the above equations one arrives at the final result @xcite @xmath1009@xmath1010 .\label{result}\ ] ] here @xmath1011 are andreev energy levels in a quantum point contact .
( represented by the term in the square brackets in eq .
( [ result ] ) versus @xmath17 in a single mode qpc at different temperatures for even ( three upper curves ) and odd ( three lower curves ) ensembles . here
we have chosen the channel transmission @xmath1012 @xcite.,width=302 ] the first line of eq .
( [ result ] ) defines the standard grand canonical result @xcite while the term in the square brackets accounts for the parity effect in our system . for @xmath1013 and at @xmath2
this term reduces to unity for even ensembles and to zero for odd ones , i.e. pc turns out to be _ totally blocked _ in the case of odd number of electrons @xcite . for even ( a ) and odd ( b ) ensembles @xcite.,width=302 ]
the physics of this blocking effect is rather transparent and can be understood as follows .
we first recall that in the limit @xmath1005 the josephson current can be expressed only via the contributions from discrete andreev energy states @xmath1014 as @xcite @xmath1015 .\label{jos2}\ ] ] using the fermi filling factors for these states @xmath1016^{-1}$ ] one arrives at the standard grand canonical results @xcite . in the case of superconducting rings with fixed number of electrons @xmath519 these filling factors should be modified .
let us set @xmath432 .
for even @xmath519 all electrons are paired occupying available states with energies below the fermi level ( see fig .
[ f38 andreev levels]a ) . in this case
one has @xmath1017 , @xmath1018 , the current is entirely determined by the contribution of the quasiparticle state @xmath1019 and eq .
( [ jos2 ] ) yields the same result as one for the grand canonical ensemble .
by contrast , in the case of odd @xmath519 one electron always remains unpaired and occupies the lowest available energy state in our case @xmath1020 above the fermi level .
hence , for odd @xmath519 one has @xmath1021 ( fig .
[ f38 andreev levels]b ) , the contributions of the two andreev states in eq .
( [ jos2 ] ) exactly cancel each other , and the current remains zero for any @xmath17 or the magnetic flux @xmath875 .
this is just the effect of parity - induced blocking of pc derived above from formal considerations . for @xmath624 eq .
( [ result ] ) demonstrates that both for even and especially for odd @xmath519 the current - phase relation for qpc may substantially deviate from that derived for the grand canonical ensemble @xcite , see fig .
[ f37 pc grand ] . for even ensembles
the supercurrent _ increases _ above its grand canonical value .
this effect is mainly pronounced for phases @xmath17 not very far from @xmath1022 and at sufficiently low @xmath7 it becomes progressively more important with increasing temperature .
on the contrary , for odd ensembles the supercurrent is always suppressed below its grand canonical value . this suppression is gradually lifted with increasing temperature , though at phases @xmath17 in the vicinity of the point @xmath1022 blocking of pc may persist up to sufficiently high @xmath7 .
( [ result ] ) also shows that in quantum point contacts with several conducting channels and at @xmath432 the current through the most transparent channel will be blocked by the odd electron .
hence , though blocking of pc remains incomplete in this case , it may nevertheless be important also for quantum point contacts with @xmath1023 .
let us now turn to another important limit of superconducting rings containing a normal wire of length @xmath1024 .
in contrast to the case @xmath1005 considered above , the josephson current in @xmath1025 structures can not anymore be attributed only to the discrete andreev states inside a weak link , and an additional contribution from the states in the continuum should also be taken into account .
furthermore , for any non - zero @xmath934 there are always more than two discrete andreev levels in the system .
accordingly , significant modifications in the physical picture of the parity effect in such sns rings can be expected .
junction with @xmath1026 and their occupation at @xmath2 for even ( a ) and odd ( b ) ensembles @xcite.,width=302 ] the key difference can be understood already by comparing the typical structure of discrete andreev levels in sns junctions ( fig .
[ f39 andreev sns ] ) with that of a quantum point contact ( fig .
[ f38 andreev levels ] ) . as before , in the limit @xmath432
all states below ( above ) the fermi level are occupied ( empty ) provided the total number of electrons in the system is even ( fig .
[ f39 andreev sns]a ) .
if , on the other hand , this number is odd the lowest andreev state above the fermi energy is occupied as well ( fig .
[ f39 andreev sns]b ) thus providing an additional contribution to the josephson current .
this contribution , however , cancels only that of a symmetric andreev level below the fermi energy , while the contributions of all other occupied andreev levels and of the continuum states remain uncompensated .
hence , unlike in the case @xmath1005 , in @xmath1025 rings one should not anymore expect the effect of pc blocking by the odd electron but rather some other non - trivial features of the parity effect . ) for sns rings with @xmath1013 : @xmath1027 ( dashed ) line and @xmath1028 ( solid ) line @xcite.,width=226 ] this conclusion is fully confirmed by our quantitative analysis @xcite .
let us restrict our attention to transparent sns junctions in which case the function @xmath1029 is the same for all transmission channels and reads @xcite @xmath1030@xmath1031 substituting this function into ( [ fbfb ] ) and repeating the whole calculation as above , we arrive at the final result @xcite which takes a particularly simple form in the limit @xmath432 and @xmath1032 : @xmath1033 this result applies for @xmath1034 and should be @xmath427-periodically continued otherwise .
we observe that at @xmath2 the current @xmath983 again coincides with that for the grand canonical ensembles @xcite , while in the case of odd ensembles the current - phase relation is shifted by the value @xmath1035 .
this shift has a simple interpretation being related to the odd electron contribution @xmath1036 from the lowest ( above the fermi level ) andreev state @xmath1037 inside the sns junction .
as we have expected , this contribution indeed does not compensate for the current from other electron states .
rather it provides a possibility for a parity - induced @xmath870-junction state @xcite in our system : according to eq .
( [ sns ] ) for single mode @xmath1025 junctions the sawtooth current - phase relation will be shifted exactly by @xmath870 , see fig .
[ f40 current phase sns ] . more generally , we can talk about a novel @xmath1038-junction state , because in the odd case the minimum josephson energy ( zero current ) state is reached at @xmath1039 , see fig .
[ f40 current phase sns ] . for any @xmath1023
this is a twofold degenerate state within the interval @xmath1034 . in the particular case
@xmath1040 the current - phase relation @xmath1028 turns @xmath870-periodic , see fig .
41 . only for the odd ensembles ( second eq .
( [ sns ] ) ) and for @xmath1041 and @xmath1042 @xcite.,width=226 ] let us recall that the @xmath870-junction state can be realized in @xmath1025 structures by driving the electron distribution function in the contact area out of equilibrium @xcite . here , in contrast , the situation of a @xmath870- or @xmath1035-junction is achieved in thermodynamic equilibrium . along with this important difference , there also exists a certain physical similarity between the effects discussed here and in refs .
@xcite : in both cases the electron distribution function in the weak link deviates substantially from the fermi function .
it is this deviation which is responsible for the appearance of the @xmath870-junction state in both physical situations .
perhaps the most spectacular physical consequence of the parity effect in sns rings is the presence of _ spontaneous _ supercurrents _ in the ground state _ of such rings with odd number of electrons .
similarly to the case of standard @xmath870-junctions @xcite such spontaneous supercurrents should flow even in the absence of an externally applied magnetic flux .
unlike in ref .
@xcite , however , here the spontaneous current state occurs for _ any _ inductance of the ring because of the non - sinusoidal dependence @xmath1028 .
( @xmath1034 ) for @xmath1013 and different values of the parameter @xmath1043 @xcite.,width=226 ] consider , for instance , the limit @xmath1044 . in the case of odd number of electrons the ground state energy of an sns ring can be written in a simple form @xmath1045 where the first term is the magnetic energy of the ring ( @xmath1046 is the ring inductance ) while the second term represents the josephson energy of an sns junction . minimizing ( [ e ] ) with respect to the flux @xmath1047
one immediately concludes that the ground state of the ring is a twofold degenerate state with a non - vanishing spontaneous current @xmath1048 ^{-1}\label{isp}\ ] ] flowing either clockwise or counterclockwise . in the limit of small inductances @xmath1049 this
current does not vanish and its amplitude just reduces to that of the odd electron current at @xmath1050 .
one finds @xcite : @xmath1051 at @xmath1052 the amplitude of the current @xmath1053 can be evaluated numerically , see fig .
[ f43 spontaneous ] .
one observes that in agreement with eq .
( [ sp1 ] )
@xmath1053 increases linearly with @xmath934 at small @xmath934 , reaches its maximum value @xmath1054 at @xmath1055 and then decreases with further increase of @xmath934 approaching the dependence ( [ sp2 ] ) in the limit of large @xmath934 . for generic bcs superconductors
the magnitude of this maximum current can be estimated as @xmath1056 na .
these values might be considered as surprisingly large ones having in mind that this current is associated with only one andreev electron state . as a function of the parameter @xmath51 at @xmath2 . in the inset
, the same function is shown on the @xmath1057 scale .
dashed lines indicate the asymptotic behavior of @xmath1058 in the limits of small and large @xmath51 @xcite.,width=302 ] note that in the above analysis we merely assumed that the normal wire is sufficiently clean and , on top of that , is in a good electric contact with superconductors .
since both these assumptions can be violated in a realistic experiment it is important to discuss the corresponding modifications of our results .
assume first that the transmissions of both ns interfaces are small @xmath1059 .
in this limit electron transport across the junction is mainly due to resonant tunneling through discrete energy levels inside the normal metal . for simplicity we will restrict our analysis to a single channel junction @xmath1013 .
the most interesting physical situation is realized in the limit of short junction @xmath1060 . in the case of a one dimensional metal of length @xmath934 , the level spacing in the vicinity of the fermi energy
is @xmath1061 .
hence , the condition for the short junction regime can also be represented in the form @xmath1062 .
electron tunneling causes a non - zero linewidth of the energy levels which is proportional to @xmath1063 .
this value is much smaller than @xmath1064 , hence , the resonances remain well separated . in this situation
it suffices to take into account only the closest level to the fermi energy inside the normal metal . as before , making use of eqs .
( [ fbfb ] ) combined with the proper expression for the @xmath1065-function @xcite one finds @xcite @xmath1066 where @xmath1067 @xmath1068 , @xmath1069 , @xmath1070 are the tunneling rates , @xmath1071 and the total transmission probability at the fermi energy @xmath1072 is given by the breit - wigner formula @xmath1073 where @xmath1074 is the energy of a resonant level .
junction with @xmath1075 : ( a ) @xmath1076 and ( b ) @xmath1077 and @xmath1078 . in both cases
@xmath1079 @xcite.,width=226 ] it follows from eqs .
( [ fbfb1 ] ) , ( [ fbfb2 ] ) that although the transparencies of both barriers are low the total transmission @xmath1072 and , hence , the josephson current shows sharp peaks provided the fermi energy becomes close to a bound state inside the junction . on the other hand , eqs .
( [ fbfb1 ] ) , ( [ fbfb2 ] ) demonstrate that even in the vicinity of resonances the behavior of the josephson current as a function of the phase difference @xmath17 and temperature @xmath7 can substantially deviate from that for transparent @xmath1025 junctions . in order to understand the physical reasons for such a difference
it is instructive to compare the structure of discrete andreev levels for ballistic ( @xmath1076 ) sns junctions with that for junctions with weakly transmitting ns interfaces @xmath1077 , see fig .
[ f44 andreev single sns ] .
the spectrum of the latter system consists only of a single non - degenerate state @xmath1080 in the interval @xmath1081 ( fig .
[ f44 andreev single sns]b ) . as a result ,
the behavior of @xmath1080 at small @xmath17 is smooth and the derivative of @xmath1082 with respect to @xmath17 has no jump at @xmath1083 .
in contrast , in the case of ballistic sns junctions discrete levels become split at arbitrary small values of the phase @xmath17 ( fig .
[ f44 andreev single sns]a ) and the derivative of the lowest andreev level with respect to @xmath17 acquires a jump at @xmath1083 . as
this feature is absent in resonant sns junctions the spontaneous current in the ground state of such systems can only develop at not very small ring inductances .
the results for pc in sns rings with resonant transmission are presented in fig .
[ f45 current phase resonant ] .
they clearly demonstrate that at sufficiently low temperatures the @xmath870-junction state should be realized in the case of odd number of electrons . for @xmath1084 , @xmath1085 and different values of the parameter @xmath1086 @xcite.,width=302 ]
finally let us turn to the case of a disordered normal wire .
the difference between pc values for odd ( @xmath986 ) and even ( @xmath983 ) ensembles is related to the minigap value @xmath1087 inside the normal metal @xcite .
this relation acquires a particularly simple form in the limit @xmath1088 : @xmath1089 the last term in this equation describes the contribution to the current from the odd electron occupying the lowest available state above the minigap @xmath1087 in the density of states of the normal metal .
let us first evaluate pc for the even ensemble @xmath983 . as before , at @xmath2 this current identically coincides with one calculated for the grand canonical ensemble .
the results for the current - phase relation @xmath1027 are displayed in fig .
[ f46 jos current odd even ] at various impurity concentrations .
pc in the odd ensemble @xmath986 at @xmath2 can be evaluated from eq .
( [ ioe ] ) . combining our results for @xmath1090 with those for the minigap @xmath1087
derived in ref .
@xcite we arrive at a typical dependence @xmath1028 displayed in fig .
[ f46 jos current odd even ] .
we observe that at sufficiently large values of @xmath1091 the absolute value of the odd electron contribution to pc @xmath1092 exceeds the term @xmath1027 and the total current @xmath986 changes the sign .
this non - trivial parity - affected current - phase relation is specific for sns rings with disorder and it substantially differs from the current - phase relations derived above for sns rings with ballistic and resonant transmissions . for the odd and even number of electrons in the ring @xcite.,width=302 ] at the same time , as in the previous cases , in the odd ensemble there exists a possibility both for a @xmath870-junction state and for spontaneous currents in the ground state of the system without any externally applied magnetic flux .
let us evaluate the ground state energy of the sns junction by integrating eq .
( [ ioe ] ) with respect to the phase @xmath17 .
one finds @xmath1093 where @xmath1094 are the ground state energies of sns junction for even and odd number of electrons in the ring .
while the energy @xmath1095 is always non - negative and reaches its minimum at @xmath1083 , in the odd case the ground state energy @xmath1096 can become negative reaching its absolute minimum at @xmath1022 .
this physical situation of a @xmath870-junction is illustrated in fig .
[ f47 omega ] .
it is easy to find out under which conditions the @xmath870-junction state becomes possible . for that purpose
it is sufficient to observe that for any impurity concentration @xmath1097 , where @xmath683 is the grand canonical critical current at @xmath2 and the prefactor @xmath1098 depends on the form of the current - phase relation .
the @xmath870-junction condition @xmath1099 is equivalent to the inequality @xmath1100 it is obvious from fig .
[ f47 omega ] that in the many channel limit the inequality ( [ picond ] ) can not be satisfied for sufficiently large @xmath148 .
on the other hand , for sufficiently short mean free paths @xmath1101 decays faster with decreasing @xmath148 as compared to the minigap @xmath1102 , and the @xmath870-junction state becomes possible .
in particular , in the diffusive limit one finds @xcite @xmath1103 and @xmath1104 , where @xmath450 is the drude resistance of a normal metal . of an sns ring as a function of the phase difference @xmath17 for the even and odd ensembles .
the solid curve corresponds to a @xmath870-junction state @xcite.,width=302 ] combining these results with the expression for the minigap @xcite @xmath1105 , from the condition ( [ picond ] ) we observe that in the odd case the @xmath870-junction state is realized provided the number of conducting channels in the junction @xmath1106 is smaller than @xmath1107 this condition is not very restrictive and it can certainly be achieved in various experiments . for sufficiently dirty junctions
it allows for a formation of the @xmath870-junction state even in the many channel limit .
the condition ( [ condn ] ) can also be rewritten as @xmath1108 , where @xmath1109 is dimensionless conductance of a normal wire .
the condition for the presence of spontaneous currents in the ground state of sns rings with an odd number of electrons is established analogously , one should only take into account an additional energy of the magnetic field produced by pc circulating inside the ring .
the ground state with spontaneous currents is possible provided the total energy of the ring @xmath1110 becomes negative , i.e. @xmath1111 + \frac{(\phi_{0}/2)^{2}}{2\mathcal{l } } < 0.\ ] ] this condition is more stringent than that for the @xmath870-junction state , but can also be satisfied provided @xmath1046 exceeds a certain threshold value which can roughly be estimated as @xmath1112 .
we conclude that in the diffusive limit the current - phase relation in the odd case is entirely different from that in the ballistic case .
also the restriction on the number of conducting channels @xmath1106 in the normal metal ( [ condn ] ) is less stringent .
this feature of diffusive sns rings is rather advantageous for possible experimental observation of the effects discussed here .
still , in practice it would be necessary to fabricate sns rings with few conducting channels in the normal wire @xmath1113 .
this condition can hardly be met for conventional normal metals .
it appears , therefore , that most promising candidates for practical realization of such structures are junctions with n - layers formed by carbon nanotubes or organic molecules . in this respect
it is important to point out that observations of clear signatures of dc josephson effect in superconducting junctions with a weak link formed by carbon nanotubes were reported by several experimental groups @xcite also intrinsic superconductivity in carbon was claimed @xcite , though at this stage it appears that more experimental support for these claims would be desirable .
regardless of this latter issue , sns junctions with carbon nanotubes are most likely objects in which it would be possible to observe the influence of the parity effect on persistent currents in superconducting nanorings .
it is well established that fluctuations play an important role in structures with reduced dimensionality . in superconducting materials
clear signatures of fluctuation effects exist already above the critical temperature @xmath4 @xcite .
also @xmath4 itself can be reduced due to fluctuation effects .
this reduction is particularly important in disordered 2d films and quasi-1d wires @xcite . in this paper
we addressed fluctuation effects which occur in ultra - thin superconducting wires at temperatures below the mean field bcs critical temperature .
superconducting properties of such systems have been intensively studied both theoretically and experimentally during past years .
the key conclusions of these investigations can be summarized as follows .
thicker superconducting wires are characterized by very small ginzburg numbers @xmath11141 and diameters typically @xmath809 100 nm . in such systems
the superconducting transition is broadened due to thermally activated phase slips ( taps ) @xcite which cause non - zero resistance @xmath3 at temperatures close enough to the critical temperature @xmath1115 . upon decreasing
temperature taps events become less likely and quantum fluctuations of the order parameter take over .
this is the regime of quantum phase slips ( qps ) which sets in at @xmath43 .
as long as the wire is sufficiently thick and the ginzburg number remains very small , @xmath11141 , qps events are rare and typically do not lead to any measurable consequences .
hence , the behavior of thicker wires remains essentially superconducting outside an immediate vicinity of the critical temperature . upon reduction of the wire diameter below @xmath70450
nm the qps rate increases drastically .
in this regime the wire resistance @xmath3 still decreases with temperature but may remain well in the measurable range down to very low @xmath7 . provided the wire diameter is decreased further the dimensionless conductance of a wire segment of length @xmath49 eventually becomes of order @xmath111610 or smaller .
in such wires qps proliferate causing a sharp _ crossover _ from a superconducting to a normal behavior . for generic parameters
this crossover is expected for wire diameters in the 10 nm range @xcite .
this crossover was indeed observed in a number of experiments in wires with thicknesses exactly in this range .
thus , _ intrinsic superconductivity in wires with diameters @xmath828 10 nm is destroyed by quantum fluctuations of the order parameter at any temperature down to _ @xmath2 .
theoretical analysis of nanowires reveals further interesting effects , like qps - binding - unbinding quantum phase transition ( qpt ) @xcite between superconducting and non - superconducting phases which is predicted to occur as the impedance of a superconducting wire becomes of order of the quantum resistance unit @xmath1117 6.5 k@xmath57 . for typical parameters
this condition is also achieved for wire diameters in the 10 nm range . to the best of our knowledge , no clear experimental evidence for this phase transition exists so far .
this can be due to rather stringent requirements : qpt can only be observed in long nanowires at sufficiently low temperatures , ideally at @xmath45 .
interesting effects may occur also in short nanowires forming weak links between superconducting electrodes . in many respects such systems
can behave similarly to josephson nanojunctions and weak links .
novel effects are also expected in superconducting nanorings . while rings formed by thicker wires demonstrate the standard behavior familiar from the bulk samples , the situation changes drastically as soon as the wire thickness
gets reduced down to @xmath1118 10 nm or below this value . in this case
qps effects become important leading to strong fluctuations of the magnetic flux inside the ring . as a result , the amplitude of persistent current decreases and its flux dependence changes from the sawtooth - like to a smoother one @xcite . for such rings
even at @xmath2 persistent current gets exponentially suppressed by quantum fluctuations provided the ring radius @xmath26 exceeds the critical scale @xmath1119 where @xmath1120 is the action of the qps core . for typical wire parameters the length @xmath928 can be of order of a micron or even smaller .
this length constitutes another fundamental scale associated with quantum phase slips .
yet another important factor which may influence persistent currents in isolated superconducting nanorings is the electron parity number .
this influence is particularly strong in nanorings containing a weak link with few conducting modes .
changing the electron parity number from even to odd may result in _ spontaneous _ supercurrent in the ground state of such rings without any externally applied magnetic flux @xcite . at @xmath2
this current is produced by the only unpaired electron which occupies the lowest available andreev state . under certain conditions
this spontaneous supercurrent can reach remarkably large values up to @xmath1121 na which can be reliably detected in modern experiments .
we gratefully acknowledge collaboration and/or helpful discussions with a.a .
bezryadin , g. blatter , a.v .
galaktionov , m.s .
kalenkov , c.n .
lau , j.e .
mooij , yu.v .
nazarov , y. oreg , a. van otterlo , l. pryadko , g. schn , s.v .
sharov , m. tinkham , d. vodolazov and g.t .
this work was supported in part by the european community s framework programme nmp4-ct-2003 - 505457 ultra-1d `` experimental and theoretical investigation of electron transport in ultra - narrow 1-dimensional nanostructures '' and by rfbr grant 06 - 02 - 17459 .
we formally define the sub - blocks @xmath112 in the green - keldysh matrix ( [ gggg ] ) by means of the following equations : @xmath1122{cc}\langle{\mathcal{t}}\hat\psi_{\uparrow}(x_{1})\hat\psi^{\dagger}_{\uparrow } ( x_{2})\rangle & -\langle\hat\psi^{\dagger}_{\uparrow}(x_{2})\hat \psi_{\uparrow}(x_{1})\rangle\\ \langle\hat\psi_{\uparrow}(x_{1})\hat\psi_{\uparrow}^{\dagger}(x_{2})\rangle & \langle{\mathcal{t}}^{-1}\hat\psi_{\uparrow}(x_{1})\hat\psi_{\uparrow } ^{\dagger}(x_{2})\rangle \end{array } \right ) , \nonumber\\ \hat f=\frac{1}{i}\left ( \begin{array } [ c]{cc}\langle{\mathcal{t}}\hat\psi_{\uparrow}(x_{1})\hat\psi_{\downarrow}(x_{2})\rangle & -\langle\hat\psi_{\downarrow}(x_{2})\hat\psi_{\uparrow}(x_{1})\rangle\\ \langle\hat\psi_{\uparrow}(x_{1})\hat\psi_{\downarrow}(x_{2})\rangle & \langle{\mathcal{t}}^{-1}\hat\psi_{\uparrow}(x_{1})\hat\psi_{\downarrow}(x_{2})\rangle \end{array } \right ) , \nonumber\\ \hat{\bar f}=\frac{1}{i}\left ( \begin{array } [ c]{cc}\langle{\mathcal{t}}\hat\psi_{\downarrow}^{\dagger}(x_{1})\hat\psi_{\uparrow } ^{\dagger}(x_{2})\rangle & -\langle\hat\psi_{\uparrow}^{\dagger}(x_{2})\hat\psi_{\downarrow}^{\dagger}(x_{1})\rangle\\ \langle\hat\psi_{\downarrow}^{\dagger}(x_{1})\hat\psi_{\uparrow}^{\dagger } ( x_{2})\rangle & \langle{\mathcal{t}}^{-1}\hat\psi_{\downarrow}^{\dagger } ( x_{1})\hat\psi_{\uparrow}^{\dagger}(x_{2})\rangle \end{array } \right ) , \nonumber\\ \hat{\bar g}=\frac{1}{i}\left ( \begin{array } [ c]{cc}\langle{\mathcal{t}}\hat\psi^{\dagger}_{\downarrow}(x_{1})\hat\psi _ { \downarrow}(x_{2})\rangle & -\langle\hat\psi_{\downarrow}(x_{2})\hat \psi^{\dagger}_{\downarrow}(x_{1})\rangle\\ \langle\hat\psi^{\dagger}_{\downarrow}(x_{1})\hat\psi_{\downarrow}(x_{2})\rangle & \langle{\mathcal{t}}^{-1}\hat\psi^{\dagger}_{\downarrow } ( x_{1})\hat\psi_{\downarrow}(x_{2})\rangle \end{array } \right ) .\nonumber\end{aligned}\ ] ] here @xmath1123 are the electron annihilation and creation operators , the symbol @xmath1124 ( @xmath1125 ) stands for the time ( anti - time ) ordering .
let us introduce the complete basis of the eigenfunctions @xmath1126 of the single electron hamiltonian @xmath1127 : @xmath1128 here @xmath1129 are the electron eneries in the normal state . defining the quasiparticle energies in the superconducting state @xmath1130 as well as the bcs coherence factors @xmath1131 we arrive at the explicit expressions for the sub - blocks @xmath1132 : @xmath1133 , \nonumber\\ i\hat{\bar g}(t,\bm{r}_{1},\bm{r}_{2 } ) & = \sum_{n}\chi_{n}(\bm{r}_{1})\chi _ { n}(\bm{r}_{2 } ) \left [ v_{n}^{2}\hat a_{n}(t ) + u_{n}^{2}\hat b_{n}(t)\right ] , \nonumber\\ i\hat f(t,\bm{r}_{1},\bm{r}_{2 } ) & = i\hat{\bar f}(t,\bm{r}_{1},\bm{r}_{2 } ) = \sum_{n}\chi_{n}(\bm{r}_{1})\chi_{n}(\bm{r}_{2 } ) u_{n}v_{n } \nonumber\\ & \times\ , \left [ \hat a_{n}(t ) - \hat b_{n}(t)\right ] , \label{g}\ ] ] where @xmath1134{cc}\theta(t)-f_{n } & -f_{n}\\ 1-f_{n } & \theta(-t)-f_{n}\end{array } \right ) , \nonumber\\ \hat b_{n}(t ) & = e^{ie_{n}t}\left ( \begin{array } [ c]{cc}f_{n}-\theta(-t ) & -(1-f_{n})\\ f_{n } & f_{n}-\theta(t ) \end{array } \right ) .\end{aligned}\ ] ] here we introduced the occupation probabilities of the quasiparticle states @xmath1135 . in thermodynamic equilibrium these filling factors
take the universal form @xmath1136
one can demonstrate that the matrix green - keldysh functions satisfy the following ward identities @xmath1137 @xmath1138 where @xmath1139{cc}\beta_{f}(t,{\bm r } ) & 0\\ 0 & \beta_{b}(t,{\bm r } ) \end{array } \right)\end{aligned}\ ] ] is an arbitrary diagonal @xmath90 matrix .
explicit expressions for the functions @xmath168 read @xmath1140 @xmath1141 @xmath1142 @xmath1143 here we defined @xmath1144 , @xmath1145)$ ] . in a number of limiting cases the kernels @xmath319 can be evaluated exactly .
first let us set @xmath2 and consider the limit of small wave vectors @xmath1146 .
this regime is relevant , e.g. , in the context of microwave absorption . at @xmath1147
we find @xmath1148 , \end{aligned}\ ] ] while at @xmath1149 one finds @xmath1150 , \nonumber\\ \chi_{j } & = -\frac{8e^{2}n_{0}\delta_{0}^{2}}{\omega\sqrt{\omega^{2}-4\delta_{0}^{2}}}\left [ \ln\left ( \frac{\omega}{2\delta_{0}}+\sqrt { \frac{\omega^{2}}{4\delta_{0}^{2}}-1}\right ) -i\frac{\pi}{2}\right ] , \nonumber\\ \chi_{l } & = \frac{2m^{2}\sigma_{d}\delta_{0}}{e^{2}}\left [ \frac{2\delta _ { 0}}{\omega}k\left ( \frac{2\delta_{0}}{\omega}\right ) \right .
\nonumber\\ & \left .
+ \,i\frac{4\delta_{0}}{\omega+2\delta_{0}}k\left ( \frac{\omega-2\delta_{0}}{\omega+2\delta_{0}}\right ) \right ] , \nonumber\\ \chi_{d } & = \frac{\sigma_{d}}{\omega}\left [ k\left ( \frac{2\delta_{0}}{\omega}\right ) -e\left ( \frac{2\delta_{0}}{\omega}\right ) \right ] \nonumber\\ & + \,i\frac{\sigma_{d}}{\omega}\left [ \left ( 1+\frac{2\delta_{0}}{\omega } \right ) e\left ( \frac{\omega-2\delta_{0}}{\omega+2\delta_{0}}\right ) \right .
\nonumber\\ & \left .
-\,\frac{4\delta_{0}}{\omega+2\delta_{0}}k\left ( \frac{\omega-2\delta_{0}}{\omega+2\delta_{0}}\right ) \right ] .\end{aligned}\ ] ] here and below @xmath1151 and @xmath1152 are complete elliptic integrals of the first and second kind respectively . utilizing the relation ( [ qsigma ] ) one can verify that the expressions for @xmath169 and @xmath170 agree with the well known results @xcite , @xcite . at temperatures close to @xmath4 one
can derive analytic expressions for the kernels in the limit @xmath1153 . at @xmath1147
these expressions read @xmath1154 \nonumber\\ & + \,i\frac{n_{0}\delta_{0}}{t}\frac{\omega}{dq^{2 } } \left [ \left ( 1-\frac{\omega^{2}}{4\delta_{0}^{2}}\right ) k\left ( \sqrt{1-\frac{\omega^{2}}{4\delta_{0}^{2}}}\right ) \right .
\nonumber\\ & \left .
+ \,\left ( \frac{\omega^{2}}{4\delta_{0}^{2}}-1+\frac{d^{2}q^{4}}{\omega^{2}+d^{2}q^{4}}\right ) \right .
\nonumber\\ & \times\ , \left .
\pi\left ( \frac{d^{2}q^{4}}{\omega^{2}+d^{2}q^{4}},\sqrt{1-\frac{\omega^{2}}{4\delta_{0}^{2 } } } \right ) \right ] , \label{chideltaap1}\ ] ] @xmath1155 , \label{chijap1}\ ] ] @xmath1156 \nonumber\\ & -\,i\frac{m^{2}n_{0}d\delta_{0}}{t}\frac{\omega^{3}}{\omega^{2}+d^{2}q^{4 } } \pi\left ( \frac{d^{2}q^{4}}{\omega^{2}+d^{2}q^{4}},\sqrt{1-\frac{\omega^{2}}{4\delta_{0}^{2}}}\right ) .\label{chilap1}\ ] ] here @xmath1157 is the complete elliptic integral of the third kind . at higher frequencies
@xmath1149 we find @xmath1158 \nonumber\\ & -\,i\frac{n_{0}}{2t}\frac{\omega } { |\omega|}\frac{\omega^{2}+d^{2}q^{4}-4\delta_{0}^{2}}{dq^{2 } } \nonumber\\ & \times\ , \left [ k\left ( \sqrt{1-\frac{4\delta_{0}^{2}}{\omega^{2}}}\right ) -\pi\left ( -\frac{d^{2}q^{4}}{\omega^{2}},\sqrt{1-\frac{4\delta_{0}^{2}}{\omega^{2}}}\right ) \right .
\nonumber\\ & \left . -\ , \pi\left ( \frac{4\delta_{0}^{2}-\omega^{2}}{d^{2}q^{4}},\sqrt{1-\frac{4\delta_{0}^{2}}{\omega^{2}}}\right ) \right ] , \label{chideltaap}\ ] ] @xmath1159 , \label{chijap}\ ]
] @xmath1160 \nonumber\\ & + \,i \frac{2m^{2}n_{0}d\delta_{0}^{2}}{t}\frac{\omega}{|\omega| } \left[k\left ( \sqrt{1-\frac{4\delta_{0}^{2}}{\omega^{2}}}\right ) \right . \nonumber\\ & -\pi\left ( -\frac{d^{2}q^{4}}{\omega^{2}},\sqrt{1-\frac{4\delta_{0}^{2}}{\omega^{2}}}\right ) \nonumber\\ & \left . -\
, \pi\left ( \frac{4\delta_{0}^{2}-\omega^{2}}{d^{2}q^{4}},\sqrt{1-\frac{4\delta_{0}^{2}}{\omega^{2}}}\right ) \right ] .
\label{chilap}\end{aligned}\ ] ]
integrating out the electromagnetic fields in eq .
( [ a105 ] ) we arrive at the action ( [ a116 ] ) where the function @xmath325 is defined by the following general expression @xmath1161 \right ) + \frac{\tilde{\chi}_{j}\tilde{\chi}_{l}}{4m^{2}}\left [ c\frac{\omega^{2}}{c^2 } + \frac{q^{2}}{l}\right ] } { \left ( \frac{c}{s}+\tilde{\chi}_{j}+\tilde{\chi } _ { d}q^{2}\right ) \left ( \frac{1}{sl}+\tilde{\chi}_{e}\frac{\omega^{2}}{c^2}+\frac { e^{2}}{m^{2}c^2}\tilde{\chi}_{l}\right ) -\tilde{\chi}_{d}^{2}\frac{\omega^{2}q^{2}}{c^2 } } .\label{calf}\ ] ] the electromagnetic potentials are linked to the fluctuating phase of the order parameter field via the following saddle point conditions @xmath1162 | superconducting properties of metallic nanowires can be entirely different from those of bulk superconductors because of the dominating role played by thermal and quantum fluctuations of the order parameter . for superconducting wires with diameters below @xmath0
nm quantum phase slippage is an important process which can yield a non - vanishing wire resistance down to very low temperatures .
further decrease of the wire diameter , for typical material parameters down to @xmath1 nm , results in proliferation of quantum phase slips causing a sharp crossover from superconducting to normal behavior even at @xmath2 .
a number of interesting phenomena associated both with quantum phase slips and with the parity effect occur in superconducting nanorings .
we review recent theoretical and experimental activities in the field and demonstrate dramatic progress in understanding of the phenomenon of superconductivity in quasi - one - dimensional nanostructures . |
gravitationally lensed quasars are very well suited to study the inner structure of agn @xcite .
the magnification of lensed quasar images depends on the mass distribution of the lens galaxy and on geometrical considerations ( distances and alignment between quasar , galaxy and observer ) .
therefore , the mean ( or macromodel ) flux ratios between images would be very useful observables to study the mass density profile in lens galaxies , assuming that all the lens galaxy mass is smoothly distributed .
however , compact objects such as stars in the lens galaxies induce strong spatial gradients in the gravitational potential that give rise to anomalies in the flux ratios compared to the predictions of macromodels .
this effect ( so - called microlensing ; see * ? ? ?
* ; * ? ? ?
* ) complicates the macro - modeling of lens systems but in exchange it has a very useful property : it is sensitive to the source size ( with smaller source regions showing larger magnifications ) .
thus , we can study the size of the emitting region by measuring flux ratios to deduce the effects of microlensing .
taking advantage of this property , the inner structure of agns can be studied by searching for the effects of microlensing on different regions of single - epoch spectra of the images of a lensed quasar @xcite .
the cores of the emission lines , which likely arise from extended regions , is insensitive to microlensing , so their flux ratios are a baseline against which we can measure the effects of microlensing on other regions of smaller size , such as the wings of the emission lines @xcite and the continuum generated by the accretion disk . according to the thin disk model @xcite , the size of the continuum varies with wavelength with a @xmath10 law , and hence the microlensing magnification will show some wavelength dependence ( so - called microlensing chromaticity ) .
the objective of this paper is to use single - epoch spectra of three gravitationally lensed quasars ( he0435 - 1223 , wfi2033 - 4723 , and he2149 - 2745 ) to discuss the presence of microlensing in the emission lines ( which would yield information on the structure of the blr ) and in the continuum ( to estimate sizes and temperature gradients in the accretion disk ) .
the paper is organized as follows . in section 2
we present the data .
section 3 is devoted to a description of our method of analysis .
we discuss our results in 4 and offer concluding remarks in section 5 .
he0435 - 12223 was observed on january 12th 2008 with the blue channel spectrograph on the mmt .
spectroscopic information for wfi2033 - 4723 and he2149 - 2745 was gathered during april - may 2008 with the fors2 spectrograph at the very large telescope ( vlt ) .
table [ obs ] summarizes the main observational characteristics of the data . for he0435 - 1223
, we also used archival data obtained with the fors1 spectrograph on the vlt .
for the three systems we also analyzed the deconvolved spectra from @xcite provided by the vizier @xcite catalogue . a detailed description of these observations and the spectrum analysis can be found in @xcite and @xcite .
lccccccccr he0435 - 1223 & bd & 1.5 & mmt / blue - channel & 300 & 2008/01/12 & 1.40 & -13.92 & 0.87 & 1800 + & bd & & vlt / fors1 & 300 & 2004/10/11 & & & & 1400 + & bd & & vlt / fors1 & 300 & 2004/10/12 & & & & 1400 + & bd & & vlt / fors1 & 300 & 2004/11/11 & & & & 1400 + wfi2033 - 4723 & bc & 2.1 & vlt / fors2 & 300 & 2008/04/14 & 1.23 & -79.05 & 0.7 & @xmath11 + & bc & & vlt / fors2 & 300 & 2005/05/13 & 1.16 & & 0.54 & @xmath12 + he2149 - 2745 & ab & 1.7 & vlt / fors2 & 300 & 2008/05/07 & 1.38 & -28.63 & 0.8 & @xmath13 + & ab & & vlt / fors2 & 300 & 2006/08/04 & 1.48 & -32.0 & 0.62 & @xmath14 + data reductions were carried out with iraf tasks . the procedure consisted of subtraction , flat fielding , and wavelength calibration . as we are interested only in flux ratios between the quasar images ,
flux - calibration is not needed .
the cosmic - rays were removed using at least three exposures .
the 1-d spectra extraction is obtained by simultaneously fitting two gaussian functions to the components for each wavelength .
the deconvolved spectra obtained from the vizier archive are already fully reduced .
the systematic errors affecting our measurements were discussed elsewhere @xcite .
the procedure we use to separate microlensing and extinction consist of measuring the displacement between the continuum and the core of the emission line flux ratios ( e.g. , see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
thus , the baseline for no - microlensing is established by using the line core fluxes .
the continuum is retrieved by fitting the regions on either side of each emission line ( @xmath15 to @xmath16 wavelength range ) and its flux is the integral below such function ( @xmath17 ) .
for instance , when @xmath18 is used as fitting function , the integrated flux is obtained as @xmath19 .
once the continuum is subtracted , the core flux is obtained by integrating the emission line profiles using dipso @xcite in starlink .
the integration is performed in a narrow interval ( from 20 to 100 depending on the line profile shape ) centered on the peak of line .
narrower integration windows are chosen in those cases in which absorption lines are present ( e.g. 20 for in he2149 - 2745 ) .
the continuum fitting error ( @xmath20 , @xmath21 ) provided by dipso ( at 1@xmath22 level ) is used as an estimation of the core flux error .
specifically , the errors for each coefficient ( @xmath20 , @xmath21 ) are used to estimate the error in the continuum as @xmath23 .
as the core flux measurement relies on the continuum fitting , its error is estimated as the error in the continuum ( see * ? ? ?
we also compare our results with magnitude differences obtained from the literature ( table [ lit ] ) , measured in the near - infrared .
longer wavelengths are expected to be less affected by microlensing because they are produced in a larger emitting region ( however , as @xcite have stated , for sources with @xmath24 the @xmath25 broad - band could be contaminated by thermal emission from the inner dusty torus ) lcccccr he0435 - 1223 bd & 0.46 & 1.689 & l & 0.26 & @xmath26 & 5 + & & & k & 0.45 & @xmath27 & 5 + & & & ks & 0.45 & @xmath27 & 9 + & & & h & 0.61 & @xmath28 & 9 + & & & f160w & 0.65 & @xmath27 & 1 + & & & j & 0.80 & @xmath29 & 9 + & & & z & 1.00 & @xmath30 & 9 + & & & f814w & 1.23 & @xmath31 & 1 + & & & i & 1.27 & @xmath32 & 4 + & & & i & 1.27 & @xmath33 & 4 + & & & i & 1.30 & @xmath34 & 2 + & & & i & 1.30 & @xmath35 & 9 + & & & iac29 & 1.43 & @xmath36 & 4 + & & & h@xmath37 & 1.52 & @xmath38 & 4 + & & & h@xmath37 & 1.52 & @xmath39 & 4 + & & & r & 1.60 & @xmath40 & 2 + & & & r & 1.60 & @xmath40 & 9 + & & & iac28 & 1.65 & @xmath41 & 4 + & & & r & 1.60 & @xmath42 & 3 + & & & f555w & 1.80 & @xmath43 & 1 + & & & v & 1.83 & @xmath42 & 3 + & & & str - y & 1.83 & @xmath44 & 4 + & & & str - y & 1.83 & @xmath45 & 4 + & & & g & 2.08 & @xmath46 & 2 + & & & g & 2.08 & @xmath47 & 9 + & & & g & 2.08 & @xmath48 & 3 + & & & str - b & 2.14 & @xmath43 & 4 + & & & str - v & 2.43 & @xmath34 & 4 + & & & str - v & 2.43 & @xmath49 & 4 + & & & u & 2.84 & @xmath50 & 9 + & & & str - u & 2.85 & @xmath51 & 4 + & & & 0.5 - 0.8kev & 3.23 & @xmath52 & 9 + & & & 0.4 - 0.8kev & 3.23 & @xmath53 & 12 + wfi2033 - 4723 bc & 0.66 & 1.66 & ks & 0.75 & @xmath54 & 9 + & & & h & 1.00 & @xmath55 & 9 + & & & f160w & 1.07 & @xmath56 & 1 + & & & j & 1.33 & @xmath57 & 9 + & & & z & 1.82 & @xmath58 & 9 + & & & f814w & 2.07 & @xmath59 & 1 + & & & i & 2.16 & @xmath60 & 9 + & & & i & 2.16 & @xmath61 & 10 + & & & r & 2.57 & @xmath62 & 11 + & & & r & 2.65 & @xmath63 & 9 + & & & r & 2.65 & @xmath64 & 10 + & & & f555w & 3.05 & @xmath65 & 1 + & & & g & 3.46 & @xmath66 & 9 + & & & g & 3.46 & @xmath67 & 10 + & & & u & 4.71 & @xmath68 & 9 + & & & u & 4.71 & @xmath69 & 10 + & & & 0.5 - 0.8kev & 3.23 & @xmath70 & 10 + he2149 - 2745 ab & 0.60 & 2.032 & l & 0.26 & @xmath71 & 5 + & & & k & 0.45 & @xmath72 & 5 + & & & f160w & 0.65 & @xmath73 & 1 + & & & f814w & 1.23 & @xmath74 & 1 + & & & i & 1.30 & @xmath75 & 8 + & & & r & 1.55 & @xmath76 & 6 + & & & r & 1.55 & @xmath77 & 7 + & & & v & 1.83 & @xmath78 & 7 + & & & v & 1.83 & @xmath79 & 8 + & & & v & 1.83 & @xmath80 & 8 + & & & f555w & 1.80 & @xmath81 & 1 + & & & b & 2.28 & @xmath82 & 6 + & & & b & 2.28 & @xmath83 & 7 + in those cases where our spectra are contaminated by the lens galaxy continuum ( e.g. he2149 - 2745 ) , we compare our line core flux ratio with the uncontaminated broad - band measurements ( e.g. castles ) .
chromatic microlensing detection allows us to study the structure of the accretion disk in the lensed quasar by estimating its size and temperature profile .
the accretion disk is modeled as a gaussian , @xmath84 , with size @xmath85 .
@xcite showed that microlensing primarily constrains the half - light radius of the disk @xmath86 , but the precise details of the disk model are unimportant , because the gaussian scale length @xmath87 we use here is related to the half - light radius by @xmath88 and this can be used to compare our results to other choices for the disk profile .
the probability of reproducing the measured microlensing magnifications is estimated by randomly placing a gaussian source on microlensing magnification maps .
the inverse polygon mapping method @xcite was used to compute maps of @xmath89 pixels of 0.5 light days for each image .
microlensing magnification statistics is degenerate with the single - mass case except in the case of a mass function with a high dynamical range , when a similar contribution to the mass density from microlenses at the opposite ends exists ( bimodality ) .
previous works @xcite support that microlensing magnification does not depend on the details of the mass function , but rather on the mean mass of the microlenses through the scaling factor @xmath90 , at least for salpeter s like laws ( but see also @xcite for more general stellar mass functions ) .
thus , we use @xmath91 microlenses and the size scales linearly with this mass , i.e. @xmath92 . the convergence ( @xmath93 ) and shear ( @xmath94 ) for each image are selected using macromodels from the literature @xcite .
the fraction of mass in compact objects is assumed to be @xmath95 , an acceptable value considering current estimates ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
bayesian posterior probabilities @xcite of @xmath87 and @xmath5 conditioned to the measured microlensing magnifications are estimated using a uniform logarithmic ( linear ) prior on @xmath87 ( @xmath5 ) .
discovered by @xcite , it consists of four images of a lensed quasar at @xmath96 and a lens galaxy at @xmath97 @xcite .
castles images obtained with hst also revealed multiple partial arcs between the images .
we present spectra for the @xmath98 and @xmath99 images taken with mmt and vlt telescopes . due to the symmetric distribution of the images ,
the time delay between images is small ( @xmath100 days ; * ? ? ?
* ; * ? ? ?
in figure [ prof_he0435bd ] we present the continuum subtracted spectra in the regions corresponding to the , and emission lines .
the spectra are normalized to match the profiles in the region of the core of the line ( see 3 ) .
a very interesting result in the high ionization line is the presence of a slight but systematic enhancement of the red wing ( @xmath101 $ ] range ) of component @xmath99 with respect to component @xmath98 .
a similar enhancement of the red wing is hinted at in component @xmath102 ( not plotted here ) with respect to component @xmath98 , but with less strength .
profiles of images @xmath98 and @xmath103 ( not plotted here ) seem to match reasonably well . in the case of the low ionization line , the red wing of image @xmath99 also seems to be brighter than that of image @xmath98 , but by a small amount .
finally , the lower ionization line shows no significant differences between image profiles .
+ + differences present at specific wavelengths in the line profiles and the dependence of its strength on the degree of ionization ( decreasing in the , , sequence ) can be explained as microlensing acting selectively on parts of a broad line region with organized kinematics . @xmath98 and @xmath103 would be the images less affected by differential microlensing and @xmath99 the most affected . from near ir observations of the h@xmath37 emission line ,
@xcite also propose microlensing to explain the differences between profiles .
they found that the h@xmath37 emission line profiles of images @xmath98 and @xmath103 match well and that the effect of microlensing seems to be more pronounced for image @xmath99 .
however , their figure 1 shows an excess in the image @xmath98 profile in the blue wing relative to @xmath99 ( instead of the red wing excess of @xmath99 that we measure ) .
this is likely due to the different approach they followed to compare the profiles .
they normalize the continuum before superposing them .
thus , their normalization factors include macro - magnification and continuum microlensing . in this way
, they are mixing these two effects with line microlensing in the comparison of two emission line profiles .
following our procedure , on the contrary , continuum microlensing is removed by continuum subtraction , and macro - magnification is corrected for by normalization to the core of the emission line ( assumed to be insensitive to microlensing ) .
in fact , the re - normalization of the emission line profiles of @xcite to match the core of the lines would likely result in an enhancement of the red wing of the @xmath99 emission line profile with respect to that of @xmath98 . in the same manner
, the blue wing enhancements reported by @xcite in the emission line will also likely disappear after normalization to the line cores . integrating the red wing excess in corresponding to image @xmath99 and @xmath98 in the @xmath104 $ ] range ,
we obtain the microlensing magnification associated with the region .
the red wing magnification was obtained as ( see * ? ? ?
* ) : @xmath105}-(m_b - m_d)_{core}=(-0.12\pm0.03)-(-0.37\pm0.02)=(0.25\pm0.04)$ ] mag . in spite of the presence of microlensing in the red wings of some line profiles , the ratio of the line cores ( see figure [ diff_he0435a ] and table [ mag_he0435 ] ) agrees within uncertainties .
we obtain an average value for the emission lines ratio of @xmath106 mag .
this confirms the expectation that the cores are not very sensitive to microlensing and that extinction is not significantly present in this lens system ( in agreement with * ? ? ?
llcrr continuum & 4170 & 4000 - 4350 & @xmath107 & + & 5140 & 4540 - 5550 & @xmath108 & @xmath109 + & 7560 & 7130 - 7880 & @xmath110 & @xmath111 + line & @xmath1121549 & 4170 - 4195 & @xmath113 & + & red wing & 4200 - 4250 & @xmath114 & + & @xmath1121909 & 5100 - 5180 & @xmath115 & @xmath116 + & @xmath1122800 & 7480 - 7580 & @xmath117 & @xmath118 + continuum observations also indicate that , in contrast with @xmath98 , @xmath102 and @xmath99 are affected by microlensing .
@xcite conclude that the @xmath98 image is the least affected by stellar microlensing while @xmath102 is affected by strong microlensing variations .
@xcite found evidence of chromatic microlensing in @xmath99 component ( 0.07 mag ) between dec .
2001 and sep . 2002 .
using lightcurves in the r filter , @xcite also observed microlensing in @xmath99 ( relative to @xmath102 ) of @xmath119 mag yr@xmath120 . @xcite
obtained lightcurves with narrow - band filters in oct .
2007 , observing a chromaticity between the bluest ( str - b ) and the reddest ( i - band ) filters of @xmath121 mag affecting @xmath102 component .
@xcite observed the system in the i , v , r bands in two epochs concluding that image @xmath102 is probably affected by microlensing . in figure [ diff_he0435a ]
we show the continuum ratio values available from the literature and those obtained from the spectra used in the present work . in the wavelength region between and
all these data consistently depict a linear trend with a variable gap with respect to the no microlensing baseline defined by the emission line ratios of about 0.1 and 0.3 mag respectively at the red ( ) and blue ( ) ends . from towards the uv ,
the broad band data available seem to fit in this linear trend albeit with a large scatter . from towards the ir ,
the chromaticity disappears and , with the exception of the outlying k band data from @xcite , the broad band data seem to concentrate around a constant offset of @xmath1220.18 mag with respect to the baseline defined by the emission line ratios .
following the method described in 3 we will use these wavelength dependent microlensing measurements to estimate the size and temperature profile of the accretion disk . to make the problem manageable
, we will consider that the global behavior of the continuum can be defined by two straight lines , one independent of lambda to describe the data redder than , and the other one following the almost linear dependence of microlensing with @xmath123 from towards the blue ( see figure [ diff_he0435b ] ) .
we will take three points corresponding in wavelength to the f160w band , 8100 ( from spectroscopic continuum ) , and u band ( see table [ map_he0435 ] ) to describe the global dependence of microlensing with wavelength in the simplest possible way . in figure [ he0435size ]
we show the 2d joint pdf of @xmath87 and @xmath5 conditioned to these three microlensing measurements .
the resulting estimates are @xmath124 light - days at @xmath125 and @xmath126 ( @xmath127 level ) , using uniform logarithmic and linear priors for the size and @xmath5 respectively .
the estimated size ( @xmath128 light - days or @xmath129 light - days , scaled to @xmath130 ) is in agreement with the average size estimated by @xcite ( @xmath131 light - days at 1026 ) , and it is consistent with the results in @xcite ( @xmath132 or @xmath133 light - days at 1208 ) .
this is also consistent with sizes predicted by the black hole - mass size correlation ( * ? ? ?
* @xmath134 at 2500 for @xmath135 )
. however , this size is large compared to predictions based on flux variations ( * ? ? ?
* @xmath136 or 0.3 light - days at 3027 ) , and recent results using r - band light - curves ( * ? ? ?
* @xmath137 or @xmath138 light - days at 3027 ) .
our results for @xmath5 are in agreement with previous results ( ranging from @xmath139 in @xcite to @xmath140 in @xcite ) .
notice that these results match those of @xcite although those were estimated using pair @xmath141 .
we follow the same procedure @xcite to estimate the size of the emitting region affected by microlensing . in this case , we only calculate the size of the region at the wavelength .
we obtained @xmath142 light - days .
this size is in agreement with that of the region emitting the blue continuum , indicating that the microlensed emission likely arises close to the accretion disk .
rc 3522 & @xmath143 + 8100 & @xmath144 + 15500 & @xmath145 + this is a quadruply imaged quasar ( @xmath146 by our estimation ) discovered by @xcite lensed by a galaxy at @xmath147 @xcite .
we present new spectra for components @xmath98 and @xmath103 which are separated @xmath148 .
@xcite found a time delay of @xmath149 days , in agreement with values estimated by @xcite . in figure [ prof_wfi2033 ]
we plot the continuum subtracted spectra for images @xmath98 and @xmath103 in the regions corresponding to the and emission lines .
the spectra have been normalized to match the profiles in the region of the core of the line ( see 3 ) .
a very interesting result is the significant enhancement ( of the @xmath98 image compared to the @xmath103 image ) of a bump - like feature present in the blue wing of the line , coincident with an emission line typical of quasar spectra .
careful inspection of our spectra reveals that there is another bump - like feature in the red wing of the line that also appears to be brighter in @xmath98 than in @xmath103 .
+ at a much lower intensity , two microlensed bumps with a separation of about @xmath150 are also noticeable at both sides of the emission lines . in the @xcite spectra
, the emission line profiles also show a relative enhancement of the @xmath98 image with respect to @xmath103 .
this was interpreted as evidence of microlensing by @xcite , who indicated that the excess might also be due to microlensing of the or lines blended with the blue wing of .
unfortunately , the s / n spectra of the red side of and of the lines are not sufficient to look for further evidence of these features .
llcr continuum & 5140 & 4700 - 5450 & @xmath151 + & 7560 & 7200 - 7800 & @xmath151 + line & @xmath1121909 & 5060 - 5100 & @xmath152 + & blue wing & 4930 - 5000 & @xmath153 + & red wing & 5160 - 5250 & @xmath154 + & @xmath1122800 & 7430 - 7470 & @xmath155 + integrating the bump excess corresponding to image @xmath98 in the @xmath156 $ ] and @xmath157 $ ] ranges , we obtain the microlensing magnification associated to the region generating the blue / red wing as ( see * ? ? ?
* ) : @xmath158 .
we obtain consistent microlensing estimates for both bumps : @xmath159 mag and @xmath160 mag .
the presence of the two bumps with similar enhancements on both sides of the line suggest a common origin for both structures ( as an alternative to a microlensed line ) . in several agns ( see * ? ? ?
* and references therein ) the presence of these features in the wings of a bel , which remind us of the double - peak characteristic of kinematics confined in a plane , have been interpreted as evidence of the accretion disk surrounding the massive central black - hole . both elements ,
the separation between the bumps @xmath161 ( @xmath162 ) and the fact that only the bumps are microlensed , would be in agreement with the hypothesis that these features arise from an inner , compact region of the blr , likely residing on the accretion disk .
an alternative possibility is that the bumps arise from a region with biconical geometry @xcite .
rc 4300 & @xmath163 + 5500 & @xmath164 + 8700 & @xmath165 + 5439 & @xmath166 + 8012 & @xmath167 + 15500 & @xmath168 + 3522 & @xmath169 + 5500 & @xmath170 + 8500 & @xmath171 + 3522 & @xmath172 + 9114 & @xmath173 + 16500 & @xmath174 + in figure [ diff_wfi2033a ] ( table [ mag_wfi2033 ] )
we present the emission line and continuum ratios for our data and for other data in the literature .
the emission line ratios from @xcite and from our data agree within uncertainties and show no evidence of chromaticity .
we obtain an average value for the emission lines ratio of @xmath175 mag ( we have not considered the discrepant emission line ratios from @xcite , because they have been obtained integrating in a window of 200 ) .
the continuum ratios , however , show a great change in amplitude at bluer wavelengths ( i.e. chromaticity ) showing a varying slope from 2003 to 2008 , which can be explained by microlensing magnification of image @xmath98 ( figure [ diff_wfi2033b ] ) .
this trend is better seen if we group the data around four slopes that correspond to different epochs or observations ( green , red , blue , and black in figure [ diff_wfi2033b ] ) . in 2008
the slope of the chromaticity changes , which likely requires a combination of microlensing in both components to be explained if a small size for the blue emission is assumed .
following the method described in 3 we estimate the size , @xmath87 , and the logarithmic slope , @xmath5 , of the size dependence with wavelength for each one of the 4 epochs described above ( table [ mag_wfi2033 ] ) . in figure [ wfi2033size ]
we present the pdf of @xmath87 and @xmath5 for each one of the epochs and the combined pdf .
notice that the slope in our continuum data is small ( meaning small chromaticity within our error bars , _ black squares _ in figure [ diff_wfi2033b ] ) , i.e. parameters are largely unconstrained for this epoch ( figure [ wfi2033size ] a ) .
the resulting estimates for the combined pdf are @xmath176 light - days ( log prior ) at @xmath125 and @xmath177 ( linear prior ) .
this disk size value ( @xmath178 light - days or @xmath179 light - days , scaled to @xmath130 ) is in agreement with the average size estimated by @xcite ( @xmath131 light - days at 1026 ) and @xcite ( @xmath134 at 2500 for @xmath135 ) .
however , we can not reconcile our results ( @xmath87 and @xmath5 ) with those of @xcite ( @xmath180 light - days at 1233 , using 0.3m@xmath181 microlenses , @xmath182 ) .
although these last authors obtain size estimates that decrease with wavelength , they do not rule out positive values for @xmath5 and suggest that the anomalous flux ratios might be caused by unusual caustic patterns . + using the same procedure
, we estimate a size of @xmath183 light - days for the region generating the emission line bumps on both sides of the line . under the hypothesis of a common origin for both bumps
, we can take this size as the radius of the accretion disk , infer a velocity of 7500@xmath184 from the separation between the bumps and , assuming keplerian circular rotation , estimate a mass of @xmath185 for the supermassive black hole .
he2149 - 2745 was discovered by @xcite ; it consists of two images @xmath102 and @xmath98 separated by @xmath186 at @xmath187 .
the lens galaxy is at @xmath188 @xcite .
image @xmath98 is separated by @xmath189 from the main lens galaxy .
chromatic microlensing was detected in spectra taken by @xcite .
these authors estimated a time delay of @xmath190 days .
+ llcrr continuum & 4170 & 4000 - 4350 & @xmath107 & + & 5140 & 5000 6200 & @xmath108 & @xmath109 + & 7560 & 8250 8650 & @xmath110 & @xmath111 + line & @xmath1121549 & 4170 - 4195 & @xmath113 & + & @xmath1121909 & 5720 5830 & @xmath115 & @xmath116 + & @xmath1122800 & 8480 8540 & @xmath117 & @xmath118 + in figure [ prof_he2149 ]
we present the continuum subtracted spectra for the @xmath102 and @xmath98 images in the regions corresponding to the , and emission lines .
the spectra match very well , after normalization using @xmath191 , except for the absorption in .
@xcite also find that the absorbed fractions of the emission line do not agree once scaled and they attribute the difference to time - variable broad absorption together with a time - delay . they found a chromatic difference between the spectra ( @xmath192 for ciii and @xmath191 for ) which they attribute to dust extinction on image @xmath98 and/or intrinsic variability combined with a time delay of @xmath193 days .
flux comparison between our spectroscopic continua and castle data shows that there is contamination by the lens galaxy .
we estimate ( using f555w and f814w fluxes ) that the contamination in the b spectra could be up to 60% , while in the deconvolved spectra it could be around 30% . in the following we will use the broad - band data to obtain the magnitude difference in the continuum . in figure [ diff_he2149a ] ( table [ mag_he2149 ] )
we present the emission line and continuum ratios for our data and for other data in the literature .
the emission line ratios show no dependence with wavelength .
the continuum flux ratios , however show chromaticity , likely induced by microlensing ( figure [ diff_he2149b ] ) .
rc t 4380 & @xmath194 + 8140 & @xmath195 + 38000 & @xmath196 + using the continuum and emission line ratios , ( see 3 ) we estimate the size , @xmath87 , and the logarithmic slope , @xmath5 , of the size dependence with wavelength . in this case
we have used convergence @xmath197 , @xmath198 , and shear @xmath199 , @xmath200 @xcite to compute the magnification maps . in figure [ he2149size ]
we present the 2d pdf of @xmath87 and @xmath5 .
the resulting estimates are @xmath201 light - days ( log prior ) at @xmath125 and @xmath202 ( linear prior ) .
we have used spectroscopy of three lensed quasars , he0435 - 1223 , wfi2033 - 4723 , and he2149 - 2745 , to study their inner structure ( blr and continuum source ) as derived from microlensing magnification .
the results are : * we have detected microlensing in the emission line profiles of two of the lensed systems , he0435 - 1223 , wfi2033 - 4723 . in the case of he0435 - 1223 , we have found an enhancement of the red wing of image @xmath99 with respect to image @xmath98 in and that , adopting the same criteria for line profile comparison , can be reconciled with previous results from infrared spectroscopy obtained by @xcite . using the measured microlensing magnification we estimate a size of @xmath0 light - days for the emitting region affected by microlensing . in the case of wfi2033 - 4723 ,
we have detected microlensing in two bumps situated on the blue and red wings of , confirming and extending previous microlensing evidence found in the blue wing @xcite . in principle
, the blue bump might be associated to an emission line usually present in the red wings of the emission line .
alternatively , we could interpret the two bumps as evidence of the double - peaked profile typical of disk kinematics , assuming that part of the emission arises from the accretion disk .
using the measured microlensing magnification we estimate a size of @xmath203 light - days for the region of the disk emitting the microlensed bumps . combining this size with the velocity inferred from the wavelength separation between the bumps we obtain an estimate of @xmath204 for the mass of the central supermassive black hole .
* the ratios of the line emission cores show no evidence of chromaticity .
this excludes both significant effects of microlensing on the regions generating the cores of the emission lines , and appreciable extinction . * using the continuum and ( core ) emission line ratios , we estimate sizes of @xmath205 , @xmath206 , @xmath207 light - days , and slopes @xmath7 , @xmath8 , and @xmath9 for he0435 - 1223 , wfi2033 - 4723 , and he2149 - 2745 respectively . in the case of he0435 - 1223 and wfi2033 - 4723 , the good agreement between the sizes of the continuum and microlensed regions in the emission line wings also support the hypothesis that the latter arise from the accretion disk . * the measured continuum microlensing amplitude ( in the three systems ) and chromaticity ( in wfi2033 - 4723 and he2149 - 2745 ) are below the predictions of the thin disk model .
this results in larger disk sizes and flatter temperature gradients than expected .
we thank the anonymous referee for useful suggestions .
v.m . gratefully acknowledges support from fondecyt through grant 1120741 and centro de astrofsica de valparaso .
e.m , j.a.m . acknowledges support from mineco and junta de andaluca through grants : aya2011 - 24728 , aya2013 - 47744-c3 - 1 , aya2013 - 47744-c3 - 3-p , and fqm-108 .
jjv is supported by the project aya2014 - 53506-p financed by the spanish ministerio de economa y competividad and by the fondo europeo de desarrollo regional ( feder ) , and by project fqm-108 financed by junta de andaluca .
jam is also supported by the generalitat valenciana with the grant prometeo/2014/60 .
is supported by phd fellowship fib - uv 2015/2016 .
this research has made use of nasa s astrophysics data system , and the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
corrales , l. 2015 , , 805 , 23 hanisch , r. j. , & biemesderfer , c. d. 1989 , , 21 , 780 lamport , l. 1994 , latex : a document preparation system , 2nd edition ( boston , addison - wesley professional ) schwarz , g. j. , ness , j .- u . , osborne , j. p. , et al .
2011 , , 197 , 31 vogt , f. p. a. , dopita , m. a. , kewley , l. j. , et al .
2014 , , 793 , 127 courbin , f. , chantry , v. , revaz , y. , sluse , d. , faure , c. , tewes , m. , eulaers , e. , koleva , m. , asfandiyarov , i. , dye , s. , magain , p. , van wincke , h. , coles , j. , saha , p. , ibrahimov , m. , and meylan , g. 2011 , , 536 , 53 richards , g.t . , keeton , c.r . ,
pindor , b. , hennawi , j.f . , hall , p.b . ,
turner , e.l . ,
inada , n. , oguri , m. , ichikawa , s .-
i . , becker , r.h . ,
gregg , m.d .
, white , r.l .
, wyithe , j.s.b . ,
schneider , d.p . ,
johnston , d.e .
, frieman , j.a . ,
brinkmann , j. 2004 , , 610 , 679 ricci , d. , poels , j. , elyiv , a. , finet , f. , sprimont , p. g. , anguita , t. , bozza , v. , browne , p. , burgdorf , m. , calchi novati , s. , dominik , m. , dreizler , s. , glitrup , m. , grundahl , f. , harpse , k. , hessman , f. , hinse , t. c. , hornstrup , a. , hundertmark , m. , jorgensen , u. g. , liebig , c. , maier , g. , mancini , l. , masi , g. , mathiasen , m. , rahvar , s. , scarpetta , g. , skottfelt , j. , snodgrass , c. , southworth , j. , teuber , j. , thone , c. c. , wambsganss , j. , zimmer , f. , zub , m. , surdej , j. 2011 , , 528 , 42 | we use single - epoch spectroscopy of three gravitationally lensed quasars , he0435 - 1223 , wfi2033 - 4723 , and he2149 - 2745 , to study their inner structure ( blr and continuum source ) .
we detect microlensing - induced magnification in the wings of the broad emission lines of two of the systems ( he0435 - 1223 and wfi2033 - 4723 ) . in the case of wfi2033 - 4723
, microlensing affects two `` bumps '' in the spectra which are almost symmetrically arranged on the blue ( coincident with an emission line ) and red wings of .
these match the typical double - peaked profile that follows from disk kinematics .
the presence of microlensing in the wings of the emission lines indicates the existence of two different regions in the blr : a relatively small one with kinematics possibly related to an accretion disk , and another one that is substantially more extended and insensitive to microlensing .
there is good agreement between the estimated size of the region affected by microlensing in the emission lines , @xmath0 light - days ( red wing of in he0435 - 1223 ) and @xmath1 light - days ( bumps in wfi2033 - 4723 ) with the sizes inferred from the continuum emission , @xmath2 light - days ( he0435 - 1223 ) and @xmath3 light - days ( wfi2033 - 4723 ) . for he2149 - 2745
we measure an accretion disk size @xmath4 light - days .
the estimates of @xmath5 , the exponent of the size vs. wavelength ( @xmath6 ) , are @xmath7 , @xmath8 , and @xmath9 for he0435 - 1223 , wfi2033 - 4723 , and he2149 - 2745 , respectively . in conclusion , the continuum microlensing amplitude in the three quasars and chromaticity in wfi2033 - 4723 and he2149 - 2745 are below expectations for the thin disk model .
the disks are larger and their temperature gradients are flatter than predicted by this model . |
topological excitations play a very important role in system dynamics , because they are responsible for mass and charge transport in solids and on crystal surfaces . as classical examples
, one can mention dislocations and crowdions described by the frenkel - kontorova ( fk ) model @xcite , where the topological excitations correspond to kinks that describe local compression or expansion of a commensurate structure .
the fk model has numerous applications in superionic conductivity , surface physics , hydrogen - bonded chains , josephson junctions , tribology , etc .
( e.g. , see @xcite and references therein ) . in the continuum limit approximation ,
the equation of motion of the fk model reduces to the exactly integrable sine - gordon ( sg ) equation .
but in continuum models , even in a model with anharmonic ( but local ) interaction , the topological excitations are always subsonic , the kink can not propagate with a velocity @xmath0 larger than the sound speed @xmath1 because of lorentz contraction of kink s width .
moreover , in the classical fk model , the kinks of the same topological charge repel from one another and , therefore , they can not carry a multiple topological charge .
however , simulation demonstrates that supersonic kinks as well as multiple kinks do exist .
for example , fig .
[ fig01 ] shows the propagation of supersonic single and double kinks in the fk model with exponential interatomic interaction .
the motion of topological solitons with _
velocities was firstly predicted , to the best of our knowledge , analytically by kosevich and kovalev @xcite in the fk model with some specific interatomic interaction in the continuum limit approximation .
later , supersonic topological solitons were observed by bishop _
@xcite in molecular dynamics study of polyacetylene .
then the supersonic kinks were studied numerically in the discrete fk model with anharmonic interaction by savin @xcite .
it was shown that for certain supersonic kink velocities , when its width coincides with that of the corresponding toda soliton @xcite , the kink propagates almost without energy losses .
_ multiple _ fast ( but subsonic ) kinks were firstly observed numerically by peyrard and kruskal @xcite in the classical highly discrete fk model .
@xcite have shown that multiple kinks exist also in continuum systems with _
nonlocal _ interaction .
the bounded states of kinks ( subsonic as well as supersonic ) in the case of anharmonic interaction were also studied numerically by zolotaryuk _
it was found that these multiple kinks are asymptotically unstable .
the dynamics of the generalized fk chain _ driven _ by a dc external force was studied numerically in @xcite , where we observed the existence of supersonic kinks and multiple ( double and triple at least ) kinks .
recently , the existence of multiple kinks for certain kink velocities in the _ discrete _ fk - type model was proven rigorously @xcite .
= the aim of the present paper is to find the criteria for the existence of supersonic and multiple topological excitations in the fk - type models .
we will show that , first , the model must be _ discrete _ as was already mentioned above .
second , because kink s motion in a discrete chain is damped due to radiation of phonons , we must apply _ a driving force _ to support the motion ( this point was lost in the previous studies @xcite ) .
third , the interatomic interaction must be _
anharmonic_. under these three conditions , the model admits both _ supersonic kinks _ and _ multiple kinks_. the paper is organized as follows .
the model is introduced in sec .
[ model ] . then in sec .
[ continuum ] the problem is studied with the help of the continuum limit approximation .
we show that if the discreteness effects are properly taken into account , the model formally allows both supersonic kinks and multiple kinks .
moreover , we prove that for a given set of parameters the model admits either the single - kink solution or the double - kink solution , but not both solutions simultaneously . numerical solution of the corresponding ordinary differential equation , which can be obtained with any desired accuracy , does show the existence of supersonic and multiple kinks . in secs .
[ supersonic ] and [ multiple ] we develop an approximate variational approach that helps to find the conditions for existence of supersonic and multiple kinks .
we show that supersonic kinks may exist in the model with anharmonic interaction only , and the multiple kinks may be stable for supersonic velocities only .
these conclusions are confirmed by simulation .
finally , sec
. [ conclusion ] concludes the paper with the comment that supersonic kinks and multiple kinks may be considered as `` disturbed '' toda solitons .
we consider the chain of atoms placed into the sinusoidal substrate potential and driven by a dc force @xmath2 applied to all atoms , so that the equation of motion is @xmath3 \nonumber \\ & & + \sin x_l(t)-f=0 , \label{mot}\end{aligned}\ ] ] where @xmath4 is the coordinate of the @xmath5th atom , @xmath6 is the external viscous damping coefficient introduced to compensate the driving force , and @xmath7 describes the exponential repulsion between nearest neighboring atoms .
we assume periodic boundary conditions with the number of atoms @xmath8 , where @xmath9 is the number of wells of the periodic potential , so we have one multiple @xmath10-kink ( @xmath10 excessive atoms ) inserted into the commensurate structure .
similarly to the toda model @xcite , one may add also an attractive linear interaction , so that the interaction reduces to the harmonic one in the limit @xmath11 ( the classical fk model ) and to the hard
core interaction in the opposite limit @xmath12 . throughout the paper , we use the dimensionless system of units , where the atomic mass is @xmath13 , the period of the external potential is @xmath14 and its amplitude is @xmath15 . in these units , a characteristic frequency of atomic vibration at a minimum of the external potential is @xmath16 , a characteristic time scale is @xmath17 , and the maximum value of the external dc force , when the minima of the sinusoidal substrate potential disappear and the topological excitations can not exist anymore , is @xmath18 .
the systematic procedure to derive the equation of motion in the continuum limit starting from the discrete lattice was proposed by rosenau @xcite . for the anharmonic fk model ( [ mot ] ) it leads , in the first order of the discreteness parameter , to the equation @xmath19 = 0 , \label{h18 } \end{aligned}\ ] ] where @xmath20 is the width of the static kink , @xmath21 is the elastic constant defined as @xmath22 the anharmonicity parameter @xmath23 is defined as @xmath24 and the parameter @xmath25 describes the discreteness effects . looking for a traveling wave solution of eq .
( [ h18 ] ) in the form @xmath26 , we obtain the equation @xmath27 where @xmath28 is the sound speed ( in our system of units @xmath29 ) .
although the travelling - wave ansatz may be too crude due to radiation of phonons by the moving kink ( e.g. , see fig .
[ fig01]a ) , eq .
( [ r2 ] ) allows to find kink s asymptotic rigorously , because the radiation has to decay due to nonzero damping coefficient @xmath6 in the model under consideration . at @xmath30 , substituting @xmath31 into eq .
( [ r2 ] ) , we obtain for the kink width @xmath32 the equation @xmath33 where @xmath34 . similarly , we can find the tail asymptotic behind the kink , @xmath35 at @xmath36 ; the width @xmath37 has to satisfy the same equation ( [ r3 ] ) but with @xmath38 .
one can see that at low velocities , @xmath39 , the discreteness effects lead to a decrease of the kink width in agreement with theory @xcite .
however , now eq .
( [ r3 ] ) has a solution for _ any _ kink velocity @xmath0 .
thus , the discreteness effects remove the restriction @xmath40 of sg - type equations . now even for the classical fk model with harmonic interaction the kink may move with any velocity @xmath0 .
a kink solution corresponds to a separatrix of the continuum limit equation . to find the separatrix of eq .
( [ r2 ] ) , let us normalize the coordinate @xmath41 , the velocity @xmath42 , and define the dimensionless discreteness parameter @xmath43 . introducing the new variable @xmath44 and the function @xmath45 , eq .
( [ r2 ] ) can be rewritten as @xmath46^{3 } \right ] \right .
\\ \nonumber % \left .
& & - \left .
\alpha w^{\prime}(\xi ) w(\xi ) + \left ( 1-\widetilde{v}^2 -\widetilde{h}^2 \cos \xi \right ) w^{\prime}(\xi ) \right .
\\ % \left . & & + \left .
\left ( \widetilde{h}^2 \sin \xi \right ) w(\xi ) + \eta \widetilde{v } \right\ } w(\xi ) -\sin \xi + f = 0 .
\label{r4}\end{aligned}\ ] ] a ( multiple ) kink solution has to satisfy the boundary conditions @xmath47 , @xmath48 , @xmath49 , or @xmath50 for example , if @xmath51 is a separatrix for the double kink ( @xmath52 ) , it has to connect the points @xmath53 and @xmath54 .
however , due to periodicity of the substrate potential the function @xmath55 must correspond to the separatrix solution as well .
thus , on the ( @xmath56 ) plane the separatrices of the multiple ( @xmath57 ) kinks must intersect at some point with @xmath58 .
one can show that in the model without the discreteness effects , @xmath59 , when the phase space of eq .
( [ r4 ] ) is two - dimensional , such intersections are forbidden @xcite .
thus , eq . ( [ r4 ] ) with @xmath59 allows neither supersonic kinks nor multiple kinks . on the contrary , at @xmath60 the phase space of eq .
( [ r4 ] ) is four - dimensional , thus the separatrices corresponded to multiple kinks may not intersect , and multiple kinks are allowed in principle @xcite .
thus , _ the discrete model formally allows both supersonic and multiple topological excitations . _
although we can not find the separatrix solution analytically , the ordinary differential equation ( [ r4 ] ) can be solved numerically with any desired accuracy .
thus , if one could find a separatrix solution corresponded to supersonic or multiple kink , this will prove their existence .
indeed , looking for a separatrix solution numerically for the @xmath61 and @xmath62 case , we found that at small discreteness , @xmath63 , so that @xmath64 and @xmath65 , the separatrix solution corresponds to the @xmath66-kink at forces as large as @xmath67 , when the kink is supersonic , @xmath68 . on the other hand , for higher discreteness , @xmath69 so that @xmath70 , we saw the @xmath71-kink at @xmath72 when @xmath73 , and the @xmath74-kink at @xmath75 when @xmath76 .
_ thus , both supersonic kinks and multiple kinks do exist _ , at least for some particular choices of model parameters .
near kink s tails , @xmath77 , e.g. for @xmath78 , where @xmath79 , we can use the expansion @xmath80 where @xmath81 , @xmath82 , _ etc . _ substituting this expansion into eq .
( [ r4 ] ) and grouping together the terms of the same power of @xmath83 , we obtain the following equation for @xmath84 [ cf . with eq .
( [ r3 ] ) ] , @xmath85 which always has two roots , one positive and one negative , for any kink velocity @xmath0 . then , equating the terms for higher powers of @xmath83 , we obtain the relations that uniquely determine the coefficients @xmath86 , @xmath87 , _ etc . _
thus , the separatrix solution of eq .
( [ r4 ] ) is _ unique _ , i.e. , for a given set of system parameters _ the model has either the single - kink solution or the double - kink one _ , but never both solutions simultaneously .
thus , we have demonstrated the existence of supersonic and multiple kinks in the driven discrete fk model with anharmonic interaction ( note that the discreteness of the model and the anharmonicity of the interaction are the necessary conditions ) .
but to find the parameter range for their existence ( the sufficient conditions ) , we have to study numerically either the continuum - limit equation ( [ r4 ] ) or , better , the original discrete model ( [ mot ] ) .
the approximate variational approach described below , essentially simplifies this task .
it is easy to show that in the case of @xmath88 , eq . ( [ r2 ] ) corresponds to an extremum of the following energy functional @xcite , @xmath89 = \int % _ { -\infty}^{+\infty } dz \left [ \frac{c^2-v^2}{2 } \left ( u^{\prime } \right)^2 - \frac{\alpha d^3}{6 } \left ( u^{\prime } \right)^3 -\cos u \right].\ ] ] substituting a simple sg - type ansatz @xmath90 into eq .
( [ fz2 ] ) , we obtain @xmath91 = although the variational approach does not describe rigorously the kink tails because of neglecting the discreteness effects , it allows us to find analytically the shape of the kink s core and , therefore , to calculate approximately the kink velocity for the model with anharmonic interaction .
indeed , looking for extrema of the function @xmath92 , we come to the equation @xmath93 , or @xmath94 \kappa + \frac{\pi}{3 } \alpha,\ ] ] where we introduced the new variable @xmath95 . for the harmonic interaction , @xmath96 , eq .
( [ fz6 ] ) has a solution for @xmath40 only , which describes relativistic narrowing of the sg kink , @xmath97^{1/2}$ ] .
but for the anharmonic interaction , @xmath98 , eq .
( [ fz6 ] ) has a solution for _ any _ kink velocity @xmath0 , including supersonic velocities @xmath99 .
we emphasize that _ supersonic excitations are possible for kinks ( local compressions ) only . _
considering the kink as a rigid quasiparticle , the kink effective mass can be introduced as ( e.g. , see @xcite ) @xmath100 ^ 2 = \frac{4}{\pi d_{\rm eff}}. \label{fz7}\ ] ] then , assuming that kink s parameters at nonzero @xmath2 and @xmath6 are the same as those for the @xmath101 case , the steady - state kink velocity can be found approximately from the equation @xmath102 using eq .
( [ fz8a ] ) , eq .
( [ fz6 ] ) can be rewritten in the form @xmath103 \kappa^3 = \kappa + \frac{\pi}{3 } \alpha.\ ] ] numerical solution of eq .
( [ fz9 ] ) allows us to find the function @xmath104 that is shown by the dashed curve in fig .
[ fig02 ] together with the dependence @xmath105 obtained by solution of the discrete equation of motion ( [ mot ] ) .
one can see that in the anharmonic model we always have @xmath106 at @xmath107 , and although the simulation velocity is lower than @xmath108 due to additional damping of the moving kink because of phonon radiation , the discrete kink still may reach a supersonic velocity .
thus , the variational approach predicts that the supersonic kinks may be expected in the _ anharmonic _ fk model only .
returning back to the _ discrete _ model , note that in the classical fk model the driven kink can not reach even the sound velocity , because it exists some critical kink velocity @xmath109 above which the driven kink becomes unstable and the system goes to the `` running '' state , where all atoms move with the velocity @xmath110 @xcite .
however , in the anharmonic fk model , the critical kink velocity may exceed the sound speed as has been observed already in the simulation @xcite .
the dependence of @xmath111 on the anharmonicity parameter @xmath112 is shown as inset in fig .
[ fig02 ] . in this calculation
we used the following algorithm @xcite : for a fixed value of @xmath2 ( we took @xmath113 ) , the friction was decreased starting from the overdamped case @xmath114 to the underdamped value @xmath115 in 256 steps . at each step of @xmath6 decreasing we waited untill the steady state was reached and then checked if the transition to the running state took place .
to study multiple kinks with the help of a variational approach , let us consider the double kink as a sum of two single kinks separated by a distance @xmath116 , @xmath117 substituting the ansatz ( [ dk1 ] ) into the functional ( [ fz2 ] ) , we obtain the effective energy @xmath118 which is now a function of two parameters @xmath119 and @xmath116 . looking for a minimum of @xmath118 over @xmath119 at @xmath116 fixed
, we found that for the classical fk model , @xmath96 , the function @xmath120 is a monotonically decreasing function of @xmath116 , i.e. , the kinks are repelled from one another . on the other hand , for the anharmonic fk model the function
@xmath121 has a minimum at some @xmath122 , so two kinks attract one another and thus have to couple into a double kink .
the `` dissociation '' energy of the double kink is very small at subsonic velocities , but becomes high enough at supersonic velocities ( see inset in fig .
[ fig03 ] ) . the `` size '' @xmath123 of the double kink decreases with @xmath124 increasing .
thus , although the variational approach with the sg - type ansatz is too crude , it nevertheless predicts that the multiple kinks could be stable for high ( supersonic ) kink velocities .
= as is well known , the sg kinks of the same topological charge always repel from one another .
the same is true for static ( @xmath125 ) kinks of the discrete fk model , including the anharmonic ( @xmath126 ) model @xcite ( contrary to the variational approximation which mistakenly predicted a weak attraction at @xmath40 ) .
thus , it must exist a threshold kink velocity @xmath127 such that at small velocities @xmath128 the steady - state solution of the model corresponds to the @xmath71-kink , while at high velocities @xmath129 , it corresponds to the double ( @xmath74- ) kink ( if @xmath130 is lower than the kink velocity at @xmath18 , that is true at low enough values of @xmath6 ) .
indeed , the simulation results presented in fig . [ fig03 ] demonstrate that the double kink is stable within the force interval @xmath131 but becomes unstable at higher as well as smaller forces , while the @xmath71-kink is stable for @xmath132 only .
similarly one could expect the existence of a second threshold velocity @xmath133 such that at @xmath134 the steady - state solution will correspond to the @xmath135-kink , _ etc . _
thus , we have shown that supersonic kinks as well as multiple kinks do exist in the driven discrete fk model , if the interatomic interaction is anharmonic . note that both supersonic kinks and multiple kinks remain stable at nonzero system temperatures as well , at least for the time scale of our numerical simulation .
notice also that at high forces the kink velocity is close to that of the toda soliton @xcite .
indeed , the toda soliton is characterized by the `` jump '' @xmath136 , where @xmath137 is the parameter coupled with the soliton velocity @xmath0 by the relationship @xmath138 . in the presence of the external substrate potential due to boundary conditions the jump @xmath139 must be equal to @xmath140 for the @xmath10-kink ; thus we obtain @xmath141 , or @xmath142 . in particular , for the anharmonicity parameter @xmath143 used in the simulation , we have @xmath144 for the @xmath71-kink and @xmath145 for the @xmath74-kink correspondingly .
thus , the supersonic and multiple kinks may be treated as toda solitons `` disturbed '' by the external periodic potential .
discussions with a. r. bishop , yu .
s. kivshar , and m. peyrard are gratefully acknowledged .
this work was supported in part by nato grant no .
htech.lg 971372 , intas grant no .
97 - 31061 , and by a grant from the hong kong research grants council . | the criteria for the existence of supersonic and multiple topological excitations ( kinks ) in the driven frenkel - kontorova model ( a chain of atoms placed into an external periodic potential ) with anharmonic ( exponential ) interatomic interactions are studied . |
our understanding of planet formation has rapidly improved over the past 15 years .
prior to the discovery of the first extrasolar planet orbiting a solar - type star ( 51 peg b ; @xcite ) , it was widely assumed that extrasolar giant planet semimajor axes would mimic those of the gas giants in our own solar system , which orbit at distances @xmath14 5 au . in the years that followed it became apparent that an _ in situ _ formation model was not universally applicable because radial velocity surveys were finding jovian planets in abundance well inside the canonical ice line .
over 350 planets have now been discovered , 282 of which reside around stars within 200 pc .
sufficiently large samples are available for the statistical properties of exoplanets to reveal themselves , providing information about the planet formation and migration processes . for solar - type stars ( f , g , and k dwarfs ) ,
jovian planets fall into two rough populations : `` hot planets '' with @xmath15 @xmath1 0.1 au ( @xcite ) and those that orbit beyond @xmath0 1 au ( figure [ f1.eps ] ) .
these observations were explained _ a posteriori _ in terms of orbital migration ( @xcite ) and planet - planet scattering ( @xcite ; @xcite ; @xcite ) , with the dearth of planets with periods between @xmath0 10 - 100 days ( the `` period valley '' ) possibly resulting from differential mass - dependent orbital migration ( @xcite ; @xcite ; @xcite ) .
while much is known about planets around sun - like stars ( e.g. , @xcite ; @xcite ; @xcite ; @xcite ; @xcite ) , comparatively little is known about planets around intermediate - mass ( i m ) stars with @xmath16 @xmath14 1.5 @xmath17 .
main sequence a- and f - type stars are problematic doppler targets because of their high jitter levels and rotationally - broadened absorption features ( @xcite ) . on the other hand , evolved i m stars have lower jitter levels as well as narrower and more numerous absorption lines resulting from their slow rotation and cool photospheres . as a consequence , nearly all radial velocity surveys of i
m stars are targeting evolved g- or k - type subgiants and giants ( @xcite , @xcite , @xcite ; @xcite , @xcite , @xcite , @xcite , @xcite ) , although at least one survey is targeting their main sequence progenitors ( @xcite ; @xcite ) .
recently @xcite showed that planets orbiting evolved a - type stars have large semimajor axes compared to planets around solar - type stars .
this trend has become even stronger with the discovery of more planets orbiting i m stars ( @xcite ) .
specifically , over 20 planets have been discovered around stars with minimum masses @xmath14 1.5 @xmath18 but none have semimajor axes @xmath19 0.6 au .
this `` planet desert '' is shown in figure [ f1.eps ] and a summary of known planets around i m stars is presented table [ evolvedtab ] .
lcccccccc hd 13189 & 2 - 6 & k2 ii & & 8 - 20 & 1.5 - 2.2 & 0.27 ( 0.06 ) & 1 , 2 + @xmath20 tau & 2.7 ( 0.1 ) & k0 iii & 13.7 ( 0.6 ) & 7.6 ( 0.2 ) & 1.93 ( 0.03 ) & 0.151 ( 0.023 ) & 3 + ngc 2423 no . 3 & 2.4 ( 0.2 ) & & & 10.6 & 2.10 & 0.21 ( 0.07 ) & 4 + 81 cet & 2.4 ( 2.0 - 2.5 ) & g5 iii : & 11 ( 10 - 13 ) & 5.3 & 2.5 & 0.206 ( 0.029 ) & 5 + hd 104985 & 2.3 & g9 iii & 11 & 8.3 & 0.95 & 0.090 ( 0.009 ) & 6 , 7 + 18 del & 2.3 & g6 iii & 8.5 & 10.3 & 2.6 & 0.08 ( 0.01 ) & 6 + hd 17092 & 2.3 ( 0.3 ) & k0 iii & 10.9 ( 2.8 ) & 4.6 ( 0.3 ) & 1.29 ( 0.05 ) & 0.166 ( 0.052 ) & 8 + @xmath21 aql & 2.2 & k0 iii & 12 & 2.8 & 0.68 & 0.0 ( fixed ) & 6 + 14 and & 2.2 ( 2.0 - 2.3 ) & k0 iii & 11 ( 10 - 12 ) & 4.8 & 0.83 & 0.0 ( fixed ) & 5 + hd 81688 & 2.1 & k0iii - iv & 13 & 2.7 & 0.81 & 0.0 ( fixed ) & 6 + hd 173416 & 2.0 ( 0.3 ) & g8 iii & 13.5 ( 0.9 ) & 2.7 ( 0.3 ) & 1.16 ( 0.06 ) & 0.21 ( 0.04 ) & 9 + hd 11977 & 1.91 ( 0.21 ) & g5 iii & 10.09 ( 0.32 ) & 6.54 & 1.93 & 0.40 ( 0.07 ) & 10 , 11 + hd 102272 & 1.9 ( 0.3 ) & k0 iii & 10.1 ( 4.6 ) & 5.9 ( 0.2 ) & 0.614 ( 0.001 ) & 0.05 ( 0.04 ) & 12 + & & & & 2.6 ( 0.4 ) & 1.57 ( 0.05 ) & 0.68 ( 0.06 ) & 12 + @xmath12 gem & 1.86 , 1.7 ( 0.4 ) & k0 iii & 8.8 ( 0.1 ) & 2.9 ( 0.3 ) & 1.69 ( 0.03 ) & 0.06 ( 0.04 ) & 13 , 14 , 15 , 16 + hd 89744 & 1.86 ( 0.18 ) & f7 iv & 2.08 ( 0.06 ) & 7.2 & 0.88 & 0.70 ( 0.02 ) & 17 , 18 + hd 210702 & 1.85 ( 0.13 ) & k1 iv & 4.45 ( 0.07 ) & 1.97 ( 0.11,0.18 ) & 1.20 ( 0.02,0.03 ) & 0.036 ( @xmath190.106 ) & 19 , 20 + @xmath22 crb & 1.84 ( 0.13 ) & k0 iv & 4.71 ( 0.08 ) & 2.01 ( 0.11,0.17 ) & 2.80 ( 0.07,0.08 ) & 0.044 ( @xmath190.123 ) & 21 , 20 + 6 lyn & 1.82 ( 0.13 ) & k0 iv & 5.2 ( 4.9 - 5.6 ) & 2.21 ( 0.11,0.16 ) & 2.18 ( 0.05,0.06 ) & 0.059 ( @xmath190.125 ) & 20 , 5 + hd 167042 & 1.72 ( 0.12 ) & k1 iv & 4.30 ( 0.07 ) & 1.70 ( 0.09,0.12 ) & 1.32 ( 0.03,0.04 ) & 0.089 ( 0.028,0.065 ) & 21 , 20 , 5 + hd 192699 & 1.69 ( 0.12 ) & g8 iv & 3.90 ( 0.06 ) & 2.40 ( 0.15,0.21 ) & 1.15 ( 0.02,0.03 ) & 0.129 ( 0.029,0.060 ) & 19 , 20 + hd 175541 & 1.65 ( 0.12 ) & g8 iv & 3.80 ( 0.09 ) & 0.70 ( 0.06,0.08 ) & 1.03 ( 0.02,0.03 ) & 0.083 ( @xmath190.283 ) & 19 + hd 5319 & 1.59 ( 0.18 ) & g5 iv & 3.26 ( 0.50,0.41 ) & 1.94 & 1.75 & 0.12 ( 0.08 ) & 22
there are several explanations for this observational result .
one possibility is that the swollen radii of evolved stars have engulfed or tidally disrupted short - period planets .
the current census of i m exoplanet - host stars mostly consists of red clump giants ( core helium and hydrogen shell burning post - rgb stars ) and subgiants ( stars with contracting inert helium cores and hydrogen - burning shells ) .
planet - hosting stars in the clump giant phase have radii between @xmath0 8 - 14 @xmath23 ( @xmath0 0.03 - 0.06 au ) so any planets orbiting inside @xmath0 0.1 au will likely have been engulfed by an expanding radius during post - main sequence stellar evolution .
however , a tidal torque from an expanding stellar surface can also decay the orbits of short period planets . @xcite
numerically trace the semimajor axis evolution of short - period planets around evolving rgb stars and show that , in their past , i m clump giants may have engulfed or disrupted the orbits of planets out to @xmath0 0.5 au .
with radii between @xmath0 2 - 5 @xmath23 ( @xmath0 0.01 - 0.02 au ) , planets orbiting i m subgiants are the least affected by stellar evolution .
stellar evolution may therefore explain the lack of short - period planets around i m clump giants , but the same result for i m subgiants suggests that the observed trend is not due to post - main sequence engulfment .
the observed semimajor axis distribution can also be explained as a result of inward orbital migration combined with mass - dependent disk dispersal lifetime .
the rocky progenitors of jovian planets ( @xmath010 @xmath24 cores ) can form at distances @xmath25 8 au for stellar masses between @xmath0 1.5 - 3 @xmath18 ( @xcite ) .
@xcite performed monte carlo simulations of jovian planet formation and migration around i m stars and , using simple stellar mass - dependent gas disk lifetime relations ( @xcite ) and type ii migration models ( @xcite ) , was able to successfully reproduce the observed dearth of short - period planets .
in this scenario inward migration is halted once rapid disk dispersal occurs , stranding migrating planets at semimajor axes that depend on stellar mass .
an alternative explanation was offered by @xcite .
they found that the protoplanetary disks of young i m stars will develop a maximum surface density at @xmath0 1 au as a result of magnetrotational instability of the inner disk , leading to a trapping and accumulation of solids that can then grow to form rocky cores and jovian planets .
this formation scenario provides a mechanism for _ in situ _ formation of jovian planets interior to the ice line , which is located near 3 au at 10 myr for a 2 @xmath18 star ( @xcite ) .
models of jovian planet formation around i m stars make few quantitative predictions that can be observationally tested . simple disk depletion plus migration models for i m planet hosts ( 1.5 - 3.0 @xmath18 ) by @xcite predict occurrence rates for jovian planets with semimajor axes @xmath19 0.5 au to be @xmath1 1.5% and for jovian planets with semimajor axes @xmath14 0.5 au to be @xmath25 7.5% .
@xcite use a semianalytic model of protoplanetary disk evolution to study snow line locations and planet formation rates around stars of varying masses . for stellar masses between 1.5 - 2.0
@xmath18 , their models predict that jovian planet occurrence rates reach frequencies of @xmath010 - 15% .
@xcite suggest that multiple planetary systems may form more efficeintly around i m stars compared to other stellar mass regimes .
the semimajor axis distribution of planets with minimum masses between @xmath0 2 - 10 @xmath4 orbiting i m stars is beginning to be better constrained by observations , but little is known about planets with masses @xmath19 1.5 @xmath4 .
this poor understanding is a direct result of the dearth of low - mass planets currently known , with only one having a minimum mass below 1.5 @xmath4 ( hd 175541b with @xmath26sin@xmath27 = 0.70 @xmath4 ) .
it is unclear , however , whether this scarcity is a result of a detection bias caused by higher jitter levels in i m stars or whether it reflects an intrinsic shortage of low - mass planets .
unfortunately , models of planet formation in this stellar mass regime have made few predictions about low - mass planetary companions . for solar - type stars ,
@xcite estimate the frequency of neptune - mass planets with periods @xmath19 50 days to be at least 30% .
this raises the exciting possibility that low - mass planets could be abundant around i m stars , especially in light of recent studies suggesting that the frequency of jovian - mass planets scales with stellar mass ( @xcite ) .
testing these theories requires an understanding of the detection limits of radial velocity surveys .
the goal of this study is twofold : to derive the occurrence rate of jovian planets around i m stars and to characterize the distributions of planet periods ( @xmath10 ) and minimum masses ( @xmath26sin@xmath27 ) .
in addition we take the opportunity to update the orbit solutions for known planet - hosting stars in our sample using new radial velocity measurements . to address the aforementioned questions we make use of a uniform sample of 31 i m subgiants taken from an ongoing radial velocity survey at lick observatory ( @xcite ; @xcite ) .
our observations span @xmath0 5 years and sample semimajor axes out to several au , enabling detailed comparisons with previous surveys targeting solar - type stars . in @xmath282
we describe our radial velocity measurements and define our sample selection .
updated keplerian orbits for five previously - known planetary systems are presented in @xmath283 and in @xmath284 we describe time - series photometric observations of these systems . in @xmath285
we derive detection limits for stars in our sample .
we discuss the frequency of jovian - mass planets in @xmath286 , and in @xmath287 we compare the mass - period distribution for solar - type stars to the results from our sample of i m subgiants .
finally , we discuss the implications of our work in @xmath288 .
our sample of subgiants is derived from a larger planet - hunting program at lick observatory targeting evolved stars ( the lick subgiant planet search : @xcite ; @xcite ) .
the original survey consists of 159 stars selected on the basis of @xmath29 @xmath30 colors and @xmath31 absolute magnitudes such that they lie one magnitude above the main sequence , have masses @xmath25 1.2 @xmath18 , and avoid red giant branch ( rgb ) , clump giant , and cepheid variable regions of the hr diagram .
known spectroscopic and proper motion binaries within @xmath0 2@xmath32 are excluded . to obtain a more uniform population for this study we selected a sample of stars from the original survey based on the following criteria : 2 @xmath19 @xmath31 @xmath19 3 , 0.8 @xmath19 @xmath33
@xmath34 @xmath19 1.0 , @xmath16 @xmath14 1.5 @xmath18 , and at least four radial velocity measurements ( the average number of measurements in our final sample is 26 ) .
4 radial velocity measurements , but it is a previously - known spectroscopic binary . ]
the magnitude cuts exclude clump giants and solar - mass subgiants , while the color cuts exclude rgb stars and yellow stragglers , the latter of which have high jitter levels and large binary frequencies .
altogether our sample includes 31 intermediate - mass subgiants ( figure [ f2.eps ] ) .
radial velocity measurements were obtained at lick observatory s 3 m shane telescope and 0.6 m coude auxiliary telescope with the hamilton spectrometer ( @xcite ; @xmath35 @xmath0 50,000 at @xmath36 = 5500 ) .
doppler shifts are measured from each spectrum using the iodine cell method described in detail by @xcite and summarized as follows . a temperature - controlled pyrex cell containing gaseous iodine
is placed at the entrance slit of the spectrometer .
the dense set of narrow molecular lines imprinted on each stellar spectrum from 5000 to 6000 provides a robust wavelength scale for each observation , as well as information about the shape of the spectrometer s instrumental response ( @xcite ) .
additional details about the observations can be found in @xcite .
we derive keplerian orbital solutions for the five previously - known planets in our sample using ` rvlin ` , an efficient orbit - fitting routine written in idl and based on a partial linearization of kepler s equations ( @xcite ) .
the radial velocity measurements were first combined into two - hour bins and the total measurement errors were estimated by adding 5 m s@xmath37 to the internal measurement errors , in quadrature , to account for the moderate jitter level of typical subgiants ( @xcite ; @xcite ; @xcite ) .
we combine our observations of hd 167042 and 6 lyn ( hd 45410 ) with those made available by @xcite to create a larger data set for improved orbital phase coverage .
the data sets were merged by simultaneously fitting for the constant offset between the two observatories .
long - period companions will manifest as constant accelerations in the radial velocity measurements . to first order
this drift can be modeled as a linear velocity trend for very long - period companions for which the period is much greater than the observing time baseline .
when the slope has an amplitude comparable to the noise it is often difficult to determine whether a trend should be included in an orbit solution .
the goodness of fit statistic ( typically the @xmath38 value ) might be reduced with the orbit - plus - trend solution , but , by that criterion for model selection , a more complicated model with multiple planets could be fit to a radial velocity set for which there is clearly insufficient evidence for additional companions . more generally ,
how does one decide whether including more parameters in a model is justified ?
fortunately this problem has received a great deal of attention in statistics and is the subject of bayesian model comparison .- test ( see , e.g. , @xcite ) .
we choose to follow a bayesian approach in this work . ]
a commonly used model selection tool is the bayesian information criterion ( bic ; @xcite ; @xcite ; @xcite ) , which rewards better - fitting models but penalizes models that are overly complex : @xmath39 where @xmath40 is the maximum likelihood for a particular model with @xmath41 free parameters and @xmath42 data points . in general the model with the smaller bic value
is preferred .
we use bic values for orbit solutions with and without a velocity trend to decide whether to include a linear slope in the model . the only planet - hosting star for which a trend produced a lower bic value was hd 167042 .
we take this as evidence for a long - period companion in that system .
the best - fitting orbits for the five known planet - hosting subgiants are shown in figures [ f3.eps]-[f7.eps ] , and the updated orbital parameters are listed in table [ orbitpars1 ] .
lccccc @xmath10 ( days ) & 420.77 ( 3.48 , 3.11 ) & 345.53 ( 1.77 , 1.63 ) & 354.29 ( 2.31 , 2.13 ) & 1261.94 ( 28.91 , 23.97 ) & 874.774 ( 16.27 , 8.47 ) + @xmath43 ( jd 2,450,000 ) & 4230.1 ( 40.6 , 41.6 ) & 4036.6 ( 21.0 , 22.1 ) & 4142.6 ( 78.3 , 100.1 ) & 3909.2 ( 332.5 , 260.9 ) & 4024.5 ( 180.2 , 130.9 ) + @xmath44 & 0.089 ( 0.028 , 0.065 ) & 0.129 ( 0.029 , 0.060 ) & 0.036 ( @xmath190.106 ) & 0.044 ( @xmath190.123 ) & 0.059 ( @xmath190.125 ) + @xmath45 ( @xmath46 ) & 85.7 ( 35.4 , 35.8 ) & 41.3 ( 23.3 , 25.9 ) & 282.3 ( 82.3 , 98.0 ) & 148.4 ( 83.2 , 94.2 ) & 314.9 ( 75.5 , 52.6 ) + @xmath47 ( m s@xmath37 ) & 32.16 ( 1.32 , 1.32 ) & 49.3 ( 2.6 , 3.2 ) & 37.45 ( 1.90 , 2.49 ) & 25.17 ( 1.12 , 1.55 ) & 31.53 ( 1.12 , 1.32 ) + @xmath48 sin@xmath27 ( @xmath4 ) & 1.70 ( 0.09 , 0.12 ) & 2.40 ( 0.15 , 0.21 ) & 1.97 ( 0.11 , 0.18 ) & 2.01 ( 0.11 , 0.17 ) & 2.21 ( 0.11 , 0.16 ) + @xmath15 ( au ) & 1.32 ( 0.03 , 0.04 ) & 1.15 ( 0.02 , 0.03 ) & 1.20 ( 0.02 , 0.03 ) & 2.80 ( 0.07 , 0.08 ) & 2.18 ( 0.05 , 0.06 ) + @xmath49 & 1.17 & 1.82 & 1.49 & 1.01 & 1.43 + bic ( trend ) & * 99.84 & 104.97 & 90.19 & 82.51 & 110.25 + bic ( no trend ) & 101.88 & * 101.72 & * 89.18 & * 80.15 & * 106.31 + @xmath50/@xmath51 ( m s@xmath37 yr@xmath37 ) & 2.14 ( 0.99 , 0.73 ) & 0.0 ( fixed ) & 0.0 ( fixed ) & 0.0 ( fixed ) & 0.0 ( fixed ) + * * * * * minimum masses and semimajor axes are calculated using the analytic approximation for @xmath26@xmath52 as a function of the observed orbital parameters and kepler s third law ( i.e. , equations 2 and 3 in @xcite ) .
we use newly determined stellar masses following the method described in @xcite based on new [ fe / h ] values ( see table [ evolvedtab ] ) , which are derived from updated iodine - free template spectra following the iterative scheme from figure 1 of @xcite .
we use a markov - chain monte carlo ( mcmc ) algorithm following the description in @xcite to estimate our orbital parameter uncertainties ( see also @xcite , @xcite ) .
mcmc is a bayesian inference technique that uses the data to explore the shape of the likelihood function for each parameter of an input model .
this method of error analysis has been shown to be more efficient and accurate than bootstrap monte carlo resampling techniques ( @xcite ) .
every step in the chain represents a variation of the period ( @xmath10 ) , eccentricity , ( @xmath44 ) , time of periastron passage ( @xmath53 ) , argument of periastron ( @xmath45 ) , velocity semi - amplitude ( @xmath47 ) , and constant velocity offset .
a velocity trend parameter ( d@xmath54/dt ) is allowed to vary for hd 167042 .
additionally , for hd 167042 and 6 lyn , a constant radial velocity offset between lick observatory measurements and those from @xcite is treated as a free parameter to account for the systematic offset between observations from different telescopes .
we allow one random parameter to be altered at each step by drawing a new value from a gaussian transition distribution .
if the resulting @xmath38 value for the trial orbit is less than the previous @xmath38 value , then the trial orbital parameters are retained . if not , then the probability of adopting the new value is equal to the ratio of the probabilities from the previous and trial steps ( the metropolis - hastings algorithm ; @xcite ; @xcite ) .
if the trial is rejected then the parameters from the previous step are adopted .
we altered the standard deviations of the gaussian transition functions so that the acceptance rates were between 20% and 40% to maximize convergence efficiency .
for each planet - star system , the initial parameters were chosen from the best - fitting orbit solutions and each chain was run for 1 - 3@xmath5510@xmath56 steps depending on the rate of convergence .
the initial 10% of each chain was excluded from the final estimation of parameter uncertainties .
the parameters were tested for convergence by running five shorter chains with 10@xmath57 steps .
we verified that convergence was reached by ensuring that the gelman - rubin statistic ( @xcite ; @xcite ) for each parameter was near unity ( typically @xmath1 1.02 ) and that a visual inspection of the history plots suggested stability .
the results of the mcmc analysis are posterior probability density functions ( pdfs ) for each orbital parameter used in the model .
an example of the results for hd 167042 is presented in figure [ f8.eps ] . for each parameter
we compute the range encompassing 68.3% of the data about the median of the distribution .
these values represent approximate upper and lower one - sigma limits for gaussian - like distributions and are included in parentheses next to each parameter in table [ orbitpars1 ] . for eccentricities near zero , the 68.3% range of data about the median is a poor estimator of the most likely values as it fails to encompass the peak of the distribution .
when that occurs we quote upper limit eccentricity values below which encompass 95.4% of the data .
we note that the results from using the bic value as a way to discriminate between adopting models with or without velocity trends yields the same results as using the posterior pdf of the velocity trend from mcmc as an indicator .
we use lomb - scargle ( ls ) periodograms ( @xcite ; @xcite ) to search for additional short period companions after subtracting the best - fitting orbit from the observations . a false alarm probability ( fap )
is computed following the method described in @xcite , which gives the probability that a peak will occur by chance assuming the data are pure noise .
the results are shown in figure [ f9.eps ] .
no periodicities in the residuals are identified with fap of 0.1% or less .
several stars in our sample exhibit very high radial velocity scatter indicative of stellar companions orbiting at small semimajor axes .
hd 8375 is one such system and , with 28 observations , is well - enough sampled to derive an accurate orbit solution and posterior pdfs using ` rvlin ` and mcmc ( figure [ f10.eps ] ) .
hd 8375 is a known spectroscopic binary ( @xcite ; @xcite ; @xcite ) but to our knowledge no orbit has been published .
the amplitude and period of the companion suggest that we can not use the negligible companion mass approximation to derive @xmath58sin@xmath27 .
instead we compute the mass function @xmath59(@xmath60 ) and minimum separation @xmath61sin@xmath27 , which are listed along with the orbital parameters of hd 8375b in table [ orbitpars8375 ] .
we find that a companion with a long - term velocity trend best - fits the data based on bic values with and without a trend .
lc @xmath10 ( days ) & 83.9408 ( 0.0016 , 0.0015 ) + @xmath43 ( jd 2,450,000 ) & 4039.3 ( 0.5,0.4 ) + @xmath44 & 0.0179 ( 0.0004 , 0.0004 ) + @xmath45 ( @xmath46 ) & 321.83 ( 1.74,1.97 ) + @xmath47 ( m s@xmath37 ) & 4939.2 ( 2.6,2.5 ) + @xmath59(@xmath60 )
( 10@xmath62 @xmath17 ) & 1.050 ( 0.002 , 0.002 ) + @xmath61sin@xmath27 ( 10@xmath63 au ) & 3.813 ( 0.002 , 0.002 ) + @xmath49 & 5.69 + bic ( trend ) & * 119.3 + bic ( no trend ) & 928.8 + @xmath50/@xmath51 ( m s@xmath37 yr@xmath37 ) & 52.9 ( 1.8,1.9 ) + * following @xcite , we can estimate the minimum mass required to accelerate the hd 8375 system at the observed rate as a function of semimajor axis : @xmath64 where @xmath26 is the minimum planet mass , @xmath27 is the inclination , @xmath65 is the acceleration , @xmath66 is the gravitational constant , and @xmath67 is the ratio of the time baseline of the observations divided by the period of the companion . for large periods relative to the time baseline , sec(2@xmath68 )
@xmath0 1 , yielding equation 1 from @xcite . for small values of @xmath67 ,
the measured value of @xmath65 gives @xmath69 values of @xmath26sin@xmath27 and @xmath15 below this curve are not permitted ; masses are not high enough at large separations to induce the observed acceleration . lowering the value of @xmath67 shifts the curve towards smaller semimajor axes and higher masses .
if the third body is a planet then it is in a circumbinary orbit , otherwise it is a star or brown dwarf in a hierarchical configuration .
in addition to the radial velocities from lick observatory , we used the t3 0.4 m and the t12 0.8 m automated photometric telescopes ( apts ) at fairborn observatory to gather time - series photometry of the five systems with orbital updates in table [ orbitpars1 ] .
the observations of these five systems were conducted at various epochs between 1993 april and 2009 june .
the apts can detect short - term , low - amplitude brightness variations in cool stars due to rotational modulation of the visibility of surface features , such as starspots and plages ( e.g. , @xcite ) , and can also detect longer - term variations associated with stellar magnetic cycles ( e.g. , @xcite ; @xcite ) .
photometry of planetary candidate host stars helps to establish whether observed low - amplitude radial velocity variations are caused by stellar activity or planetary - reflex motion ( e.g. , @xcite ) . @xcite and @xcite have found periodic radial velocity variations in solar - type stars that are caused by starspots .
the apt observations are also useful to search for possible transits of the planetary companions ( e.g. , @xcite ; @xcite ; @xcite ) .
the t12 apt is equipped with a two - channel precision photometer employing two emi 9124qb bi - alkali photomultiplier tubes ( pmts ) to make simultaneous measurements of a star in strmgren @xmath70 and @xmath71 passbands .
the t4 apt has a single - channel precision photometer that uses an emi 9124qb pmt to measure a star sequentially through @xmath70 and @xmath71 filters .
the apts observe each target star ( star d ) in a quartet with three ostensibly constant comparison stars ( stars a , b , and c ) .
we compute @xmath70 and @xmath71 differential magnitudes for each of the six combinations of the four stars : @xmath72 , @xmath73 , @xmath74 , @xmath75 , @xmath76 , and @xmath77 .
we then correct the strmgren @xmath70 and @xmath71 differential magnitudes for differential extinction with nightly extinction coefficients and transform them to the strmgren system with yearly mean transformation coefficients . finally , we combine the strmgren @xmath70 and @xmath71 differential magnitudes into a single @xmath78 passband to improve the precision of the observations .
@xcite presents a detailed description of the automated telescopes and photometers , observing techniques , and data reduction procedures needed for long - term , high - precision photometry .
we use the two most constant comparison stars in each quartet to compute the quantity @xmath79 , the target star s absolute night - to - night variability , statistically corrected for any unrecognized intrinsic variability in the two comparison stars and also corrected for the measurement uncertainty of the differential magnitudes .
we follow the method described in @xcite and compute @xmath79 as @xmath80 where @xmath81 is the standard deviation of the target star s differential magnitudes computed as the _ mean _ of the two best @xmath72 , @xmath73 , @xmath74 time series ( e.g. , @xmath82 ) , @xmath83 is the standard deviation of the differential magnitudes of the two best comparison stars ( e.g. , @xmath77 ) , and @xmath20 is the measurement precision of an individual differential magnitude .
we estimate @xmath20 by examining the standard deviations of the comparison star differential magnitudes ( @xmath75 , @xmath76 , and @xmath77 ) for each of the five target stars .
the lowest values are close to 0.0010 mag , representing the standard deviations of the most stable comparison stars , which we take to be @xmath20 .
the photometric results for the five stars are summarized in table [ photometry ] .
columns 4 and 5 give the duration of the photometric observations in days and the total number of measurements , respectively .
for all stars except 6 lyn , the duration of the photometric measurements approximately equals or exceeds the orbital period of the planetary companion , though not by more than a couple of cycles .
the standard deviations in columns 6 and 7 , @xmath84 and @xmath85 , vary somewhat from star to star due to differences in factors such as stellar brightness , airmass , and seasonal photometric quality .
our method of computing the absolute variability of each target star takes these differences into account .
finally , the value of each star s absolute variability level , @xmath79 , is given in column 8 of table [ photometry ] .
all values are significantly less than 0.001 mag ( 1 mmag ) and are consistent with each star being constant . in cases where both the comparison stars and the target star are especially stable
, random errors in their measurements can result in @xmath86 , i.e. , @xmath79 becomes imaginary .
this is the case for @xmath22 crb , for which we assume @xmath79 to be zero and the star to be constant .
hd 210702 has the largest value of @xmath79 ( 0.00074 mag ) , though it is still less than 1 mmag .
we classify hd 210702 as variable but append a colon to indicate some uncertainty .
we find nothing significant in our periodogram analyses of all five stars .
we conclude that there is no photometric evidence for brightness variability levels in any of these five stars that could call into question the existence of their planetary companions .
ccccccccc hd 167042 & 12 & 5412855004 & 876 & 275 & 0.00182 & 0.00205 & 0.00046 & constant + hd 192699 & 12 & 5419255004 & 812 & 106 & 0.00214 & 0.00267 & 0.00012 & constant +
hd 210702 & 12 & 5410054994 & 894 & 106 & 0.00218 & 0.00253 & 0.00074 & variable : + @xmath22 crb & 4 & 4909450250 & 1156 & 222 & 0.00094 & 0.00105 & 0.00000 & constant + 6 lyn & 12 & 5443754928 & 491 & 238 & 0.00153 & 0.00163 & 0.00011 &
information about the @xmath87 of planets around stars is critical for quantifying the occurrence rates of multiple jovian planetary systems , computing frequencies of different planet populations ( @xcite ) , and correcting for detection biases in radial velocity surveys ( @xcite ) .
we use monte carlo simulations of radial velocity measurements to assess the detection limits of stars in our sample .
we derive upper limits for the velocity semiamplitude ( @xmath88 ) as a function of period , which we convert to upper limits for minimum companion mass as a function of semimajor axis using stellar masses .
our method for computing detection limits is similar to that of @xcite .
we use the rms value of our radial velocities to determine which keplerian orbits of hypothetical planets are inconsistent with the observed rms values . for each star we begin with an initial orbital period and a large velocity semiamplitude ( @xmath89 ) .
we generate a series of synthetic radial velocity measurements with random phases and circular orbits on the same dates that the observations were taken to create a distribution of rms values for that specific period and @xmath89 .
noise is added to each synthetic measurement by drawing from a gaussian distribution with a standard deviation equal to the total measurement - plus - jitter error at that date .
if the distribution is inconsistent with the observed rms value at some confidence level ( we chose 95.4% in this analysis ) , then @xmath89 is decreased and a new distribution is created ( figure [ f11.eps ] ) . if the observed rms value is consistent with the distribution then we adopt that @xmath89 value as the velocity semiamplitude upper limit , @xmath88 , for that period .
this process is then repeated over all relevant periods .
the detection of a planet around a star increases the probability that other planets are in that system , as finding a planet suggests that the stellar system in question is dynamically and was historically amenable to planet formation and retention . for known planetary systems within 200 pc , @xcite find that additional planets are present at least 28% of the time .
it is therefore instructive to analyze the detection limits for the known planet - hosting stars in our sample so that certain regions of mass - semimajor axis phase space may be ruled out as not harboring additional planetary companions . to derive upper limits we generate 2500 artificial keplerian orbits for each @xmath89 .
the @xmath89 value is lowered in steps of 0.1 m s@xmath37 until the rms distribution is consistent with the observed rms value at the 95.4% level .
200 periods are evenly sampled from a log(@xmath10 ) distribution between 2 to 5000 days .
results for the five i m subgiants are presented in figure [ f12.eps ] and table [ detlimits ] .
our observations are sensitive to planet minimum masses between a few times that of neptune ( 0.053 @xmath4 ) to several jupiter masses for semimajor axes between 0.04 au to several au . with a high probability
we are able to exclude the existence of additional planets in the upper portions of each plot in figure [ f12.eps ] .
however , neptune - mass planets in circular orbits can not be ruled out by our observations . in general our detection limits
are sensitive to the observed rms value , the internal measurement errors , and the assumed stellar jitter level .
it is therefore unlikely that more observations will significantly improve the resulting detection limits as stellar jitter is the dominant contribution to the rms budget .
i m main sequence stars and giants both have higher mean jitter levels than subgiants ( e.g. , @xcite ; @xcite ) ; transit or microlensing surveys may therefore be the only tools available to study sub - neptune - mass planets in circular orbits around i m stars , although the larger radii and shorter evolutionary lifetimes of i m stars will make the detection of low - mass planets difficult even using these techniques .
overall our sample contains a variety of radial velocity characteristics ( see the `` notes '' column of table [ detlimits ] for a summary ) : five targets are known planet - hosting stars , nine exhibit linear velocity trends , and four have large rms velocity scatter @xmath14 900 m s@xmath37 ( which roughly corresponds to the semiamplitude of a 13 @xmath4 object orbiting a 1.5 @xmath18 star at 0.1 au ) .
the targets with large velocity scatter are all known spectroscopic binaries and include hd 8375 ( discussed in 3.1 ) , hd 65938 ( @xcite ; @xcite ) , hd 179799 ( @xcite ; @xcite ; @xcite ) , and hd 210211 ( spectroscopic triple system ; @xcite ; @xcite ; @xcite ) .
the two systems hd 90043 and hd 200964 both exhibit long - period ( @xmath0 1.5 yr ) radial velocity variations which are likely caused by planetary - mass companions . both stars are chromospherically quiet and do not show strong photometric variability .
ls periodograms confirm strong periodicities for both systems with false alarm probabilities @xmath19 0.1% .
our radial velocity coverage is currently insufficient to obtain unique orbital solutions so we withhold a detailed discussion of these objects for a future publication .
however , we note that single - planet orbital solutions fail to accurately reproduce the observed variations ; instead , two - planet solutions produce significantly better fits .
the inner planet of hd 90043 has a minimum mass of @xmath0 2.5 @xmath4 and is located at @xmath0 1.4 au ( @xmath10 @xmath0 1.2 yr ) , while the inner planet of hd 200964 has a minimum mass of @xmath0 2.0 @xmath4 and is located at @xmath0 1.7 au ( @xmath10 @xmath0 1.7 yr ) .
no additional stars in our sample exhibit periodicities with fap @xmath19 0.1% .
we derive detection limits for our full sample using the same parameters as in @xmath28[sg_detlimits ] for all stars with more than five radial velocity measurements .
the results are presented in table [ detlimits ] .
we subtract the best - fitting linear trends for stars exhibiting constant accelerations . excluding the large rms velocity stars , the median ( minimum ) mass upper limits for our sample are @xmath0 \{0.21 , 0.34 , 0.50 , 0.64 , 1.3 } @xmath4 at semimajor axes of \{0.1 , 0.3 , 0.6 , 1.0 , 3.0 } au at the 95.4% confidence level .
our sample of subgiants is therefore typically sensitive to saturn - mass planets out to @xmath0 0.1 au and jupiter - mass planets out to @xmath0 1 au . excluding stars with large rms velocity values
, planets @xmath14 3 @xmath4 can be excluded out to 1 au for all stars in our sample , and planets @xmath14 6 @xmath4 can be excluded out to 3 au for all but one of our stars . for the majority of our targets , however , our survey is sensitive to a few jupiter - mass planets within 3 au , or 1 @xmath4 planets within a few au .
lccccccccccccc 45410 & 31039 & 6 lyn & 1.82 & 0.15 & 0.27 & 0.37 & 0.49 & 0.86 & 9.3 & 64 & 5.01 & & p + 90043 & 50887 & 24 sex & 1.91 & 0.12 & 0.20 & 0.29 & 0.37 & 0.73 & 8.4 & 50 & 4.37 & & pp + 142091 & 77655 & @xmath22 crb & 1.84 & 0.07 & 0.12 & 0.17 & 0.22 & 0.37 & 6.2 & 62 & 5.15 & & p + 167042 & 89047 & & 1.72 & 0.11 & 0.18 & 0.25 & 0.34 & 0.59 & 7.7 & 68 & 5.27 & 2.1 & p , lt + 192699 & 99894 & & 1.69 & 0.20 & 0.34 & 0.49 & 0.62 & 1.40 & 10.5 & 50 & 5.05 & & p + 200964 & 104202 & & 1.70 & 0.13 & 0.22 & 0.33 & 0.42 & 0.81 & 8.7 & 91 & 5.02 & & pp + 210702 & 109577 & & 1.85 & 0.17 & 0.32 & 0.39 & 0.51 & 0.98 & 8.8 & 51 & 4.76 & & p + 587 & 840 & & 1.72 & 0.20 & 0.33 & 0.42 & 0.61 & 1.05 & 8.17 & 22 & 3.90 & 5.9 & lt + 8110 & 6289 & & 1.64 & 0.56 & 1.04 & 1.41 & 2.13 & 5.35 & 21.3 & 10 & 2.02 & & + 8375 & 6512 & & 1.66 & 0.42 & 0.68 & 0.96 & 1.25 & 2.49 & 19.0 & 25 & 4.05 & 52.9 & sb , lt + 22682 & 17049 & & 1.65 & 0.34 & 0.71 & 0.88 & 1.19 & 1.98 & 13.2 & 10 & 3.85 & & + 25975 & 19302 & & 1.59 & 0.21 & 0.32 & 0.44 & 0.63 & 0.95 & 7.71 & 16 & 4.69 & & + 27536 & 20263 & ek eri & 1.87 & 0.77 & 1.36 & 2.09 & 2.95 & 5.06 & 32.6 & 36 & 3.24 & & + 34538 & 24679 & & 1.53 & 0.28 & 0.51 & 0.70 & 0.86 & 1.66 & 11.2 & 12 & 4.13 & & + 37601 & 26942 & 24 cam & 1.93 & 0.28 & 0.42 & 0.68 & 0.86 & 1.54 & 10.1 & 17 & 4.01 & 9.2 & lt + 45506 & 30815 & & 1.84 & 0.16 & 0.29 & 0.43 & 0.53 & 0.84 & 6.80 & 14 & 4.21 & 21 & lt + 57707 & 35751 & & 1.81 & 0.40 & 0.91 & 0.95 & 1.17 & 2.25 & 10.8 & 6 & 2.24 & & + 65938 & 39198 & & 1.66 & & & & & & 2060 & 4 & 0.99 & & sb + 73764 & 42528 & & 1.91 & 0.48 & 0.73 & 0.96 & 1.33 & 1.99 & 13.1 & 9 & 3.30 & & + 103484 & 58110 & 6 vir & 1.98 & 0.70 & 1.10 & 1.62 & 2.71 & 3.90 & 25.8 & 40 & 4.09 & & + 109272 & 61296 & & 1.83 & 0.22 & 0.62 & 0.70 & 0.73 & 1.26 & 8.81 & 16 & 3.87 & & + 111028 & 62325 & 33 vir & 1.69 & 0.19 & 0.34 & 0.47 & 0.60 & 1.07 & 9.28 & 23 & 4.03 & 9.9 & lt + 123929 & 69185 & & 1.61 & 0.50 & 0.59 & 1.56 & 2.40 & 2.19 & 12.3 & 7 & 2.62 & & + 135944 & 74690 & & 1.62 & 0.17 & 0.31 & 0.43 & 0.52 & 0.95 & 7.68 & 16 & 4.10 & 53 & lt + 153226 & 82989 & & 1.76 & 0.19 & 0.33 & 0.50 & 0.64 & 1.04 & 7.85 & 18 & 3.93 & & + 179799 & 94521 & & 1.74 & 42.8 & 85.0 & 129 & 177 & 355 & 2510 & 8 & 3.67 & & sb + 184010 & 96016 & & 1.80 & 0.37 & 0.64 & 0.93 & 1.13 & 2.35 & 15.2 & 26 & 4.07 & & + 185351 & 96459 & & 1.91 & 0.17 & 0.31 & 0.43 & 0.52 & 1.11 & 7.09 & 17 & 3.93 & & + 202568 & 104941 & & 1.57 & 0.20 & 0.31 & 0.40 & 0.54 & 1.04 & 7.10 & 11 & 3.82 & 5.2 & lt + 210211 & 109281 & & 1.82 & 125 & 228 & 330 & 438 & 794 & 7930 & 8 & 2.25 & & sb + 210404 & 109338 & & 1.75 & 0.68 & 1.21 & 1.77 & 2.44 & 14.8 & 26.3 & 11 & 1.21 & 209 & lt +
several studies have provided hints that the occurrence rate of jovian planets increases with stellar mass . @xcite
combine the results of the california and carnegie planet search around fgk and m - type stars with the frequency of planets using the entire lick i m subgiant sample .
they found that the frequency of jovian planets appears to rise with stellar mass , reaching estimated rates of at least 9% for i m stars .
similarly , @xcite found a higher frequency of massive planets ( @xmath26 sin@xmath27 @xmath14 5 @xmath4 ) around i m stars than around solar - type stars .
these previous estimates of jovian planet frequency , however , have combined multiple published samples with different radial velocity sensitivities and different sample selection criteria . while our sample size is considerably smaller in this work , we benefit from having a uniform data set observed over a long ( @xmath0 5-year ) baseline with the same telescope and instrument setup
we also have the advantage of having well - characterized detection limits for the entire sample .
five stars in our sample are known planet - hosting subgiants and we identify two more ( hd 200964 and hd 90043 ) as being strong candidates for having planetary companions ( @xmath285.3 ) .
the number of jovian planet - hosting stars in our sample of 31 subgiants is therefore at least 7 .
bayes theorem can be used to derive the posterior pdf for the frequency of planets within @xmath0 3 au : @xmath90 where @xmath91 is the posterior pdf for the probability of a star hosting a planet ( @xmath92 ) given @xmath41 detections in a sample of @xmath93 stars , @xmath94 is the pdf for observing @xmath41 detections in the sample , and @xmath95 is the prior pdf for the probability that a star hosts a planet .
the detection of a planet represents a bernoulli trial , so @xmath94 follows a binomial distribution for the unknown parameter @xmath92 .
if we assume a uniform prior for @xmath95 between 0 and 1 , the posterior pdf @xmath91 simply follows a binomial distribution .
the resulting posterior pdfs are shown in figure [ f13.eps ] .
the median fraction of planet - hosting subgiants in our sample is 24@xmath96% , with upper and lower limits representing the range encompassing 68.3% of the distribution about the median .
if we exclude subgiants exhibiting large rms velocity scatter , which is indicative of a stellar companion , then the fraction increases to 26@xmath3% .
these values can be compared to quantitative predictions from theoretical modeling of planet formation . for
the mass range of our lick subgiant sample ( 1.5 - 2.0 @xmath18 ) , @xcite predict jovian planet occurrence rates to be @xmath0 10 - 15% .
our observed frequency is significantly higher , which may suggest that planet formation around i m stars is more efficient than previously thought .
the lower range of our results are , however , marginally consistent with the upper range predicted by @xcite .
our results can be compared to the giant planet frequencies derived by other authors for solar - type stars .
our survey is sensitive to giant planets with masses @xmath25 1 @xmath4 within a few au . @xcite
derive a planet frequency of @xmath0 4% for solar - type stars covering similar planet mass and semimajor axis ranges ; the giant planet frequency they derive is significantly lower than the frequency we find for i m stars . in a similar analysis , @xcite derive a frequency of @xmath00.5% for planets with masses @xmath14 5 @xmath4 within 2.5 au for solar - type stars .
over the same planet mass and semimajor axis range our program did not detect any planets . using the binomial theorem ,
zero out of 28 stars translates into a 2 @xmath13 upper limit of 10.1% for the frequency of @xmath14 5 @xmath4 planets within 2.5 au .
although poorly constrained , our frequency of the most massive giant planets around i m stars is consistent with the results from @xcite for solar - type stars .
finally , we note that , intriguingly , the planet occurrence rates for solar- and intermediate - mass stars resemble the occurrence rates of debris disks found through excess infrared emission ( @xmath015% for sun - like stars : @xcite ; @xcite ; @xmath2530% for i m stars : @xcite ; @xcite ; @xcite ) , although at the moment no correlation has been found between the presence of planets and debris disks ( @xcite ; @xcite ; @xcite ) .
the mass - period distribution of extrasolar planets around solar - type stars has been extensively studied in the literature ( e.g. , @xcite ; @xcite ; @xcite ) . the simplest and most common parametric technique to model the distribution of giant planets is to fit a double power law to a mass and period histogram : @xmath97 where @xmath5 is the number density of objects , @xmath9 is the planet ( minimum ) mass , and @xmath10 is the planet s orbital period .
values of @xmath11 and @xmath12 are generally consistent among authors ; we take the values from @xcite as being representative of studies in the literature for sun - like stars .
cumming et al .
find @xmath11 = 0.31 @xmath98 0.2 and @xmath12 = 0.26 @xmath98 0.1 for masses @xmath14 0.3 @xmath4 and periods @xmath19 2000 days .
qualitatively this means that the mass and period distributions for planets around solar - type stars rise sharply ( in linear space ) for lower masses and shorter periods , respectively .
can also be expressed as @xmath5 @xmath6 @xmath99 . in logarithmically - spaced bins the number of planets remains nearly constant .
] the probability of finding a low - mass , short - period planet is therefore much higher than finding a high - mass , long - period planet .
the masses and periods of planets being discovered around i m stars is qualitatively quite different from those being discovered around solar - type stars .
for example , out of @xmath14 20 planets discovered around i m stars , none orbit at semimajor axes less than 0.6 au ( @xmath0 130 days for a 1.75 @xmath17 star ) and none have minimum masses below 0.70 @xmath4 ( see figure [ f1.eps ] and table [ evolvedtab ] ) .
this has generally been attributed to a different mass - period distribution for planets around i m stars , but the influence of higher jitter levels on detection limits in surveys targeting i m stars is usually not taken into account and no quantitative analysis has yet been performed .
we use a monte carlo method to rigorously test the null hypothesis that the mass - period distribution for planets around solar - type and i m stars is the same .
the question we seek to answer is the following : given the values of @xmath11 and @xmath12 for solar - type stars , a planet occurrence rate , and the detection limits for our sample of 28 stars , what is the probability of finding 7 or more planets all having masses greater than 1.5 @xmath4 and semimajor axes greater than 1 au ?
value is small , so in this study we derive the probability of finding 7 _ or more _ planets with masses _ greater than _ 1.5 @xmath4 and semimajor axes _
greater than _ 1 au . ]
we first test whether we can reproduce the observed number and properties of planets in our sample using the same jovian planet frequency observed around solar - type stars . for each trial
we randomly draw a new value of @xmath11 and @xmath12 from a gaussian distribution centered on the values from cumming et al . with a standard deviation equal to the quoted uncertainties . for each star in our sample
we use the 10.5% planet frequency derived by cumming et al . to determine whether that particular star harbors a planet in our simulation .
if a star harbors a planet , we randomly draw a mass and period for that planet from the power law distribution in equation [ eqn : mp ] .
finally , we check to see whether that planet would have been detected around that star based on the star s detection limits .
we repeat this process @xmath100 times , saving the results after each trial .
the results of our first monte carlo simulation are shown in the left panels of figure [ f14.eps ] . out of @xmath100 trials , 24 produced 7 or more planets with any mass or semimajor axis , yielding a probability of 0.024% ( figure [ f14.eps ] : top left ) .
from the same simulation , however , no trials produced 7 or more planets _ with masses above 1.5 @xmath4 and semimajor axes greater than 1 au _ , yielding a probability of less than 10@xmath101 ( figure [ f14.eps ] : bottom left ) .
the input model values of @xmath11 and @xmath12 and/or the input planet frequency are therefore inconsistent with the number and properties of planets we observed in our sample of intermediate - mass stars . to determine whether increasing the planet frequency changes the results , we perform a similar monte carlo simulation except instead of the 10.5% occurrence rate we randomly draw a new planet frequency for each trial following the binomial distribution for 7 detections out of a sample of 28 ( yielding a median planet frequency of @xmath0 26% ) .
for this exercise , the values of @xmath11 and @xmath12 are the same as the cumming et al . values , but the probability of a star hosting a planet changes for each trial
. the results of this simulation are presented in the right - hand panels of figure [ f14.eps ] .
the probability of detecting 7 or more planets with any mass and semimajor axis rose to 6.3% , but the probability of detecting 7 or more planets with masses above 1.5 @xmath4 and semimajor axes larger than 1 au is 0.002% .
even when accounting for the different planet occurrence rates , we can rule out the sun - like mass - period distribution for planets around i m stars at a confidence level of over 4 @xmath13 .
these simulations can easily be extended to exclude a wider range of @xmath11 and @xmath12 parameter space with varying degrees of confidence .
we do this by running the same monte carlo simulation drawing from a binomial planet frequency distribution ( with 7 detections out of 28 stars ) over the following ranges of exponents : 2 @xmath19 @xmath11 @xmath19 8 ( @xmath102 = 0.5 ) and 2 @xmath19 @xmath12 @xmath19 8 ( @xmath103 = 0.5 ) . for each @xmath11-@xmath12 pair
we run @xmath104 trials .
the results are displayed in figure [ f15.eps ] ; each value of the grid represents the probability of detecting 7 or more planets with masses above 1.5 @xmath4 and semimajor axes greater than 1 au .
this technique allows us to exclude large regions of @xmath11-@xmath12 space as being inconsistent with our observations .
negative values and small positive values of @xmath11 and @xmath12 are rejected with high confidence levels , while larger values of @xmath11 and @xmath12 are able to reproduce the observed properties more often .
qualitatively this makes sense : we did not detect any low - mass ( small @xmath11 ) or short - period ( small @xmath12 ) planets .
we note that the inclusion of detection limits is critical in our analysis as it allows us to `` detect '' or
miss simulated planets based on whether they fall above or below our detection threshold .
little is known about the saturn- or neptune - mass population of planets around i m stars .
most planets orbiting i m stars discovered to date have masses @xmath25 2 @xmath4 ( table [ evolvedtab ] ) , leading to suggestions that more massive stars tend to produce more massive planets ( @xcite ) .
it is unclear , however , whether this is a _ bona fide _ property or a result of an observational bias caused by higher jitter levels in i m stars which might mask the signals of lower - mass planets .
moreover , few authors have assessed the limiting planet masses attainable by doppler surveys of evolved i m stars . despite the dearth of jovian planets at small semimajor axes , could a population of saturn - mass planets exist interior to 1 au around i m stars ? would the higher jitter levels of evolved i m stars prevent the discovery of neptune - mass planets ?
we address these questions in this study by deriving detection limits for our sample of 31 i m ( 1.5 @xmath19 @xmath2 @xmath19 2.0 ) subgiants .
typical detection limits for our stellar sample include planet masses down to @xmath0 \{0.21 , 0.34 , 0.50 , 0.64 , 1.3 } @xmath4 within \{0.1 , 0.3 , 0.6 , 1.0 , 3.0 } au at the 95.4% confidence level , excluding targets with large radial velocity scatter .
we can therefore rule out the existence of hot saturns within 0.1 au and jovian planets out to 1 au for most stars in our sample .
these detection limits suggest that the notably high masses of planets from our sample ( @xmath25 2 @xmath4 ) compared to planet masses around solar - type stars may be caused by a real difference in planet population characteristics .
for example , if a population of planets with random masses existed at 1 au then we would expect an observational bias to result in an observed mass distribution that was truncated near the typical mass detection limit . as jovian planets would have been detected out to @xmath0 1 au , the higher planet masses uncovered so far may be indicative of a real trend .
we test this idea quantitatively by comparing the mass - period distribution of planets around sun - like stars to the observed number and properties of planets in our sample . even when correcting for the higher planet occurrence rate found in our sample , the values of @xmath11 and @xmath12 in equation [ eqn : mp ] fail to reproduce the number ( 7/28 ) , masses ( @xmath14 1.5 @xmath4 ) , and semimajor axes ( @xmath14 1 au ) of planets from our sample at a confidence level of @xmath14 4 @xmath13 .
we conclude that the frequency and mass - period distribution of planets around i m stars is different from those around solar - type stars . increasing the mass of the host star by a mere factor of 1.5 - 2 results in an entirely new planet population which is characterized by a high frequency ( @xmath0 26% ) of high - mass planets ( @xmath26sin@xmath27 @xmath14 1.5 @xmath4 ) at large semimajor axes ( @xmath15 @xmath14 1 au ) .
the detection limits of the residuals of the planet - hosting stars in our sample show the strength of subgiant jitter levels and therefore trace the sensitivity levels attainable to doppler surveys targeting subgiants .
the residual rms velocities range from @xmath06 - 10 m s@xmath37 .
the detection limits indicate that planetary companions with minimum masses above \{0.2 , 0.4 , 0.5 , 0.7 , 1.4 } @xmath4 within \{0.1 , 0.3 , 0.6 , 1.0 , 3.0 } au can not exist in these systems .
if these detection limits are representative of the typical jitter levels of subgiants then the higher rms velocity values observed in other subgiants from our sample may suggest as - yet - unrecognized low - mass companions .
given our jitter - dominated detection limits , the prospects of discovering neptune - mass planets ( @xmath0 0.053 @xmath4 ) in circular orbits around i m stars using the doppler technique is not encouraging .
if the dominant source of jitter in subgiants is from @xmath92-mode oscillations , observing strategies that include longer integration times or repeated exposures over hour - long timescales may help to partially overcome this hurtle ( see @xcite ) .
our ongoing survey of intermediate - mass subgiants at keck observatory will address this issue by using the high rv precision achievable with the hires spectrometer and larger telescope aperture of keck compared to the lick 3 m .
we derive updated orbit solutions using new observations for five of the known planet - hosting i m subgiants in our sample , which were originally announced by @xcite , @xcite , and @xcite .
our results are in excellent agreement with those previously reported in the literature .
our parameter uncertainties are typically a factor of @xmath0 2 smaller than the published values as a result of longer baselines and more radial velocity measurements .
we also note that there is no evidence for periodicity in the residuals of the previously - known planet - hosting stars .
we acquired time - series photometric observations of the five known planet - hosting stars with the automatic photometric telescopes at fairborn observatory .
we find no evidence for brightness variation levels in any of the five stars that could call the existence of their planetary companions into question .
the eccentricities of three of the five planets in our sample with accurate orbital solutions are consistent with zero and emphasize a low - eccentricity trend for planets around i m subgiants , all but two of which have eccentricities below 0.3 ( table [ evolvedtab ] ) .
this is in contrast to planets around solar - type stars , which have approximately uniform eccentricity distributions between 0.0 and 0.8 for semimajor axes @xmath25 0.3 au ( @xcite ; @xcite ) .
interestingly , @xcite find that the eccentricity distributions for planets with masses @xmath19 1 @xmath4 peaks at @xmath44 @xmath19 0.2 , while the planets with masses @xmath14 1 @xmath4 have more uniformly distributed eccentricities between 0.0 @xmath19 @xmath44 @xmath19 0.6 .
planets around i m stars show the opposite trend : high - mass planets tend to have circular orbits .
we will explore this effect in detail in a future publication in this series .
information about planets orbiting i m stars has been driven by observations rather than theory . there is a growing need for stellar mass - dependent theoretical models of planet formation that make testable predictions about the physical and orbital characteristics of planets in this stellar mass regime .
specifically , the abundance of low - mass planets , the eccentricity distribution , and the fraction of multiple planetary systems are but a few of the many outstanding questions to be addressed in this young field .
we also encourage high - contrast imaging campaigns to include more a - type stars in their surveys .
five planets have already been directly imaged orbiting a stars : hr 8799 b , c , and d ( @xcite ) , formalhaut b ( @xcite ) , and @xmath12 pic b ( @xcite ) .
moreover , @xcite made their discovery after observing only a few early - type stars , in contrast to the many hundreds of late - type stars that have yielded null detections .
these results , combined with the higher inner planet occurrence rate we measure , suggest that more planet - hunting imaging surveys in this mass regime will yield fruitful results .
we extend our gratitude to the many cat observers who have helped with this project over the years , including chris mccarthy , raj sareen , howard isaacson , joshua goldston , bernie walp , julia kregenow , jason wright , and shannon patel .
we also gratefully acknowledge the efforts and dedication of the lick observatory staff , and the time assignment committee of the university of california for their generous allocations of observing time .
b. p. b. thanks chris beaumont and eric ford for productive discussions about mcmc techniques .
j. a. j. is an nsf astronomy and astrophysics postdoctoral fellow and acknowledges support from the nsf grant ast-0757887 .
we appreciate funding from nasa grant nng05gk92 g ( to g. w. m. ) .
g. w. h. acknowledges long - term support from nasa , nsf , tennessee state university , and the state of tennessee through its centers of excellence program .
d. a. f. is a cottrell science scholar of research corporation and acknowledges support from nasa grant nng05g164 g that made this work possible .
this research made use of the simbad database operated at csd , strasbourg france , and the nasa ads database . | we present an analysis of @xmath05 years of lick observatory radial velocity measurements targeting a uniform sample of 31 intermediate - mass subgiants ( 1.5 @xmath1 @xmath2 @xmath1 2.0 ) with the goal of measuring the occurrence rate of jovian planets around ( evolved ) a - type stars and comparing the distributions of their orbital and physical characteristics to those of planets around sun - like stars .
we provide updated orbital solutions incorporating new radial velocity measurements for five known planet - hosting stars in our sample ; uncertainties in the fitted parameters are assessed using a markov chain monte carlo method .
the frequency of jovian planets interior to 3 au is 26@xmath3% , which is significantly higher than the 5 - 10% frequency observed around solar - mass stars .
the median detection threshold for our sample includes minimum masses down to \{0.2 , 0.3 , 0.5 , 0.6 , 1.3 } @xmath4 within \{0.1 , 0.3 , 0.6 , 1.0 , 3.0 } au . to compare the properties of planets around intermediate - mass stars to those around solar - mass stars we synthesize a population of planets based on the parametric relationship @xmath5 @xmath6 @xmath7@xmath8ln@xmath9@xmath8ln@xmath10 , the observed planet frequency , and the detection limits we derived .
we find that the values of @xmath11 and @xmath12 for planets around solar - type stars from cumming et al .
fail to reproduce the observed properties of planets in our sample at the 4 @xmath13 level , even when accounting for the different planet occurrence rates .
thus , the properties of planets around a stars are markedly different than those around sun - like stars , suggesting that only a small ( @xmath0 50% ) increase in stellar mass has a large influence on the formation and orbital evolution of planets . |
partly because of the possibility that topological defects such as cosmic strings may have been formed in the early universe , and also because of the fact that it had already been noted that some solutions in @xmath0 gravity corresponding to spinless particles do not have closed timelike curves ( ctc ) if the total energy momentum ( @xmath1 ) is timelike @xcite , gott s solution @xcite has stimulated work discussing whether this spacetime is physically reasonable or not .
the relation between cosmic strings and @xmath0 particles comes from the property that the spacetime of an infinitely long and stationary gauge cosmic string asymptotically tends to minkowski spacetime with a deficit angle @xcite , and in the cases of interest the core is small enough that one can consider just minkowski with a conical singularity ( none of these properties holds for gauge but supermassive @xcite or global strings @xcite ) .
thus , gott s solution approximates the spacetime of two infinitely long parallel gauge cosmic strings , but it can also be thought of as the spacetime of two ( spinless ) particles in @xmath0 .
the first objections to gott s spacetime were due to the belief that it did not have an associated initial value problem , and to the fact that its total @xmath1 is timelike , in some similarity with tachyons @xcite . approximately at the same time , cutler showed that in gott s spacetime there are regions without ctc , in particular in these regions there are achronal , edgeless , not asymptotically null surfaces , so that it can be thought that the spacetime evolves from an initial data in any of these surfaces @xcite ( these surfaces must be suitably chosen , in this sense see also @xcite ) .
the apparent analogy with tachyons comes from the fact that parallel transport of vectors around a gott pair is the same as for a tachyon , but this is not true for parallel transport of spinors @xcite , basically because a gott pair satisfies the dominant energy condition ( also the weak and strong ones ) while a tachyon does not .
therefore , spatial @xmath1 must not necessarily be considered as unphysical ( for more discussions on ctc in @xmath0 gravity and on the non tachyonic character of a gott pair , see @xcite ) .
but then it remains intriguing that all known exact solutions describing spinless particles do not have ctc if their total @xmath1 is timelike @xcite .
kabat has suggested that this is a general feature , specifically , that spacetimes with spinless particles and timelike total @xmath1 do not have ctc @xcite . to this
we should add that t hooft has shown that although a gott pair can be produced from initial data with timelike @xmath1 momentum in a compact surface , a ` crunch ' will occur before the appearance of ctc @xcite .
time machine constructions have been associated with compactly generated cauchy horizons ( cgch ) @xcite .
this is , on one side , because if for certain initial data on a surface @xmath2 a domain of dependence without a cauchy horizon is obtained , and changing the data in a compact region of @xmath2 a cauchy horizon appears , beyond which ctc exist , then it is compactly generated . on the other side , in certain points of a cgch ( the so called base points )
strong causality is violated . in this work
we will follow this approach and take the question of whether a time machine can be constructed in an @xmath0 open spacetime with timelike total @xmath3 as equivalent to asking whether such a spacetime can have a cgch . the answer will be negative .
note that working with a cgch we get rid of a difficulty present in other formulations of kabat s conjecture .
this arises from the fact that it is not a priori obvious that , in a spacetime with ctc , a foliation in surfaces in which ` matter contributes positively ' exists , so that one can calculate the total @xmath1 via holonomy , without ` counting matter more than once ' . specifically
: we are interested in spacetimes arising from initial data , i.e. , of the form @xmath4 where @xmath5 is a simply connected , non compact , closed , achronal and edgeless surface and its future domain of dependence ( a stably causal region ) is denoted by @xmath4 . the dominant energy condition , i.e. , that @xmath6 is a future directed timelike or null vector for all future directed timelike or null @xmath7 , choosing @xmath7 as the normal to @xmath2 , ensures that total @xmath1 is independent of time ( a conserved quantity ) and of the foliation .
if there exists a cauchy horizon , @xmath8 , then the definition can be extended to the horizon if the matter ` crosses it ' , e.g. , assuming that there are no lightlike sources ; specifically , that @xmath9 is future directed and timelike for all future directed timelike @xmath7 ( deca ) . in this work
we will assume this energy condition but without requiring that @xmath9 is future directed ( decb ) , since in a cgch the weak energy condition ( wec ) is violated @xcite .
there are some previous results in connection with kabat s conjecture : seminara and menotti have shown , assuming additional rotational symmetry and the wec , that if there are no ctc at infinity then there are no ctc at all @xcite ; headrick and gott have shown that if a ctc is deformable to infinity , then the holonomy of the ctc itself can not be timelike , except for a rotation of @xmath10 @xcite .
nevertheless , in a noncompact cgch the wec is violated and the first result is not related to time machine construction in the sense made precise above , while the total holonomy of a ctc is in principle not related to the total @xmath1 , due to the problem mentioned in the preceding paragraph . in section ii and section iii we will summarize and discuss some known results that are crucial in our proof , which will be given in section iv .
since in section iv we will analyze the dynamics of a cgch in @xmath0 , we need here to recall some properties of cauchy horizons , obtained in @xmath11 gravity , but equally applicable to @xmath0 gravity .
let @xmath2 be a ( partial ) cauchy surface for @xmath12 , an orientable , time orientable spacetime with a future cauchy horizon @xmath13 .
then : 1 .
@xmath8 is compact iff @xmath2 is compact ( see , for instance , @xcite ) .
@xmath8 is differentiable everywhere except in a set of zero measure .
we will assume implicitly differentiability of the horizon each time it is needed .
that is , we will assume , e.g. , that the set of non differentiability is not dense ( in this sense , see @xcite ) .
@xmath8 is generated by null geodesics that are complete in the past but may be incomplete in the future @xcite .
let us denote them , generically , by @xmath14 , with @xmath15 some interval of @xmath16 and @xmath17 some affine parameter and , unless otherwise stated , we always refer to generators directed to the past .
4 . @xmath18 is defined as compactly generated if all these geodesics enter some compact , connected region @xmath19 and remain there forever .
that is , for each @xmath20 , there exists @xmath21 such that @xmath22 for @xmath23 @xcite .
5 . in a non compact cgch
the wec is violated , i.e. , there exist points in @xmath19 in which @xmath24 , with @xmath25 the tangent to the generators @xcite .
the base set , @xmath26 , is defined as the set of terminal accumulation points of null generators .
it can be seen that @xmath27 is non empty ( this follows from the completeness of the generators in the past and the compactness of @xmath19 ) , that strong causality is violated in @xmath27 ( also from the completeness of the generators in the past ) , and that @xmath27 is comprised by future and past inextendible null generators contained in @xmath27 , although not necessarily closed ( ` fountains ' ) @xcite ( as we will see in the last section , the last statement does not hold in @xmath0 ) .
from now on we will assume that the spacetime is open and that the total @xmath1 is timelike .
carroll et al . have shown that in such spacetimes , if they are composed of ( spinless ) particles , there can not exist any subsystem with spatial @xmath1 @xcite . in particular , a gott time machine can not be created out of the decay of a single cosmic string because there is not enough energy for that @xcite . in principle
this property is not obtained as a partial result in the version of kabat s conjecture that we prove here , since a gott pair satisfies the wec and , indeed , it can explicitly be seen that it does not have a cgch @xcite .
we mention it because a slight generalization is crucial in our proof .
so , we need here to summarize the analysis given in @xcite .
suppose , then , that the matter is composed by particles ( assuming implicitly , in this way , the deca ) .
the total @xmath1 as defined by holonomies is constructed starting from a trivial loop in @xmath2 ( at constant but arbitrary time ) and deforming it until it encircles all the particles . in the process
, the corresponding holonomic operator describes a curve ( let us call it @xmath28 ) in the lie group , which starts at the indentity ( corresponding to the trivial loop ) and finishes at the total @xmath1 .
we remark that ( up to similarity transformations ) the total @xmath1 does not depend on the way in which the deformation is carried out , although @xmath28 does not share this property , and is therefore not unique .
coordinates for the double covering of @xmath29 , @xmath30 , can be chosen by decomposing every element in a rotation through angle @xmath31 followed by a boost of rapidity @xmath32 and direction defined by the polar angle @xmath33 . in this
coordinates , the metric of @xmath30 ( naturally given by the structure constants ) is @xmath34 which shows that @xmath30 has the geometry of anti - de sitter spacetime .
a conformal diagram of ( the universal covering of ) this spacetime is shown in fig . [ fig1 ] , with one dimension suppressed and @xmath35 . systems with timelike ( spacelike ) total @xmath1 lie in region ii ( iii ) . since we have assumed that the total @xmath1 is timelike , the corresponding holonomic operator is equivalent ( through a similarity transformation ) to a rotation through a certain angle @xmath36 , which is defined as the total mass . since the topology of @xmath2 is assumed to be @xmath37 and ( it is assumed that ) the initial data are geodesically complete , @xmath38 where @xmath39 is the gaussian curvature , the equality and inequality follow from the gauss - bonnet and cohn - vossen theorems , respectively . at this point
we remark that the matter need not be composed of particles ( if it is not , the loop must encircle the support of @xmath40 or be deformed to infinity if this support is not compact ) and that the same analysis holds if one just assumes deca .
this condition implies that @xmath28 is timelike and future directed , a condition that in turn implies that no subsystem can have spacelike @xmath1 ; since , if @xmath28 crosses from region ii to region iii , it can not return back to region ii and lead to timelike total @xmath1 , as we did assume ( see @xcite for further details ) .
we should also remember that seminara and menotti have explicitly shown ( without using the causal structure of anti - de sitter spacetime , or the gauss - bonnet and cohn - vossen theorems ) similar properties @xcite . specifically ,
that in an open spacetime with timelike total @xmath1 satisfying deca the mass increases as the loop encircles more and more matter and if , having reached the total mass of @xmath10 , more matter is encircled then the total @xmath1 turns null or spacelike .
now recall that there is already a ` standard ' formulation for asymptotic flatness in @xmath0 , with analogues to the adm @xcite and bondi masses @xcite . in the hamiltonian formulation
it is required that asymptotically the spacetime approaches that of a ( spinless ) particle , i.e. , minkowski with a deficit angle : this angle defines the ` adm ' mass .
note that for such a spacetime the total holonomy of a loop that is deformed to infinity is equivalent to a rotation through an angle which coincides with the deficit angle , so the ` adm ' mass coincides with that defined via holonomies .
it was emphasized in @xcite that in @xmath0 , in contrast to @xmath11 , the total mass is not only bounded from below but also from above .
this was noted in a hamiltonian formulation , in which case the total mass must be strictly less than @xmath10 and it was argued that it is not a feature arising from such formulation analyzing the limit of a particle s spacetime when the mass approaches and subsequently exceeds @xmath10 , in which case the conical structure becomes cylindrical and subsequently geodesically incomplete .
what we want to remark is that , on account of eq .
( [ bonete ] ) , and the discussion that preceded it , if geodesically completeness is assumed , then the total mass is effectively bounded , it must be @xmath41 .
let us also remark that not every compact system is asymptotically flat in the sense described in @xcite , since a spacetime with spacelike or null total @xmath1 is not , even if it has @xmath40 with compact support and thus curvature of compact support ( the weyl tensor vanishes identically in @xmath0 ) .
let us assume now that decb holds .
then @xmath28 is timelike but not necessarily future directed and there are some subtleties that we need to discuss .
firstly , we shall assume that the total mass is non negative .
therefore , near the identity @xmath28 can be chosen non past directed by simply choosing the point where the loop is initially expanded as one in which there is non negative mass .
now note that without assuming deca the reasoning that led to the non existence of subsystems with spacelike @xmath1 is , in principle , no longer true ( see fig .
[ fig2 ] ) .
we shall overcome this difficulty by imposing that , once that @xmath28 has been chosen non past directed at the identity , it remains in region ii .
considering the previous sections , the proof of the version of kabat s conjecture we give here reduces to showing that open @xmath0 spacetimes with a cgch must have a subsystem with null or spacelike @xmath1 , or timelike @xmath1 of mass zero or @xmath10 . imposing timelike total @xmath3 and the decb , we will have arrived at a contradiction , except for the last two cases , which will turn out to be unstable , in an appropriate sense .
such a subsystem is , precisely , that encircled by the closed null geodesic whose existence we will now prove .
so , in @xmath0 , not only strong but also stable causality is violated : the base set necessarily contains at least one closed null geodesic @xmath42 .
the proof follows from the poincar - bendixon - schwartz theorem @xcite applied to the dynamical system defined by the past directed null generators @xmath43 in the two - dimensional compact manifold @xmath19 .
we start by noting some properties of this dynamical system .
@xmath19 is positively invariant ( this results from the definition of @xmath19 ) .
2 . the past directed generators @xmath43 in @xmath19 exist globally ( they are complete in the past , and we are always referring to this direction in time ) .
3 . from completeness of the generators in the past ,
there are no fixed points , i.e. , there does not exist @xmath44 such that @xmath45 .
the first item allows us to think of our dynamical system as one in a compact manifold .
the second item allows us to introduce what is usually called the ` @xmath46 limit set ' of a point @xmath47 .
this set is defined as the set of points @xmath44 such that the generator passing through @xmath48 satisfies : for every open neighborhood @xmath49 of @xmath50 and every @xmath21 ( in the domain in which they are defined ) there exists @xmath51 such that @xmath52 .
the ` @xmath46 limit set of @xmath19 ' is , similarly , defined as the union of the @xmath46 limit sets of all @xmath53 .
that is , the @xmath46 limit set of @xmath19 is _ by definition _ the base set @xmath27 . with this in mind
, we shall replace @xmath46 by @xmath27 .
the poincar - bendixon - schwartz theorem shows the following : let @xmath54 be a compact , connected , orientable two dimensional manifold with @xmath55 and complete orbits , such that , for @xmath47 , @xmath56 contains no fixed points ( our dynamical system does satisfy these conditions ) .
then either * @xmath57 ; or * @xmath58 is a closed orbit @xmath42 , and @xmath59 winds towards @xmath42 , where @xmath60 is the torus , a manifold without boundary , excluded in our case .
thus we have a closed null geodesic @xmath42 . in the proof of the poincar - bendixon - schwartz theorem
the 2-dimensionality of @xmath19 is crucial . indeed , although not related to the failure of this theorem or the properties of cgch in @xmath0 gravity , it has been emphasized that ( in @xmath11 ) the base points are not , in general , made up by closed null geodesics ( ` fountains ' ) @xcite .
we will now show that the @xmath1 encircled by @xmath42 is spatial , null or timelike of mass zero or @xmath10 .
for that purpose consider an arbitrary base point @xmath61 , and an orthonormal base @xmath62 .
parallel transport of such a base around the loop @xmath42 defines a new orthonormal base @xmath63 , related to the previous one by a proper lorentz transformation @xmath64 , whose equivalence class defines the @xmath1 encircled by @xmath42 . in particular , let us take a base such that the tangent vector to @xmath65 at @xmath66 , @xmath25 , belongs to @xmath67 .
@xmath42 being a geodesic , @xmath25 is parallel transported and , since it is a non broken geodesic , @xmath68 ( defined by @xmath69 ) is proportional to @xmath25 , i.e. , @xmath70 . in other words , @xmath25 is an eigenvector of @xmath71 .
it is straightforward to see that if @xmath71 is timelike it has no null eigenvector , except when @xmath71 is the identity , in which case the eigenvalue is obviously @xmath72 and all the vectors are eigenvectors .
when @xmath71 is null it has exactly one null eigenvector , with eigenvalue @xmath72 and when it is spatial it has two null eigenvectors , with eigenvalues @xmath73 and @xmath74 , so the first is attractive and the second repulsive .
then , the situation is different from @xmath11 , because in that case @xmath71 defines a map on the ( past ) sphere of null directions , and every orientation preserving map on the sphere has at least one fixed point , so that in @xmath11 every proper lorentz transformation has at least one fixed null direction @xcite .
on the other hand , a homeomorphism on the circle which preserves orientation has a fixed point iff its _ rotation number _ is zero @xcite ; in particular for rotations this number coincides with the angle of rotation , so one recovers that if @xmath71 is timelike it has null eigenvectors iff it is the identity .
the point is that , applying the analysis of the previous paragraph to the closed null geodesic @xmath42 , we have shown that the @xmath1 it encircles is spatial , null or timelike of mass zero or @xmath75 .
although it is not related to our proof , remember that the eigenvalue can not be @xmath76 because otherwise it can be shown that there would be ctc in @xmath77 @xcite ( see also proposition 6.4.4 of @xcite ) , a stably causal region .
so , when the @xmath1 encircled by @xmath42 is spacelike the corresponding eigenvalue is @xmath78 . summarizing , if the total @xmath1 is timelike and decb holds , there are no subsystems with null or spacelike @xmath1 . thereby , the @xmath1 encircled by @xmath42 can only be timelike of mass zero or @xmath75 .
if we had supposed deca we would have been able to discard the case of zero mass because @xmath42 would encircle a simply connected flat ( vacuum @xmath79 the identical vanishing of the weyl tensor ) region and thus causally well behaved .
however , since wec is violated , zero mass does not necessarily correspond to vacuum .
nevertheless , both cases , subsystems with zero or @xmath10 mass , are unstable , in the sense that in every neighborhood ( with the lie group manifold topology ) of these points all timelike elements do not have fixed null directions ( these two points belong to the boundary of region ii ) . by slightly altering the distribution of masses
( or @xmath28 , equivalently ) there will be no subsystem with @xmath80 or @xmath10 mass , see fig .
carroll et .
al have shown , from energy considerations , that a gott time machine can not be constructed in @xmath0 dimensional open gravity with timelike sources and total energy momentum . in this paper
we have shown that , in a precise sense , a time machine can not be constructed at all , providing a proof of a suitable version of kabat s conjecture .
note that it makes sense to talk about total @xmath1 of a time machine : this quantity remains unalterated if the initial data is changed in a compact region of @xmath2 , since it is defined by a loop that encircles ( in particular ) such a region .
the proof is constructive in some aspects , e.g. , it shows the existence of closed null geodesics in a cgch , a property which is interesting in its own . in a non compact cgch
the wec is violated , the known classical fields obey this condition but they do not when quantized ( even in minkowski ) , although an averaged version , the averaged null energy condition , has been proven to hold in some cases @xcite .
therefore , it can be said that in order to create a time machine quantum matter is needed , and it is natural to ask whether the laws of physics allow ctc or a cgch .
there have been different and opposite conclusions to this question ( see , for instance , @xcite ) and it seems reasonable to say that it will be difficult to have a complete answer within semiclassical gravity since not even the usual quantum field theory can be extended from @xmath81 to the base set of a cgch @xcite .
the author would like to thank r. j. gleiser , j. pullin , and g. a. raggio for encouraging this work and helpful suggestions which improved its presentation , and conicor for financial support .
this work was supported in part by grants from the national university of crdoba , and from conicor , and conicet ( argentina ) .
s. deser , r. jackiw , and g. t hooft , ann . phys . * 152 * , 220 ( 1984 ) .
h. waelbroeck , gen .
* 23 * , 219 ( 1991 ) .
r. j. gott iii , phys .
lett . * 66 * , 1126 ( 1991 ) .
d. garfinkle , phys .
d * 32 * , 1323 ( 1985 ) .
laguna and d. garfinkle , phys .
d * 40 * , 1011 ( 1989 ) . g. w. gibbons , m. e. ortiz , and f. r. ruiz , phys .
d * 39 * , 1546 ( 1989 ) ; d. harari and p. sikivie , phys .
d * 37 * , 3438 ( 1988 ) ; r. gregory , phys .
lett * b 215 * , 663 ( 1988 ) ; a. g. cohen and d. b. kaplan , phys .
lett * b 215 * , 67 ( 1988 ) .
s. deser , r. jackiw , and g. t hooft , phys .
* 68 * , 267 ( 1991 ) . c. cutler , phys .
d. * 45 * , 487 ( 1992 ) .
a. ori , phys .
d * 44 * , r2214 ( 1991 ) .
s. m. carroll , e. farhi , a. h. guth , and k. d. olum , phys .
d. * 50 * , 6190 ( 1994 ) .
m. p. headrick and j. r. gott iii , phys .
d. * 50 * , 7244 ( 1994 ) .
d. n. kabat , phys .
d * 46 * , 2720 ( 1992 ) .
g t hooft , class . quantum grav .
* 9 * , 1335 ( 1992 ) ; * 10 * , 1023 ( 1993 ) .
b. s. kay , m. j. radzikowski , and r. m. wald , commun .
phys . * 183 * , 533 ( 1997 ) . s. w. hawking , phys .
d. * 46 * , 603 ( 1992 ) .
p. menotti and d. seminara , ann . phys . * 240 * , 203 ( 1996 ) .
r. m. wald , _ general relativity _
( the university of chicago press , 1984 ) .
s. w. hawking , g. f. r. ellis , _ the large scale structure of spacetime _
( cambridge university press , 1976 ) .
p. t. crhusciel and j. isenberg , kannaskis chaos , 113 ( 1993 ) .
s. m. carroll , e. farhi , and a. h. guth , phys .
* 68 * , 263 ( 1992 ) ; * 68 * , 3368 ( e ) ( 1992 ) .
p. menotti and d. seminara , phys . lett . * b 301 * , 25 ( 1993 ) ; * b 307 * , 404 ( e ) ( 1993 ) .
a. ashtekar and m. varadarajan , phys .
d. * 50 * , 4944 ( 1994 ) .
a. ashtekar , j. bicak , and b. schmidt , phys .
d. * 55 * , 669 ( 1997 ) .
a. schwartz , amer . j. math . *
85 * , 453 ( 1963 ) .
p. t. chrusciel and g. j. galloway , * gr - qc/9611032*. r. penrose , w. rindler , _ spinors and spacetime _
( cambridge university press , 1984 ) .
r. l. devaney , _ an introduction to chaotic dynamical systems _ ( addison - wesley , 1985 ) .
r. m. wald and u. yurtsever , phys .
d. * 44 * 403 ( 1991 ) ; g. klinkhammer , phys .
d. * 43 * , 2542 ( 1991 ) .
s. w. kim and k. s. thorne , phys .
d. * 43 * , 3929 ( 1991 ) . | it is shown that in @xmath0 dimensional gravity an open spacetime with timelike sources and total energy momentum can not have a stable compactly generated cauchy horizon .
this constitutes a proof of a version of kabat s conjecture and shows , in particular , that not only a gott time machine can not be formed from processes such as the decay of a single cosmic string as has been shown by carroll et al . , but that , in a precise sense , a time machine can not be constructed at all . |
galaxies display a wide variety of physical and observational properties .
it is well known that the distribution of galaxy optical colours is bimodal , e.g. blue cloud versus the red sequence @xcite .
the bimodality of the galaxy population exists at least out to @xmath11 1 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
a number of different mechanisms ( taking place in different environments ) have been proposed for the observed bimodality of the galaxy population , including , but not limited to , galaxy merging ( major and minor ) , galaxy strangulation and harassment , ram - pressure stripping as well as agn feedback ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
such mechanisms could regulate the observed optical colours of galaxies by influencing their key physical parameters such as star formation history ( sfh ) , mean age of stellar populations , the amount of dust attenuation , dust geometry and metallicity @xcite .
+ + besides , there are substantial differences between galaxy populations in the field and those in clusters and groups . according to @xcite ,
galaxy morphology is a strong function of galaxy density , i.e. the morphology - density relation , and numerous studies since then have shown the dependence of galaxy properties on the local environment @xcite .
for example , the red population is substantially dominated by early - type galaxies and thus preferentially found in high - galaxy density environments , while blue galaxies are predominantly late - type systems and mostly found in low - galaxy density environments , i.e. the colour - density relation . moreover
, vast majority of galaxies in the blue cloud are actively forming stars while the red sequence consists mainly of passive galaxies with little or no ongoing star formation .
there are also additional contributions to the red cloud from ( a ) heavily obscured star - forming or edge - on galaxies and ( b ) galaxies with passive disks , e.g. red spirals showing signs of low - level of star formation , which are known to be considerably redder and more massive than their blue / star - forming counterparts @xcite .
it is noteworthy that the morphology - density and color - density relations evolve with redshift ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
analyses of the dust attenuation in active / star - forming galaxies suggest that in contrast to passive galaxies , they are heavily affected by dust @xcite .
it has been shown that the bulk of the dust in late - type galaxies is in the cold phase and as consequence emits at @xmath9100 , i.e. the far - infrared ( fir ) and wavelengths @xcite .
such wavelengths are covered by the instruments on board the _ herschel space observatory _ @xcite .
thus , the data collected by _
herschel _ is uniquely suited to probe the dusty component , e.g. its characteristics and origin , in all type of galaxies , in particular early - type galaxies which contain significantly less dust than late - type systems .
the existence of dust in early - type galaxies has been first reported from studying the absorption of stellar light @xcite and since then several studies have been conducted in order to shed light on the quantitative dust content of eary - type galaxies@xcite . however , data provided by _
herschel _ have enabled us to study dust properties , e.g. its total luminosity , mass and temperature in early - type galaxies in an unprecedented manner due to a better sensitivity , resolution and/or the long wavelength coverage necessary @xcite . among various surveys ,
the _ herschel _ astrophysical terahertz large area survey ( h - atlas ; * ? ? ?
* ) is the widest extragalactic survey undertaken in with _
herschel_. the large coverage of h - atlas helps to have a better statistical view of the dust content and its characteristic among galaxies spanning a broad range of luminosities , colours and morphologies .
results from @xcite as part of the h - atlas science demonstration phase ( sdp ) and based on the uv - optical colour classification , show that the majority of sources ( @xmath12 95 per cent ) with detections at low redshift ( z @xmath0 0.2 ) , are blue / star - forming galaxies with uv - optical colours @xmath0 4.5 . this earlier study suggested that the submm - detected / optically - red galaxies ( @xmath9 4.5 ) , with a contribution of @xmath13 5 per cent to the total number of detections , are more likely to be star - forming galaxies and that their red colors are due to obscuration by dust . from a stacking analysis at wavelengths
, @xcite performed a large - scale statistical study of the properties of optically selected galaxies ( based on the rest - frame color @xmath14 ) at z @xmath13 0.35 , and found that approximately 20 per cent of low - redshift galaxies in h - atlas are red . in the mean time , there have been several h - atlas studies trying to shed light on the existence and properties of dust in early - type galaxies . for instance
@xcite used data from the h - atlas sdp to study dust properties and star formation histories in a sample of low redshift galaxies ( z @xmath13 0.5 ) detected at wavelengths .
followed by classification of their sample based on optical morphology , @xcite found that @xmath12 4.1 per cent of all detections are early - type systems and that @xmath12 3.8 per cent ( 19 out of 496 ) of spiral galaxies with detections are passive . in another study and by using samples of early - type galaxies at low redshifts ( 0.013 @xmath13 z @xmath13 0.06 ) ,
@xcite found that early - type galaxies with h - atlas detections ( based on phase 1 version 2.0 internal release of the h - atlas catalogue ) , are not only bluer in the uv - optical colours but also are significantly brighter in nuv in comparison to their h - atlas none - detected counterparts .
the aim of this work is to examine in more detail the nature of detected red galaxies using the data of h - atlas .
the main difference between this work and those conducted by @xcite and @xcite is that all sources in our sample are detected in h - atlas and classified by means of the uv - optical colour index .
our main objectives are : to segregate intrinsically red galaxies from heavily obscured star forming galaxies , and subsequently discuss the origin and the role of the dust in passive systems .
the main improvements compared to our previous study come from : * a larger area coverage ( by a factor of @xmath15 ) and therefore a better statistics * the inclusion of complimentary wavelengths in the mid - infrared ( mir ) bands * the extraction of various physical parameters from multi - wavelengths observations of sources by means of the sed fitting .
the paper is organized as following : in section [ sec : data ] , we present the data from h - atlas phase 1 and select a sample of low redshift galaxies , all detected with _
herschel _ in the spire 250 band . in section [ sec
: analysis ] , we select sub - samples of optically blue and red galaxies and analyze their physical characteristics such as star formation activities and dust properties as inferred from fitting their spectral energy distributions . our main finding and conclusion
are given in section [ sec : conclusion ] . throughout the paper ,
we assume a concordance cdm cosmology with @xmath16 = 70 km s@xmath17 mpc@xmath17 , @xmath18 = 0.3 and @xmath19 = 0.7 .
we use data from the h - atlas phase 1 version 3.0 internal release which contains the ids of sigma spire detections at 250 and is reduced in a similar way to the sdp data , as described by @xcite , @xcite , @xcite and @xcite .
the phase 1 i d catalogues have been produced in a similar way to @xcite and will be presented in ( * ? ? ? * ; in prep ) . initially observed time - line data from spire and pacs instruments were processed by using the herschel interactive processing environment ( hipe ) based on a custom reduction scripts .
high - pass filtering was then applied to the data time - lines in order to correct the thermal drift in bolometer arrays .
cross - scan time - line observations were projected by using the naive map - making method of hipe . for point like sources ,
catalogue of sigma fluxes were produced from the 250 psf filtered map , using the madx algorithm ( maddox et al . in prep ) , as described in @xcite . for extended sources ,
larger apertures were chosen such that they match the extent of the source emission .
for each 250 source , corresponding 350 and 500 flux densities were estimated by using the 350 and 500 maps ( noise- weighted / beam - convolved ) at the source position extracted from the 250 map .
finally 100 and 160 aperture flux densities were measured following matching each 250 source to the nearest pacs sources within a radius of 10 arcsec .
a likelihood - ratio analysis @xcite was performed by @xcite to match 250 sources to the sdss dr7 @xcite sources brighter than r = 22.4 mag within a 10 arcsec radius .
the probability that an optical source is associated with the source has been used to define the reliability of an association . according to @xcite ,
objects with reliability @xmath20 0.8 are considered to be true matches to sources .
the h - atlas fields are along the celestial equator centred at ra of 9h(g09 ) , 12h(g12 ) and 14.5h(g15 ) .
144 2 out of the 161 2 covered by h - atlas overlap with the galaxy and mass assembly ( gama i ) survey @xcite .
the gama survey re - processes and combines optical data from the sloan digital sky survey ( sdss dr6 ; * ? ? ?
* ) , nir data from the ukirt infrared deep sky survey ( ukidss ) large area survey ( las dr4 ; * ? ? ?
* ) , and uv from the galaxy evolution explorer ( galex ; * ? ? ?
the pre - processing of the gama , sdss and ukidss archive data is descibed in detail in @xcite .
for all galaxies with @xmath21 19.4 mag in g09 and g15 as well as @xmath21 19.8 mag in g12 , redshifts have been measured using the anglo australian telescope and for brighter galaxies , redshift estimates are taken from other existing redshift surveys such as sdss , the 2df galaxy redshift survey ( 2dfgrs ) and the millennium galaxy catalogue ( mgc ; * ? ? ?
* ; * ? ? ?
furthermore , the gama-_wise _ ( the wide - field infrared survey explorer ; * ? ? ?
* ) catalogue adds coverage in four mir bands at 3.4 , 4.6 , 12 and 22 @xcite . in summary
, we have at our disposal uv , optical and mir data as well as redshift estimates for the galaxies within the h - atlas / gama - overlapping area where all selected sources in our sample satisfy the following criteria : * they all have sigma detected at spire 250 .
* they fall within the redshift range @xmath22 .
we only select objects with a sufficiently reliable spectroscopic determination ( i.e. @xmath233 ; * ? ? ?
all galaxies have a reliability parameter ( ` reliability ` @xmath20 0.8 ) of being associated with an optical counterpart in the sdss r - band catalogue , for which multi - wavelength photometry is available . as such , in addition to the 250 emission , all sources ( 7131 objects ) have corresponding fluxes ( all corrected for galactic extinction ) via aperture matched photometry in other bands ranging from uv to mir . *
since a crucial aspect of our selection of red galaxies is based on the uv - optical ( ) colour , we remove from our sample those galaxies for which their nuv fluxes as estimated in gama , differ by more than @xmath90.5 magnitude from those retrieved through _ galex _ gr6 data release based on the all - sky imaging survey ( ais ) data products ( nuv depth @xmath24 20.8 mag ) . in addition , all selected sources have nuv magnitude errors , as provided by _
galex_-gr6 , which are @xmath0 0.2 mag .
this guarantees that all sources in our sample have enough signal - to - noise ratio in uv .
the above constraints on uv fluxes , reduces our sample to 4016 sources . *
finally since the physical parameters inferred for each galaxy are based on sed fitting techniques , an extra criterion has been applied in order to exclude sources ( 234 in total ) with poor quality sed fits ( see sec . [
sec : sed ] ) .
after applying these selection criteria , we find @xmath25 galaxies with detections in at least nuv + _ u , g , r , i , z _ and 250 bands .
distributions of the sdss @xmath26-band and nuv magnitudes for all galaxies as well as those qualified to be included for the subsequent data analysis are shown in panels of fig.[fig : petrosian - r ] . according to the first panel ,
approximately @xmath1 13 per cent of the initial sources were excluded following the requirement of a uv detection for inclusion in the sample .
but that does not seems to exclude systematically any particular type of sources as a kolmogorov - smirnov test ( ks test ) suggests a @xmath5 70.0 per cent probability that the distribution of sources detected at 250 is similar to the one being observed simultaneously in the 250 @xmath27 nuv bands .
however by limiting errors in the nuv band to @xmath13 0.2 mag , more sources ( @xmath1 31 per cent ) are excluded in particular faint objects in the nuv band .
a subset of sources have also detections in _ galex _ fuv , pacs ( 100 , 160 ) and spire ( 350 , 500 ) bands . _ wise _ data are available and recently have been cross - matched , with extended sources from _ wise _ accounted for correctly , for all gama fields .
yet at the time of analysing galaxy seds in this work , _ wise _ data were only available for the g12 and g15 fields .
thus , out of the 3782 sources , 2622 ( @xmath1 70 per cent ) have also aperture - matched _
wise_-mir data .
though the vast majority of galaxies at low redshift with detection are star - forming and optically blue , a small fraction of them are red in optical bands ( e.g. @xmath28 , @xmath14 ) .
we separate blue and red galaxies in the sample using the uv - optical index .
this is more robust than optical colour indices as it is more sensitive to recent star - formation activity ( e.g. , * ? ? ?
@xcite separate red and blue galaxies in the h - atlas sample at @xmath29 4.5 , estimated through fitting a double gaussian to the colour distribution of galaxies , with redshifts @xmath22 ( i.e. similar to the present work ) , in the h - atlas sdp data .
hence any source with nuv@xmath30 mag is considered as _ red _ , while _ blue _ objects are those with nuv@xmath31 mag . as fig.[fig : petrosian - r]a shows , the majority of the galaxies in our sample have apparent @xmath26-band magnitudes @xmath13 17.5 mag and nuv magnitudes @xmath5 19.0 mag . in order to ensure that none of the emission has been contaminated by synchrotron emission from radio - jets hosted by active galactic nuclei ( agns ) , we find and exclude radio agn as follows .
we cross matched the sdss position of our sources with those from the full , unfiltered radio - source catalogue of @xcite .
the radio catalogue consists of all sources detected in the h - atlas phase 1 field by the nrao vla sky survey ( nvss ; * ? ? ?
* ) and , as such , contains 7823 sources .
the outcome is 117 matches having separations of @xmath32 arcsec . in order to determine whether the radio emission was consistent with the presence of a radio - loud agn , we calculated @xmath33 , defined as : where @xmath35 and @xmath36 are fluxes at 250 and 1.4-ghz for all matched sources respectively . if @xmath37 then part of the radio - emission is due to agn activity @xcite .
conservatively , we exclude any source which satisfies this criterion in order to ensure none of the submm emission may be contaminated by radio agn activity .
out of 117 sources with radio counterparts , only 13 sources ( 1 and 12 galaxies ) have @xmath37 and are thus excluded from the subsequent analysis . the morphology of all 117 galaxies were examined from their sdss @xmath26-band images , following independent visual inspection by three team members .
galaxies were classified into three categories of elliptical ( e ) , spiral ( s ) and uncertain ( u ) .
the number of sources in each morphological type is 37 , 48 and 32 for the e , s and u galaxies respectively ( see fig.[fig : red - morphology - pie - chart ] ) .
many of sources classified as u are too small in the sdss images to judge their morphology and can be of any type , i.e. spiral , elliptical or merging galaxies . in order to test the validity of this morphological classification , we compared our classification to an independent morphological classification based on the index @xmath38 which we obtained from the sdss dr7 galaxy catalogue @xcite .
different studies have adopted different thresholds of theindex above / below which a galaxy is considered as early / late type .
for instance , @xcite adopts @xmath38=2.0 to divide their sample into early and late types though indices of @xmath392.5 have been also used to describe early - type sources ( e.g. , * ? ? ? * ; * ? ? ?
[ fig : red - morphology ] ( panel a ) displays the distributions of indices for all galaxies in our sample , i.e. the sample as well as the morphologically classified galaxies - values ) are reported in the ks - test table .
[ table : kstest ] . ] . from this figure , it is clear that the distribution of indices for the -e sample peaks around @xmath40 .
this is larger than those estimated for the s galaxies ( either or ) .
the index distribution of the -u galaxies lies somewhat between those of the s and e samples .
an inspection of the ellipticity parameter ) where @xmath41 and @xmath42 are the galaxy s semi - major and semi - minor axes as measured in the sdss . ] of all galaxies in the sample ( fig .
[ fig : red - morphology ] , panel b ) reveals that , not surprisingly , in sources of type s , @xmath43 0.5 whereas in galaxies of type e , @xmath44 0.5 .
in fact the disk structure is extremely pronounced in highly inclined spiral galaxies and therefore the majority of galaxies in the s category are those having larger ellipticities .
this is better shown in fig .
[ fig : red - morphology]c where histograms of galaxy inclination angles ( @xmath45 ) for , -s , -e as well as s+u samples are plotted .
inclinations are determined from the relation in which @xmath47 is the ratio of the smallest to the largest axis of an oblate spheroid of rotation .
we assume @xmath47 = 0.20 which is an appropriate value to use for the intrinsic flattening of the distribution of the light of galactic spheroids ( e.g. ; * ? ? ?
* ) . unlike and -e galaxies ,
the majority of -s galaxies are highly inclined .
note that , even in the combined -u @xmath27 -s sample , there is still and excess of galaxies with relatively large inclination angles in comparison to the and -e samples . to illustrate this , we show in fig .
[ fig : red - morphology]c the distribution of inclination angles as expected from a random sampling .
the observed difference between the distribution of -(s@xmath27u ) galaxies in comparison to a sample of simulated inclinations , suggests that the fraction of highly inclined systems in -(s@xmath27u ) sample is more than one would expect for a random distribution .
this shows that the inclination angle play a non - negligible role in the observed red colour of -s systems .
the main conclusion is that the -e sample consists of intrinsically red objects while the -s sample contains galaxies where inclination could be a dominant factor in determining the observed red optical colours .
although these inclined sources are not the main interest of this paper , we do discuss some of their ensemble properties in sec .
[ subsub : sfr - red - s ] . in order to explore the environmental density of galaxies and
see if it plays an important role in shaping their observed properties , we compute the projected surface density around each galaxy .
this is based on counting the number of nearest neighbours , i.e. the density within the distance to the nth nearest neighbour .
hence , the surface density to the fifth nearest neighbour is calculated as : where @xmath49 is the projected co - moving distance to the fifth nearest neighbour within a volume - limited density - defining population and relative velocity @xmath501000 km s@xmath17 @xcite .
the density - defining population ( ddp ) are galaxies brighter than @xmath51 .
densities are calculated for galaxies with @xmath0 19.4 ( where is the r - band petrosian magnitude ) , @xmath52 and with reliable redshifts ( @xmath53 ; * ? ? ?
although eq .
[ eq : density - sigma5 ] is a 2d estimate , the redshift information of each galaxy is used to remove the background and foreground sources .
[ fig : density - sigma5 ] displays histograms of the projected densities for and galaxies within the redshift range of @xmath52 and for all systems having @xmath54 @xmath55 .
this decreases the overall number of galaxies by @xmath1 4.2 per cent ( out of 117 galaxies , two have @xmath56 and three have @xmath57 @xmath55 ) .
although the highest observed density ( 1.5 @xmath58 2.5 ) is populated by a small fraction of the -e type systems which indeed are relatively massive galaxies , there is no significant difference between the distribution for the sources in any morphological type with respect to the one corresponding to the sample .
this indicates that all galaxies , irrespective of their morphologies , reside in environments with similar densities .
it is worth mentioning however that within the redshift range considered here , the survey area does not contain very dense , cluster - like , environments .
estimated according to eq.[eq : density - sigma5 ] in ( thick solid line ) and sources .
e ( red dashed line ) , s ( green dotted line ) and u ( cyan line ) labels represent the morphology of individual source . each histogram is normalized by its integral . ]
we derive the basic properties of galaxies by fitting their seds which makes use of the data ( . [
sec : data ] ) going from the nuv up to all available herschel bands .
the sed of each galaxy is fitted using magphys ( multi - wavelength analysis of galaxy physical properties ; * ? ? ?
magphys infers the galactic properties by matching the observed sed with a large library of calculated seds .
these templates are constructed by considering the spectral evolution of stellar populations that are born with a @xcite initial mass function ( imf ) in combination with infrared dust spectral libraries as described in @xcite .
the model assumes that the energy from uv - optical radiation emitted by the stellar populations is absorbed by dust and re - radiated in the fir .
it uses also the two - component dust model of @xcite in order to account for the attenuation of starlight by dust .
the model also accounts for the enhanced attenuation of stellar radiation for stars located in star forming regions in comparison to older stars found elsewhere within the galaxy .
as the magphys analysis is based on ab magnitudes , all available photometry ( aperture matched ) has been converted to the ab magnitude system before estimating their associated fluxes in units of jansky ( jy ) .
additional errors have been added to non - fluxes before running magphys to account for the total flux measurements and calibrations between the different surveys .
these include adding 10 per cent of the flux values in quadrature for all optical - nir bands and 20 per cent for the uv bands . for each output parameter , magphys produces a probability density function ( pdf ) , in addition to the median value of each pdf .
the 16th and 84th percentiles of the pdf have been considered as a measure of the uncertainty .
@xcite showed that it is insufficient to identify bad sed fits based on a simple @xmath8 threshold , instead deriving a threshold which depends on the number of bands of photometry available , above which there is @xmath59per cent chance that the photometry is consistent with the magphys model .
sources exceeding this varying threshold are identified as bad fits , and excluded from the subsequent analysis .
we use the h - atlas sed fits over the entire phase 1 area , derived using the same method as in @xcite , with updated pacs coverage and including data from _
wise_. for the purpose of our study , we have focused on a number of galactic parameters that are inferred by fitting the observed seds with magphys .
these are : the galactic stellar mass ( @xmath60 ) , the dust mass ( @xmath61 ) , the star formation rate ( sfr ) , and the fraction of total dust luminosity contributed by the diffuse interstellar medium ( ; 0 @xmath0 @xmath0 1.0 ) .
large values of indicate that dust is heated by the old stellar populations while lower values suggest that ongoing star formation has a more prominent role in heating the dust .
an example of a sed fit for a source in our sample is shown in fig .
[ fig : sample - sed ] .
we find that the distribution of 2 in our sources , does not show any correlation with galaxy colour indices .
it is worth mentioning that the comparison of the results from magphys , with and without the mir constraints from _ wise _ , shows that including the _ wise _ data modifies the output results from magphys .
the inclusion of _ wise _ data improves the fits of the seds and provides better estimates of some of the parameter , and notably of the sfr .
for this reason , we include in the following sections only those galaxies for which _ wise _ data are available ( e.g. @xmath62 of the main sample ) .
this in turn , reduces the size of our sample from 3782 to 2622 sources with 78 having nuv@xmath30 mag and therefore are .
figure [ fig : fraction_of_red_gals ] displays the mass distribution of galaxies in the and samples ( in different categories ) . in our sample ,
@xmath1 73 per cent of sources have stellar masses @xmath63 , while the same number for the galaxies is @xmath64 per cent , accounting for @xmath1 4.2 per cent of the total number of sources with @xmath65 .
as expected , bins associated to largest stellar masses are accupied by the -e galaxies ( see table .
[ table : magphys - vs - hrs ] ) .
it is important to compare the inferred parameters derived from magphys to other determinations .
we compare the estimated dust - to - stellar mass ratio ( @xmath66 ) for all sources as computed by magphys to those derived for a sample of @xmath24300 nearby galaxies from the hrs ( _ herschel _ reference survey ; * ? ? ?
the total dust mass of a given galaxy as estimated by magphys is the sum of the three components which includes the mass contributed by dust in thermal equilibrium in stellar birth clouds , as well as warm and cold dust components in the ambient interstellar medium @xcite .
[ fig : magphys - vs - hrs ] displays the distribution of @xmath66 inferred from magphys for our sample against for all and sources .
overlaid are the @xmath66 estimates from the hrs using all spire bands . for hrs non - detections ( triangles ) , the upper - limit fluxes have been determined assuming a 3@xmath10 signal over a circular aperture of radius 0.3@xmath67 , 0.8@xmath67 and 1.4@xmath67 of the optical radius for the hrs e , s0 and spirals , respectively .
note that in determining dust masses @xmath61 , both magphys and @xcite adopt a dust emissivity index @xmath68 2.0 for cold dust but different dust mass absorption coefficients @xmath69 .
@xcite use a dust mass absorption coefficient @xmath70 of 0.192 m@xmath7kg@xmath17 at 350 whereas @xcite assume @xmath71 0.077 m@xmath7kg@xmath17 at 850 .
given the scaling relations @xmath72 and @xmath73 one finds that @xmath74 in magphys can be scaled down ( assuming @xmath68 2.0 ) to 0.45 m@xmath7kg@xmath17 at 350 and that dust masses as measured by @xcite are @xmath1 2.36 times larger than those estimated by magphys .
thus in fig .
[ fig : magphys - vs - hrs ] , the hrs sample are scaled down for @xmath1 0.37 dex to account for the differences between the two measurements of dust masses .
it can be seen that the @xmath66 ratios for both the or galaxies agrees reasonably well with estimates from the hrs detected objects .
furthermore , the sources of type e exhibit , on average , @xmath66 ratios that are noticeably lower than those of galaxies .
this is even more clear in the right panel of fig.[fig : histo - mdust ] which displays the distributions @xmath66 in all sources .
the mean values as summarized in table .
[ table : magphys - vs - hrs ] suggest that the -e objects have values of the dust - to - stellar masses that are approximately an order of magnitude lower than those in the sources .
this is partly because the -es have high stellar masses but as is visible in the left panel of fig .
[ fig : histo - mdust ] , they also have a lower dust content in comparison to the -s and systems .
note that the distribution of specific dust mass of the -s galaxies does not match the distribution of the star forming galaxies .
we will discuss this further in sec .
[ sec : magphys - sfr ] . | we combine herschel / spire sub - millimeter ( submm ) observations with existing multi - wavelength data to investigate the characteristics of low redshift , optically red galaxies detected in bands .
we select a sample of galaxies in the redshift range 0.01 @xmath0 z @xmath0 0.2 , having sigma detections in the spire 250 waveband .
sources are then divided into two sub - samples of and galaxies , based on their uv - optical colours .
galaxies in the sample account for @xmath1 4.2 per cent of the total number of sources with stellar masses m@xmath2 10@xmath3m@xmath4 .
following visual classification of the galaxies , we find that @xmath5 30 per cent of them are early - type galaxies and @xmath5 40 per cent are spirals .
the colour of the -spiral galaxies could be the result of their highly inclined orientation and/or a strong contribution of the old stellar population .
it is found that irrespective of their morphological types , and sources occupy environments with more or less similar densities ( i.e. , the @xmath6 parameter ) . from the analysis of the spectral energy distributions ( seds ) of galaxies in our samples based on magphys
, we find that galaxies in the sample ( of any morphological type ) have dust masses similar to those in the sample ( i.e. normal spiral / star - forming systems ) .
however , in comparison to the _ red_-spirals and in particular systems , _
red_-ellipticals have lower mean dust - to - stellar mass ratios . besides galaxies in the _
red_-elliptical sample have much lower mean star - formation / specific - star - formation rates in contrast to their counterparts in the sample .
our results support a scenario where dust in early - type systems is likely to be of an external origin .
[ firstpage ] 2 deg@xmath7 2 @xmath8 sigma @xmath95@xmath10 galaxy |
both because of its proximity and its unique nature , the large magellanic cloud is an excellent laboratory for studying the effects of a variety of processes thought to affect the formation of stars , e.g. interactions with neighboring galaxies , the influence of bar dynamics , stellar feedback , and metallicity of the environment .
thus , knowledge of the star formation history of the lmc is a step towards understanding the physical origin of stellar populations .
this star formation history is recorded in the lmc s numerous clusters and field stars .
because clusters are relatively easy to study , measuring their age and metallicity distribution is a natural way to study the star formation history of the lmc .
researchers have shown that the lmc s clusters are predominantly younger than 4 gyr in age , with only a small number of old ( @xmath214 gyr ) globular clusters and one cluster , eso 121+sc03 , with an age of @xmath28 gyr ( mateo , hodge , & schommer 1986 ) . while the cluster results suggest that the lmc has formed the bulk of its stars over the past 4 gyr , recent observations of lmc field stars taken with the _ hubble space telescope _ imply that stars formed even during the lull in cluster formation , assuming a standard salpeter imf ( holtzman et al .
1997 , hereafter h97 , geha et al .
these provocative results open a number of questions , such as what might cause a switch between cluster and field star formation modes , and indicate that field stars need to be accounted for in a complete description of the lmc s star formation history .
moreover , as a complete record of
star formation is available in a single pointing and field stars have an almost continuous age distribution , field observations appear ideally suited for identifying major star formation events and for mapping spatial variations in the star formation history , both keys for understanding the physical origin of stellar populations . the clear disadvantage of using field stars rather than clusters to study the star formation history , however , is that the information they hold is difficult to extract .
this difficulty was recognized by schwarzschild ( 1958 ) , who wrote : `` the random sample represented by the nearby stars certainly includes a large range of masses , possibly some variety in initial composition , and most likely a great mixture of stellar ages . a great variety of evolutionary tracks therefore is involved ... and their disentanglement seems somewhat hopeless . ''
luckily , modern tools now allow astronomers to attack the problem with brute force on a scale that schwarzschild could not have done .
observational facilities such as _ hst _ and a variety of excellent ground - based telescopes gather data efficiently and accurately , while the stellar evolutionary models used in comparison with the color - magnitude diagrams ( cmds ) have substantially improved ( e.g. schaller et al .
1992 , bertelli et al .
1994 , hereafter b94 , chaboyer & kim 1995 , vandenberg 1998 ) . aided by a number of techniques that take advantage of easily available fast computing facilities ,
work on the star formation histories of the magellanic clouds and of other local group galaxies is proceeding at a rapid pace ( e.g. gallart et al .
1996 , aparicio , gallart , & bertelli 1997 , dohm - palmer et al .
1998 , tolstoy et al . 1998 ) , often with dramatic results ( e.g. smecker - hane 1997 ) .
our present work seeks to add to the growing body of knowledge of the star formation history of the lmc through observations of six fields taken with _
these observations were taken with the primary purpose of producing cmds of the lmc s inner old globular clusters ( olsen et al .
1998 , hereafter paper 1 ) .
however , the images also contain significant numbers of field stars ; viewed as contaminants in paper 1 , they are the focus of this paper . the locations of the fields are shown in fig . [ targets ] ; five of them lie in the bar while the other lies in the disk .
as the fields are well - separated , the observations are nicely situated for exploring spatial variations in the star formation history . in particular
, we will address the question of whether the bar and disk have had different star formation histories , as would be expected if the bar induces star formation or if the event that formed the bar also triggered a significant star forming episode . while the majority of published lmc field star studies have avoided the bar because of the severe effects of photometric crowding , the resolution afforded by _
hst _ enables us to produce cmds which include stars well below the oldest main sequence turn - off even in the bar . in the following
we briefly describe the observations and reductions in section 2 , referring the bulk of this discussion to paper 1 .
section 3 describes the method used to extract the star formation histories .
the method , which is based on the work of dolphin ( 1997 ) , determines the star formation history by fitting a linear combination of model isochrones to the histogram of points in the cmd ( the hess diagram " ) , accounting for observational errors .
tests of the method using simulated cmds are described in section 4 .
section 5 discusses the application of the method to the observed cmds and evaluates its success .
we present our conclusions in section 6 .
as described in paper 1 , observations of the six lmc globular clusters ngc 1754 , ngc 1835 , ngc 1898 , ngc 1916 , ngc 2005 , and ngc 2019 were taken during cycle 5 of _ hst _ using wfpc2 and the f555w and f814w filters . by taking both long and short exposures through each filter , we were ensured reliable photometry of the brightest stars in the fields while still reaching @xmath8 .
we centered the clusters on the planetary camera , as shown in the sample image ( fig .
[ image ] ) , leaving the wide field ( wf ) chips containing primarily field stars .
paper 1 fully describes the image reductions , photometry , and production of cmds . in summary , cosmic ray rejection ,
correction for the @xmath9-dependent cte effect ( holtzman et al .
1995 ) , geometric distortion correction , and bad pixel flagging were applied to the images before performing photometry . for the photometry , we used version 2.5 of the profile - fitting photometric package dophot ( schechter , mateo , & saha 1993 ) , modified by eric deutsch to handle floating - point images . for the wf chips ,
aperture corrections were applied as a function of position in the frame using measurements of isolated bright , unsaturated stars taken in the profile - subtracted images . in agreement with the report of whitmore & heyer ( 1997 ) , we found an offset in mean magnitudes between the short- and long - exposure photometry .
we corrected for this effect by adjusting the short - exposure magnitudes to match , on average , those of the long exposures .
finally , we merged the short- and long - exposure lists of photometry , transformed the magnitudes from the wfpc2 system to johnson @xmath3/kron - cousins @xmath10 , and produced cmds by matching stars with measurements in both @xmath3 and @xmath10 by position .
table [ wfphot]a contains a sample portion of the photometry , the full version of which is published in the electronic version of the _ astronomical journal_. fig .
[ rawcmd ] shows the cmds obtained from the combined wf frames of each field .
the main features of the cmds are a lower main sequence ( ms ) seen to @xmath11 , a broad upper ms , a prominent red clump ( rc ) containing intermediate - age horizontal branch ( hb ) stars , and an old red giant branch ( rgb ) .
the stars falling to the blue of the main sequence envelope at @xmath12 in the ngc 1835 , ngc 2005 , and ngc 2019 fields could be blue hb stars ; however , most old , metal - poor hb stars will be difficult to distinguish from the broad upper main sequence . while the old red giant branch in the ngc 1754 field forms a tight sequence , in the other fields ( excluding ngc 1916 ) the giant and subgiant branches appear broader , suggesting a range of turnoff ages .
the ngc 1916 field cmd looks remarkably different from the others , having a much broader lower ms and rgb and a stretched - out rc . as was discussed in paper 1
, this appearance is most likely the result of strong differential reddening , clear evidence for which is seen in ground - based images of the region ( a. walker , personal communication ) . while we can not rule out a contribution from significant depth in the field stars local to ngc 1916 , the simplest hypothesis is that differential reddening entirely accounts for the anomalous appearance of the ngc 1916 field cmd .
some of the old rgb and lower ms stars are expected to be globular cluster stars spilling over into the wf chips . from the estimated core and tidal radii of the clusters ( paper 1 )
, we predict the numbers of contaminants listed in table [ contam ] . in paper 1
, we described using these estimates to iteratively subtract field stars from the cluster cmds and vice - versa .
[ subcmd ] shows the field star cmds after the cluster stars have been removed .
the ngc 1754 and ngc 1898
field cmds have been altered significantly , with the ngc 1754 field cmd losing most of its rgb , while the others appear relatively unchanged .
the subtraction was not perfect , as some regions of the cmds have a choppy appearance ; it appears that the large number of contaminants present in some of the fields brings out the faults of the subtraction technique .
we did not find this problem when subtracting field stars from the cluster cmds , as there are fewer contaminating field stars in the pc .
moreover , the estimated number of contaminants in the fields depends strongly on the somewhat uncertain tidal radii of the king model fits to the cluster profiles . for these reasons , we decided it was easier to model the field star cmds without correcting for contamination from the clusters ; all of the star formation histories we derive thus include a contribution from the globular clusters .
the study of the star formation history of the lmc using field star cmds has a history dating to more than 20 years ago , the biggest developments of which are summarized by gallagher et al .
the earliest analyses were conducted by visual examination of the main features of the cmds ( the upper main sequence , rgb , and asymptotic giant branch ) and by comparing the cmds to fiducial sequences of clusters of known age and metallicity .
bertelli et al .
( 1992 ) developed a more sophisticated technique which involved solving for parameters governing the star formation history of their lmc fields .
the solutions were made by comparing the average density of stars in broad regions of the observed cmds with predictions from stellar evolutionary models .
similar techniques have been developed and refined by gallart et al .
( 1996 ) , among others , and used to study the star formation histories of local group dwarf galaxies .
recently , h97 and geha et al . (
1998 ) used the luminosity functions of three lmc fields to fit several model star formation histories .
a common feature of the methods described in the previous paragraph is that they do not use all of the information available in the field star cmds ; they use either selected portions of the cmds or luminosity functions to extract the star formation histories . by disregarding the full distribution of colors and magnitudes , those approaches avoid the problems of modelling the uncertain later phases of stellar evolution , but at a cost in age resolution .
higher resolution is desirable for addressing questions such as the possible effect of interactions on the lmc star formation history and for exploring differences among cmds taken in different fields .
the theoretical framework for a more detailed comparison of models and observations has been assembled by a number of researchers ( e.g. tolstoy & saha 1996 , dolphin 1997 , ng 1998 ) , but has been mostly untested in full capacity against observations . as our cmds are well - populated by stars in most of the phases of evolution seen in the models , and we wish to address the question of whether there are differences among the star formation histories they represent , our cmds provide good motivation and a favorable environment for exploring the potential of a detailed comparison with the models . to first approximation , the field star cmds are a linear combination of isochrones of several ages and metallicities governed by a particular star formation rate as a function of time ( sfr(@xmath5 ) ) , a universal initial mass function ( imf ) , an age - metallicity relation ( z(@xmath5 ) ) , a single value of the reddening ( @xmath13 ) , and a single distance modulus ( @xmath14 ) . additional factors that may affect the appearance of the cmds are the presence of binary stars , the spread in z at each age @xmath5 , variation of the imf , and the spread in @xmath14 and @xmath13 within the fields
. for this work we ignore these additional factors ; the first three are disregarded for the sake of simplicity .
we consider the assumption of uniform distance to be a safe one , as the lmc is known to be thin compared to its distance ( caldwell & coulson 1986 ) .
also , from the tight appearance of the cluster sequences in 5 of the 6 fields ( paper 1 ) , the amount of differential reddening must be small .
for the ngc 1916 field , however , differential reddening is obviously strong ; we thus exclude this field from the current analysis . given these assumptions , the problem of determining the star formation history
is reduced to constructing a model cmd whose sfr(@xmath5 ) best reproduces the distribution of stars in the observed cmd for a given z(@xmath5 ) , imf , @xmath13 , and @xmath14 .
one approach to this problem is to construct model cmds from a small number of template star formation histories and to compare them with the observed cmds ( e.g. gallart et al .
1996 , aparicio et al .
however , in order to take full advantage of the excellent photometry of lmc stars possible with @xmath15 , the number of templates we would need to construct would be exceedingly large .
for example , if we wished to extract the star formation history using 10 age bins spanning the range of 10 myr 15 gyr with 5 possible values for the sfr within each bin , and we wished to test a small number of different imf slopes , reddenings , and distance moduli , we would need to construct @xmath16 model star formation histories .
as each model cmd needs to contain at least as many individual stars as the observed cmd , our available computing resources would be greatly overmatched .
a more convenient approach is to treat sfr(@xmath5 ) as a free parameter for which we find the best solution for a given set of input parameters . with this approach ,
we construct model cmds representing a constant sfr in each age bin of the desired solution , constrained by the input z(@xmath5 ) relation , imf , reddening , and distance modulus . by
solving for the coefficients of the linear combination of models that best fits the distribution of stars of the cmd , we recover sfr(@xmath5 ) .
the number of model star formation histories we need to construct is then vastly decreased , and is simply equal to the number of combinations of z(@xmath5 ) relations , imf slopes , reddenings , and distance moduli we wish to test times the number of age bins in the solution .
an additional advantage of this approach is that a fit parameter can be used to test the goodness - of - fit of the models and errors can easily be calculated from monte carlo simulations .
an implementation of such an approach was first discussed by dolphin ( 1997 ) , who used @xmath17 as the fit parameter and singular value decomposition ( press et al .
1992 ) to solve for the best - fit star formation history from a suite of models with a number of different imfs , reddenings , distance moduli , and abundances .
dolphin applied the method to simulated cmds to test the accuracy with which the input parameters could be recovered , but did not apply the method to actual observations .
we used the principles of this automated approach to construct a method to extract the star formation histories from our field star cmds .
the method consists of four basic steps , which we describe in the following subsections .
the b94 isochrones are well - suited for our purposes because they model every post - main sequence evolutionary phase , incorporate the recent opal opacities ( rogers & iglesias 1992 ) , and are calculated for @xmath3 and @xmath10 , our observed passbands .
their main drawback is that they are difficult to interpolate , as they are not tabulated according to common evolutionary points . to properly model the cmds , we needed to interpolate the isochrones in metallicity as well as age .
we performed this interpolation by breaking the isochrones into segments based on the characteristic evolutionary phases listed in tables 7@xmath712 of b94 , an electronic copy of which was kindly provided by g. bertelli .
each of these segments was then broken into a fixed number of mass bins .
a new isochrone could then be interpolated point - by - point from adjacent isochrones , with the segments ensuring that the isochrone shape was preserved .
we encountered a few problems , however .
a problem of a minor nature was that we discovered that the @xmath4 and @xmath3 for the masses of the evolutionary phases listed in b94 s tables 7@xmath712 were often inconsistent with those tabulated in the isochrones .
instead of using @xmath4 and @xmath3 from tables 7@xmath712 , we therefore used the @xmath4 and @xmath3 from the isochrones for the points nearest the masses in the tables .
another small difficulty is that the number of evolutionary points listed in b94 s tables 7@xmath712 occasionally changes for isochrones of different ages and different metallicities . for isochrones affected by these changes
, we used our best judgement in deciding the correct mapping .
a more serious problem is that large changes in the curvature of the isochrones do not always correspond with evolutionary points listed in the tables .
[ probint ] illustrates this problem .
the heavy lines are isochrones with log ages of 8.0 and 8.1 years and [ fe / h]@xmath2 - 0.4 , while the light line is an interpolated isochrone with log age of 8.05 . while the interpolated isochrone faithfully reproduces the shape of the adjacent ones over most of the isochrone , even on the blue loop , it deviates on the asymptotic giant branch .
this is because there is no evolutionary point in table 10 of b94 designating the finish of the blue loop and the start of the agb . while we found that the interpolated isochrones have a number of areas where they deviate from the correct shape , the deviations occur in short - lived phases of stellar evolution .
we therefore did not attempt to correct the problems , as few of the stars in our observed cmds will occupy these phases .
we produced model cmds using reddenings of @[email protected] with steps of 0.02 , distance moduli of [email protected] with steps of 0.1 , and imf slopes of x=1.5 , 2.0 , 2.35 , and 3.0 , where the imf is given by the proportionality @xmath18 .
the combinations of parameters are summarized in table [ mpars ] .
the ranges of @xmath13 were chosen to span the likely range for these fields , the distance moduli include both the `` short '' and `` long '' distance scales of the lmc , and the imfs were picked to span the range of a number of derivations of the imf in the lmc . the age - metallicity relation adopted is linear in age with a slope of -0.073 dex gyr@xmath19 and a zero point of -0.3 dex .
this model , shown in fig .
[ chemevol ] , agrees with the age - metallicity relation suggested by lmc clusters ( olszewski et al .
1996 ) , with ours having a slightly higher metallicity at older ages that reflects our readjustment of the abundances of the inner globular clusters ( paper 1 ) over olszewski et al .
however , because of the large gap in the lmc cluster age distribution , our adopted age - metallicity relation must be regarded as very uncertain over the @xmath25@xmath710 gyr range .
moreover , it is possible that the chemical evolution of the clusters and field stars is to some degree decoupled .
as discussed by gallagher et al .
( 1996 ) in the context of the lmc , the degenerate effect of age and metallicity on stars in the cmd makes the distribution of metallicities a critical assumption , particularly for older stars .
lack of knowledge about the field star metallicities thus necessarily weakens our final conclusions .
we constructed the model cmds using 36 logarithmic age bins .
the bin widths , set to @xmath20.08 in the log , and number of bins were chosen to approximately match the resolution with which we could hope to extract the star formation history from the observed cmds .
we chose to use logarithmic age bins to compensate for the fact that isochrones are spaced more closely together at larger ages , decreasing the resolution with which one can discriminate between populations of different ages . within each age bin , we selected 20000@xmath730000 stars with randomly distributed ages and masses , with limits on the ages imposed by the bin edges and limits on the masses set approximately equal to the maximum and minimum masses of stars on the isochrones within the age bin .
the number of stars was chosen to greatly exceed the number expected in the observed cmds so as to limit counting statistics in the models , yet be smaller than the number of artificial stars available for simulating the errors . in order to increase the sampling of stars with higher masses , we selected the masses from a flat imf ( x=0.0 ) but assigned each star a weight as if it were selected from the desired imf of the model . after all stars were selected , we calculated the total weighted mass in stars within each age bin , adding to this total the integral of the imf over the range 0.08 @xmath20 120 m@xmath21 but excluding the mass range occupied by the simulated stars . the total mass represented in the age bin
was then used to calculate the star formation rate within each age bin . for each model star , we linearly interpolated @xmath4 and @xmath3 from the nearest isochrones of appropriate age and metallicity .
photometric errors were applied to these stars through the use of the artificial star tests described in paper 1 . for each star in the simulated cmd
, we randomly selected a nearby artificial star .
if the artificial star was not recovered during photometry , we removed the star from the simulated cmd .
otherwise , the star was moved according to the photometric shifts experienced by the artificial star . to increase the number of artificial stars from which we could simulate the errors
, we allowed the program to pick artificial stars of a restricted range in magnitude but of any color .
because stars experience systematic photometric shifts that depend on color , we corrected for the different systematic shifts before moving the model stars .
we stored the simulated cmds as hess diagrams with bins of 0.0625 in @xmath4 and 0.2 in @xmath3 . at this stage ,
the weights to bring the cmds into keeping with the desired imf were applied and the histograms were scaled so that each age bin represents a star formation rate of 1 m@xmath21 yr@xmath19 . figs .
[ models]a - d show a sample raw simulated cmd , the simulated cmd with photometric errors , and the binned cmd , all summed over the 36 age bins . for deriving sfr(@xmath5 ) and identifying the best - fit model ,
the @xmath17 parameter tested by dolphin ( 1997 ) is not ideal , as @xmath17 is highly sensitive to outlying points .
we instead chose to minimize the following parameter , which is derived from the lorentzian distribution ( press et al .
1992 ) : @xmath22 where @xmath23 are the observed number of stars in the cmd at point @xmath24 in the @xmath25 grid , @xmath26 is the number of stars predicted at grid point @xmath24 by the model with parameters @xmath27 , and @xmath28 is the uncertainty in the number of stars at @xmath24 , which we calculated from the models assuming poisson statistics .
the lorentzian , with its large tails , is much less sensitive to outliers than @xmath17 .
we expect to find outliers when comparing our observations to the models for several reasons .
a primary reason is that the model isochrones used to generate the model cmds do not perfectly describe the process of stellar evolution , especially for the evolved phases .
in addition , we are assuming that a single reddening , distance modulus , and imf is appropriate for all of the stars in a single cmd , and that all stars of a single age have the same metallicity , while in reality there will be some spread in these quantities .
finally , it is unlikely that we have exactly reproduced the photometric errors through our artificial star tests , and there could be significant tails in the photometric distributions that we have not modeled correctly .
however , as long as the bulk of the stars are described by our single parameters and our models of the photometric errors apply to most of the stars , the remaining stars may be treated as outliers . for the ngc 1916 field ,
the assumption of a single reddening is clearly strongly violated , so that differential reddening must be included for a viable model fit .
for each model with different @xmath13 , distance modulus , and imf we can solve for sfr(@xmath5 ) by minimizing equation 1 over all of the model age bins .
because our models are normalized to represent a star formation rate of 1 m@xmath21 yr@xmath19 , the best - fit linear combination will produce sfr(@xmath5 ) in units of m@xmath21 yr@xmath19 over the area of the field .
a partial code for producing the solutions was kindly provided by a. dolphin , which we rewrote and amended for our purposes .
the program minimizes @xmath29 using the _ numerical recipes _
routine amoeba ( press et al .
1992 ) . given a starting guess at the correct values of the fitted parameters and the likely scale of the parameters
, amoeba searches the @xmath30-dimensional parameter space for a minimum in the surface , stopping when small changes in the parameters change the value of the function by less than a given amount , which we set to be slightly larger than the machine precision .
because amoeba stops when a minimum is found , the minimum may not be the desired global minimum .
to more thoroughly search the @xmath29 surface for a global minimum , amoeba was restarted several times with the new input guess and scale of the parameters set equal to the previous output .
after 4 - 5 restarts , amoeba returned to the same minimum as the previous run , indicating that no other nearby minimum exists . in this fashion , we established solutions of sfr(@xmath5 ) for each model with different @xmath13 , distance modulus , and imf . for each imf , the best - fit @xmath13 and distance modulus
were decided by picking the model with the lowest value of @xmath29 .
dolphin ( 1997 ) has shown , using a method very similar to the one described in this work , that it is possible to faithfully extract the star formation history from a simulated cmd representing a typical local group dwarf galaxy .
an outstanding question , however , is the age resolution with which the star formation history may be solved .
we explored this issue , as well as the ability of our method to solve for sfr(@xmath5 ) and identify the correct @xmath13 and @xmath14 , by producing a cmd representing a simple star formation history and running it through the solution procedure .
the test case was constructed by taking a model with a uniform star formation rate and adjusting the star formation rate in each age bin by a randomly selected amount . to avoid using pre - existing knowledge of the star formation history to our advantage during the solution ,
the input star formation history was kept secret until after the solutions were made .
choosing @xmath13 = 0.08 , @xmath14 = 18.5 , and x = 2.35 for the imf slope , we simulated a cmd for the test case by selecting 1000 stars in each age bin from the interpolated b94 isochrones , to which we applied the completeness fractions and photometric uncertainties derived for the ngc 1754 wf frames , using the artificial star tests of paper 1 .
we binned this simulated cmd to the resolution of the models .
we also smoothed both the test cmd and the model cmds with a gaussian having a dispersion of one bin width , as we found that this limited the stochastic effects imposed by the binning .
[ sfrin ] shows the input star formation history for the test case . making a starting guess at the star formation rate in each bin
, we used 4 iterations of amoeba to find the best - fit coefficients of the models in each of the 36 age bins and each value of @xmath13 in the range [email protected] , @xmath14 in the range 18.4 - 18.7 , and x = 2.35 .
[ fitsurf ] shows the computed @xmath29 surface as a function of @xmath13 and @xmath14 .
the minimum of the surface occurs at @xmath13 = 0.08 and @xmath14 = 18.5 , in agreement with the input values .
[ tests]a shows the star formation history derived for the minimum @xmath29 , compared with the input .
the output contains wild disagreements with the input star formation history , mimicking a `` bursting''-type star formation history .
guessing that the problem is one of age resolution , we redid the solutions after grouping several age bins of the models together .
[ tests]b - d show the best - fit solutions to the star formation history after grouping every 2 , 3 , and 4 models together .
these solutions reproduce much better the input model ; the disagreements are all within the uncertainties calculated from repeated test runs using resampled cmds , which we describe below .
two factors contribute to the limiting age resolution of the star formation history solutions : the number of stars that fall within each age bin and the spacing of the models in the cmd .
the first factor is a function mainly of the total number of stars in the observed cmd while the second is a function of the age of the models and the photometric errors .
grouping the models together has the effect of reducing the contribution of both factors ; the number of stars within each age bin is increased while the spacing between the models also increases . to describe the effects that the number of stars and the model spacing have on the errors in the solution , we ran repeated tests using the model star formation history described above with varying numbers of stars in the simulated cmds .
table [ tpars ] lists the parameters and number of runs of the tests . for each set of runs
, we calculated the dispersion in the recovered minus input star formation rates separately for each age bin of the solution .
the top panel of fig .
[ resolution ] shows the dispersion in ( @xmath31 ) , taken as a median over all age bins in the solution , plotted against the number of stars per age bin in the cmds used in each set of tests .
the figure demonstrates that the median uncertainty in the star formation rates increases exponentially with decreasing numbers of stars per age bin , as is expected from poisson statistics .
the lower panel of fig .
[ resolution ] shows that the dispersion about the median over the age bins , indicated by the error bars in the top panel , is due to the effect of isochrone spacing . at the youngest and oldest ages ,
the uncertainty in the recovered star formation rate is higher than for intermediate ages , where model isochrones of similar ages are more easily distinguishable .
we will bear these effects in mind when discussing the age resolution of the star formation histories derived for the observed cmds .
the sfr(@xmath5 ) solutions for the lmc field star cmds were made following the procedure outlined in section 4 . as was done for the tests described in section 5 , we binned the observed cmds to the same resolution as the models and smoothed both the observed and model color - magnitude grids with a gaussian having a 1-pixel dispersion .
we produced separate solutions for each combination of the parameters listed in table [ mpars ] and for 18 , 12 , and 9 age bins . for ngc 1835 and ngc 2005 , we found that we needed to add additional values of the reddening and distance modulus to bracket the full range of best - fit solutions . for each imf
, we found the reddening and distance modulus that best match the observed cmd by choosing the minimum of the @xmath29 surface , as was done for the test .
a sample surface is shown in fig .
[ n1754fitsurf ] . while the minimum @xmath29 is well - defined , its value is not as low as those found during the tests , a sign that the model isochrones as input do not agree perfectly with the observations .
we calculated the errors in sfr(@xmath5 ) , @xmath13 , and @xmath14 through monte carlo simulations using bootstrapped samples of the observed cmds .
we ran 25 simulations for each value of the imf ; with each run we resampled the observed cmds and produced solutions of sfr(@xmath5 ) for each value of @xmath13 and @xmath14 , exactly as was done for the analysis of the true observed cmds .
again , we chose the best - fit solution from the set by picking the minimum of the @xmath29 surface . figs .
[ sfh]a - c show the derived star formation histories and associated errors using 18 , 12 , and 9 age bins in the solution and the salpeter imf , while table [ ebv_dm ] lists the best - fit values of @xmath13 and @xmath14 for these solutions .
there are several common features in the derived star formation histories .
all of the fields show significant star formation in the last 3 - 4 gyr with similar amplitude , with the ngc 1754 field , on account of its lower star density , proceeding at an overall rate @xmath22@xmath73 times lower than the rest .
the star formation has the appearance of a burst , peaking at @xmath21 gyr , in all of the fields .
while little can be said about the very young stars because of the small size of the fields and the close spacing of the young isochrones , there appears to be a decline in sfr(@xmath5 ) over the last @xmath20.5@xmath71 gyr .
this apparent recent decline could simply be evidence that the stars formed in the past 1 gyr are not thoroughly mixed ; the crossing time in the bar is @xmath280 myr , implying a mixing time of a few hundred million years .
all of these features appear independent of the number of age bins used in the solution .
the ngc 1754 field star formation history appears to agree with that derived by h97 for a field near ngc 1866 ; both show an increase in the recent ( @xmath324 gyr ) star formation rate by a factor of @xmath23 , although the exact number depends on the age - resolution of the solution . however , we find that the star formation did not instantly increase , but did so steadily .
we do not find evidence for the short burst of star formation occurring @xmath22 gyr ago that was found by gallagher et al .
( 1996 ) . while the ngc 1754 field seems to
only recently have formed a large number of stars , star formation in the bar fields was active as long as @xmath25@xmath78 gyr ago .
the existence of this older population in the bar fields naturally explains the differences in rgb and sgb morphology seen in the cmds , as the bar fields clearly have broader rgbs and sgbs than the ngc 1754 field .
the older population is most clearly seen in the solution containing 9 age bins , but remains in the solutions with higher age resolution .
this result implies that there is a significant difference in the star formation histories between the inner and outer regions of the lmc .
because a bar does not confine any given star to the bar pattern , we of course can not directly age - date the lmc bar through its stellar populations .
however , the older population of stars that we see within the bar could be readily explained by a bar instability triggering star formation as early as 5@xmath78 gyr ago .
this interpretation disagrees with that of elson , gilmore , & santiago ( 1997 ) who have suggested , on the basis of a wfpc2 cmd of a field in the bar , that the bar formed @xmath21 gyr ago , or 1@xmath72 gyr after the lmc disk .
our results agree with elson et al . in
that we do find a peak in the star formation rate occurring @xmath21 gyr ago in the bar fields ; however , we also see this peak in the disk field .
our interpretation also disagrees with the suggestion of ardeberg et al .
( 1997 ) that the bar may be younger than 0.5 gyr . again noting that 0.5 gyr is of order the mixing time in the bar
, it is quite possible that the large young population seen by ardeberg et al . in their field
represents a local star - forming event .
our results also do not fully agree with those of hardy et al .
( 1984 ) , who found that no significant star formation has occurred in the bar before 3 gyr ago
. however , hardy et al . may not have been able to detect an older population , as their cmd was not deep enough to reach the to and lower ms
. an encouraging feature of each of the star formation histories of figs .
[ sfh]a - c is a significant number of stars in the last age bin , where we expect the contaminating globular cluster stars to be found .
how much of the old stellar populations can be attributed to the clusters spilling into the wf frames ? while we found we could not cleanly statistically subtract the cluster stars from the field star cmds , we can use the estimates of their numbers derived previously to address the question . in table
[ contam ] we compare the number of cluster stars in the wf frames estimated from the king model fits to the number of stars in the last age bin of the star formation histories of fig .
we find that the cluster stars contribute significantly to each field , while in the ngc 1754 and ngc 1898 fields they dominate the oldest bin .
this finding leads us to suppose that the narrow rgb of the ngc 1754 field is produced almost entirely by cluster contamination , while the broad rgbs and sgbs of the bar fields truly represent an old field population .
so far , we have implicitly assumed that the best - fit model star formation histories provide good fits to the observations .
however , our models make a number of simplifications over what is realistically expected ; e.g. we assumed uniform reddening , no binary stars , and a simple linear age - metallicity relation .
to get an idea of whether the derived star formation histories provide sensible matches to the data , we overlaid isochrones representing peaks in the derived star formation histories on the field star cmds .
[ isochrones ] shows the ages at which the isochrones were chosen and the corresponding matches with the cmds .
the inability to tell by eye whether the chosen peaks in the star formation rate correspond to increases in the stellar density in the cmd reinforce the need for an automated technique to derive the star formation histories .
we can tell , however , that there is a clear difference between the ngc 1754 cmd and the cmds of the bar fields . while the ngc 1754 field appears to have a single old turnoff , the bar fields have many stars @xmath20.5 magnitude brighter than the oldest turnoff , which we interpret as the 4 - 8 gyr population in figs .
[ sfh]a - c . overlaying the isochrones also shows that our adopted age - metallicity relation does not provide an ideal match to the cmds .
the most striking disagreement is that the red giant branches of the older isochrones fall too far to the red to match the observed giant branch .
while the observed giant branch could be fit by older isochrones if the distance modulus were decreased , it would then be difficult to match the blue envelope of the upper main sequence , which is reproduced well with the current models .
a possible explanation is that the metallicity of the model old populations is too metal rich . because the metallicity of the oldest age bin of the model was chosen to equal the average abundance of the old globular clusters ( paper 1 )
, this would imply that the field contains an old population that is considerably more metal - poor than the clusters .
however , because the globular cluster sequences fall quite closely on top of the field star rgbs , this would mean that the measured globular cluster abundances are too metal - rich by @xmath20.5 dex .
as the globular cluster abundances were measured from the slopes of the rgbs , which are empirically calibrated with milky way clusters using high - dispersion spectroscopic abundances , we would be surprised if they were found to be so grossly incorrect .
another possibility , however , is that the b94 isochrones do not correctly predict the color of the red giant branch for low metallicities ; our comparison suggests that the model rgbs are too red by @xmath20.08 magnitudes , given that we have solved for the correct reddening and distance modulus of the field stars . in fig .
[ residuals ] we show the residuals produced by the models of fig . [ sfh]a .
the residuals are shown as fractions of the observed star density in the cmds .
the greyscale stretches from -1 to 1 ; white pixels are regions where too many stars are predicted by the models , black pixels represent an excess of observed stars , and grey pixels represent the regions where the models closely match the observations .
the residuals show that over large areas of the cmds , the models reproduce the observations fairly well .
however , as expected from the comparisons with isochrones , there are several areas where the fits are poor , most notably the area of the red giant branch , the horizontal branch and red clump stars , and a small section of the upper main sequence .
these problems are not as apparent if the model and observed luminosity functions are compared instead of the cmds .
these luminosity functions are shown in fig .
[ lf ] ; the errors in the points include both poisson counting statistics and the uncertainties in the derived star formation rates of the models .
the use of luminosity functions smooths over many of the disagreements between the models and observations ; application of the @xmath17 statistic shows that the models are not strongly excluded , except for the case of the ngc 1835 field .
the star formation histories derived in section 5.1 may be incorrect in detail because of the inconsistent match of the models and observations . to explore the effects that the systematic disagreements between the models and observations might have on the star formation histories
, we set the reddening and distance modulus to extreme values and examined the resulting star formation histories .
[ extremes]a shows the star formation history of the ngc 1835 field for the case of 18 age bins in the solution after setting @xmath13 and @xmath14 to extremes . in general , decreasing either @xmath13 or @xmath14 increases the relative contribution of the old stellar populations , while increasing either one has the opposite effect .
this can be understood by fig .
[ extremes]b , where we plot isochrones having the extreme values of @xmath13 and @xmath14 over the cmd . for the low values of @xmath13 and @xmath14
, the old isochrones cover the lower main sequence and rgb while the young isochrones do not fit the main sequence as well .
for the high values , the old isochrones do not fit the turnoff and rgb , but the young isochrones follow the main sequence envelope .
we note that decreasing @xmath33 , the slope of the imf , has the same effect on the star formation history as decreasing @xmath13 or @xmath14 ( see fig .
[ imfvar ] ) .
this is because with a shallower model imf , more evolved stars are needed to reproduce the observed luminosity function ( see also discussion in h97 ) .
thus , we might expect that the effect of the model disagreements would be to suppress the old populations in the derived star formation histories .
because the disagreements are largest for the older stars , we should be able to quantify this expectation by using our method to solve for the star formation history of one of the globular cluster cmds shown in fig .
12 of paper 1 . by using a cluster cmd with field stars
statistically removed , we expect to recover a star formation history dominated almost entirely by the oldest bin . in the left panel of fig .
[ pcisoc ] , we show the star formation history derived for the ngc 2019 field star - subtracted cluster cmd . we have adopted a reddening of @xmath13=0.06 for the cluster , as derived in paper 1 , and a distance modulus of 18.5 . indeed , the star formation history is characterized by a large peak in the oldest age bin with almost all of the remaining bins consistent with a star formation rate of zero within the errors . in the right panel of fig .
[ pcisoc ] , we show the model isochrone having age and metallicity equal to the average of the oldest bin of the solution .
while the isochrone fits the lower main sequence and turnoff luminosity , the model rgb falls too far to the red , in a fashion similar to the fits shown in figs .
[ isochrones ] , and the hb morphology is not reproduced .
we find that the integral of the model luminosity function over the range 17 @xmath34 23 is @xmath250% lower than that of the observed lf .
this result agrees with our expectation that the star formation rates in the oldest age bins of figs .
[ sfh]a@xmath7c have been underestimated due to the model disagreements . furthermore , because we do not see significant star formation in the age range 4@xmath78 gyr in fig .
[ pcisoc ] , it appears unlikely that the model disagreements could be the cause of the different star formation histories derived for the ngc 1754 and bar fields .
forty years ago , martin schwarzschild remarked on the hopelessness of disentangling a color - magnitude diagram of stars that contains a wide mix of ages and compositions .
modern computational power , aided by improved astronomical facilities , no longer makes the problem unapproachable . in this work ,
we have applied an automated technique to the task of deriving star formation histories of five fields in the lmc from deep _ hst _ cmds .
however , assumptions still need to be made to make the problem tractable .
we have assumed throughout that the field star cmds can be parametrized by a single reddening , distance modulus , and imf ; that binary stars may be ignored ; that the imf does not vary in time ; and that a simple linear age - metallicity relation , based on lmc clusters , describes the enrichment history .
the star formation histories derived for the fields for a salpeter imf ( figs . [ sfh]a - c ) contain a number of features that match our common - sense expectations .
all of the fields have a significant number of old stars confined to the last age bin .
as these fields lie within 1 of five old globular clusters , contamination from globular cluster stars is expected ; in two of the fields , the contamination can account for the entire population of oldest stars detected in the field .
all of the fields show significant recent ( @xmath35 gyr ) star formation , in agreement with a large number of other lmc field star studies and with the lmc cluster age distribution
. in particular , the star formation history we derive for the ngc 1754 field agrees well with that derived by h97 for a field near ngc 1866 , once the contribution due to contaminating globular cluster stars is ignored .
the decline in the star formation rate between the ages of 0.5 and 1 gyr seen in each of the fields is naturally explained by inadequate mixing of stars formed most recently ; the crossing time in the bar is @xmath280 myr , implying a mixing time of a few hundred million years .
we find that the four bar fields ( ngc 1835 , ngc 1898 , ngc 2005 , and ngc 2019 ) experienced significantly more star formation in the age range 4@xmath78 gyr than did the ngc 1754 field , which lies in the disk ; this is most clearly seen in the solution containing 9 age bins ( fig .
[ sfh]c ) .
the existence of this population of stars unique to the bar naturally explains the broader giant branches and the existence of the stars 0.5 magnitudes brighter than the turnoff in each of the bar fields but not in the ngc 1754 field .
this result implies that the lmc s inner region is older than has been suggested by e.g. elson et al .
( 1997 ) and ardeberg et al .
. however , we also find disagreements between the best - fit models and observation ; the biggest disagreement is in the region of the rgb , where the model sequences fall @xmath20.08 magnitudes to the red of the observed sequence . to explore the consequences of this disagreement , we set the reddenings and distances of the models to extreme values and examined the resulting star formation histories .
in addition , we examined the success of our method in deriving the star formation rate of the ngc 2019 cluster cmd .
both explorations suggest that the model disagreements cause us to underestimate the contribution of the old populations .
thus , we expect that the star formation histories of figs .
[ sfh]a@xmath7c are roughly correct _ within _ the assumptions made about the imf , the effect of binary stars , and the age - metallicity relation . of these assumptions ,
the age - metallicity relation is on the weakest footing because of poor knowledge of lmc metallicities in the @xmath25@xmath710 gyr range .
while we think the method used here to derive star formation histories has potential , it can clearly be considerably improved .
helpful next steps would be to include binary stars in the models , explore the effect of different models of chemical evolution , and to test additional sets of isochrones .
future analyses will hopefully benefit from future improved knowledge of lmc metallicities and from better understanding of the later phases of stellar evolution .
i would like to thank paul hodge for his guidance throughout this project .
i also thank bob schommer , nick suntzeff , alistair walker , and chris smith for helpful discussions and for thorough readings of the original manuscript .
i thank the anonymous referee for several excellent comments which improved the manuscript .
andy dolphin provided the routines around which i built the program to extract the star formation histories .
this work was supported by stsci grant go05916 to nick suntzeff .
aparicio , a. , gallart , c. , & bertelli , g. 1997 , , 114 , 669 ardeberg , a. , gustafsson , b. , linde , p. , nissen , p .- e .
1997 , , 322 , l13 bertelli , g. , mateo , m. , chiosi , c. , & bressan , a. 1992 , , 387 , 320 bertelli , g. , bressan , a. , chiosi , c. , fagotto , f. , & nasi , e. 1994 , a&a supp . , 106 , 275 caldwell , j.a.r . , & coulson , i.m .
1986 , , 218 , 223 chaboyer , b. & kim , y .- c .
1995 , , 454 , 767 dohm - palmer , r.c . ,
skillman , e.d . , gallagher , j. , tolstoy , e. , mateo , m. , dufour , r.j . ,
saha , a. , hoessel , j.g . , & chiosi , c. 1998 , , 116 , 1227 dolphin , a. 1997 , new astronomy , 2 , 397 elson , r.a.w . , gilmore , g.f . , & santiago , b.x .
1997 , , 289 , 157 gallagher , j.s .
1996 , , 466 , 732 gallart , c. , aparicio , a. , bertelli , g. , & chiosi , c. 1996 , , 112 , 1950 geha , m.c .
1998 , , 115 , 1045 hardy , e. , buonanno , r. , corsi , c.e . ,
janes , k.a . , & schommer , r.a .
1984 , , 278 , 592 holtzman , j. et al .
1995 , , 107 , 156 holtzman , j.a .
1997 , , 113 , 656 mateo , m. , hodge , p. , & schommer , r.a .
1986 , , 311 , 113 ng , y.k .
1998 , a&a supp . , 132 , 133 olsen , k.a.g . ,
hodge , p.w . ,
mateo , m. , olszewski , e.w .
, schommer , r.a . , suntzeff , n.b . , & walker , a.r .
1998 , , 300 , 665 olszewski , e.w . ,
schommer , r.a . ,
suntzeff , n.b . , & harris , h.c .
1991 , , 101 , 515 olszewski , e.w .
, suntzeff , n.b . , & mateo , m. 1996 , , 34 , 511 press , w.h . , teukolsky , s.a . , vetterling , w.t .
, & flannery , b.p .
1992 , numerical recipes in c , ( cambridge : cambridge univ . press ) , 2nd edition rogers , f.j .
, & iglesias , c.a .
1992 , , 79 , 507 schaller , g. , schaerer , d. , meynet , g. , & maeder , a. 1992 , a&a supp . , 96 , 269 schechter , p.l . , mateo , m. , & saha , a. 1993 , , 105 , 1342 schwarzschild , m. 1958 , structure and evolution of the stars , ( new york : dover ) , p. 2
smecker - hane , t. 1997 , in aip conf .
393 , star formation near and far , eds .
holt & l.g .
mundy , ( woodbury : aip press ) , 571 tolstoy , e. , & saha , a. 1996 , , 462 , 672 tolstoy , e. , gallagher , j.s . ,
cole , a.a . ,
hoessel , j.g . ,
saha , a. , dohm - palmer , r.c . ,
skillman , e.d . ,
mateo , m. , & hurley - keller , d. 1998 , , 116 , 1244 vandenberg , d.a .
1998 , in iau symp .
189 , fundamental stellar properties : the interaction between observation and theory , eds .
bedding , a.j .
booth , & j. davis , ( dordrecht : kluwer ) , 439 whitmore , b. , & heyer , i. 1997 , wfpc2 instrument science report 97 - 08 lccccccccc 0 & 214.82 & 76.38 & 20.2346 & 0.0330 & 20.1296 & 0.0390 & 1 & 1 & + 1 & 184.19 & 81.18 & 19.5396 & 0.0360 & 19.4586 & 0.0330 & 1 & 1 & + 2 & 273.77 & 82.37 & 20.2416 & 0.0300 & 20.1716 & 0.0340 & 1 & 1 & + 3 & 96.77 & 91.67 & 19.3856 & 0.0240 & 18.4896 & 0.0250 & 1 & 1 & + 4 & 506.60 & 105.65 & 19.3696 & 0.0300 & 18.2548 & 0.0260 & 1 & 10 & + 5 & 513.20 & 109.20 & 19.8426 & 0.0360 & 18.8726 & 0.0310 & 1 & 1 & y + 6 & 296.19 & 110.51 & 20.0626 & 0.0270 & 19.6656 & 0.0250 & 1 & 1 & + 7 & 119.41 & 118.82 & 19.1496 & 0.0250 & 19.1116 & 0.0220 & 1 & 1 & + 8 & 247.40 & 127.18 & 19.6766 & 0.0350 & 19.2936 & 0.0290 & 1 & 1 & + 9 & 238.33 & 132.13 & 19.5236 & 0.0310 & 18.6766 & 0.0260 & 1 & 1 & y + & & & & & & & & & + [ wfphot ] cccr 0.02 & 18.4 - 18.7 & 1.5 & ngc 1835 + 0.04 & 18.3 & 1.5 & ngc 2005 + 0.04 & 18.4 - 18.7 & 1.5 & all + 0.06 & 18.3 & 1.5 & ngc 2005 + 0.06 & 18.4 - 18.7 & 1.5 & all + 0.08 & 18.3 & 1.5 & ngc 2005 + 0.08 & 18.4 - 18.7 & 1.5 & all + 0.10 & 18.3 & 1.5 & ngc 2005 + 0.10 & 18.4 - 18.7 & 1.5 & all + 0.12 & 18.3 & 1.5 & ngc 2005 + 0.12 & 18.4 - 18.7 & 1.5 & all + 0.02 & 18.4 - 18.7 & 2.0 & ngc 1835 + 0.04 & 18.3 & 2.0 & ngc 2005 + 0.04 & 18.4 - 18.7 & 2.0 & all + 0.06 & 18.3 & 2.0 & ngc 2005 + 0.06 & 18.4 - 18.7 & 2.0 & all + 0.08 & 18.3 & 2.0 & ngc 2005 + 0.08 & 18.4 - 18.7 & 2.0 & all + 0.10 & 18.3 & 2.0 & ngc 2005 + 0.10 & 18.4 - 18.7 & 2.0 & all + 0.12 & 18.3 & 2.0 & ngc 2005 + 0.12 & 18.4 - 18.7 & 2.0 & all + 0.02 & 18.4 - 18.7 & 2.35 & ngc 1835 + 0.04 & 18.3 & 2.35 & ngc 2005 + 0.04 & 18.4 - 18.7 & 2.35 & all + 0.06 & 18.3 & 2.35 & ngc 2005 + 0.06 & 18.4 - 18.7 & 2.35 & all + 0.08 & 18.3 & 2.35 & ngc 2005 + 0.08 & 18.4 - 18.7 & 2.35 & all + 0.10 & 18.3 & 2.35 & ngc 2005 + 0.10 & 18.4 - 18.7 & 2.35 & all + 0.12 & 18.3 & 2.35 & ngc 2005 + 0.12 & 18.4 - 18.7 & 2.35 & all + 0.02 & 18.4 - 18.7 & 3.0 & ngc 1835 + 0.04 & 18.3 & 3.0 & ngc 2005 + 0.04 & 18.4 - 18.7 & 3.0 & all + 0.06 & 18.3 & 3.0 & ngc 2005 + 0.06 & 18.4 - 18.7 & 3.0 & all + 0.08 & 18.3 & 3.0 & ngc 2005 + 0.08 & 18.4 - 18.7 & 3.0 & all + 0.10 & 18.3 & 3.0 & ngc 2005 + 0.10 & 18.4 - 18.7 & 3.0 & all + 0.12 & 18.3 & 3.0 & ngc 2005 + 0.12 & 18.4 - 18.7 & 3.0 & all + [ mpars ] cccccc 0.08 & 18.5 & 2.35 & 250 & @xmath2185 & 10 + 0.08 & 18.5 & 2.35 & 500 & @xmath2370 & 10 + 0.08 & 18.5 & 2.35 & 1000 & @xmath2740 & 25 + 0.08 & 18.5 & 2.35 & 2000 & @xmath21480 & 10 + 0.08 & 18.5 & 2.35 & 3000 & @xmath22220 & 10 + lcc ngc 1754 & 0.06 @xmath36 0.01 & 18.50 @xmath36 0.05 + ngc 1835 & 0.04 @xmath36 0.01 & 18.50 @xmath36 0.05 + ngc 1898 & 0.06 @xmath36 0.01 & 18.40 @xmath36 0.05 + ngc 2005 & 0.06 @xmath36 0.01 & 18.45 @xmath36 0.05 + ngc 2019 & 0.06 @xmath36 0.01 & 18.50 @xmath36 0.05 + | we present results on the analysis of background field stars found in _ hst _
wfpc2 observations of six of the old globular clusters of the large magellanic cloud . treated as contaminants by the globular cluster analysts , we produce @xmath0 color - magnitude diagrams of the field stars and
use them to explore the lmc s star formation history .
the photometry approaches @xmath1 , well below the turnoff of an ancient ( @xmath214 gyr ) lmc population of stars .
the field star cmds are generally characterized by an upper main sequence broadened by stellar evolution , an old red giant branch , a prominent red clump , and an unevolved lower main sequence .
the cmds also contain a few visual differences , the most obvious of which is the smeared appearance of the ngc 1916 field caused by heavy differential reddening .
more subtly , the base of the subgiant branch near the old turnoff appears extended in @xmath3 and the red giant branch appears broad in @xmath4 in four of the fields , but not in the ngc 1754 field .
we use a maximum - likelihood technique to fit model cmds drawn from bertelli et al .
( 1994 ) isochrones to the observed cmds .
we constrain the models by the age - metallicity relation derived from lmc clusters , test four imf slopes , and fit for the reddening , distance modulus , and star formation rate .
we find that we can just resolve structure in sfr(@xmath5 ) with time steps of @xmath20.15 in log age , implying a resolution of @xmath6 years at an age of 1 gyr . for a salpeter imf ,
our derived star formation history for the ngc 1754 field is characterized by an enhanced star formation rate over the past 4 gyr , qualitatively resembling that derived by others for a variety of lmc fields .
the remaining four fields , however , appear to have had high levels of star formation activity as long as 5@xmath78 gyr ago ; these fields lie in the lmc bar while the ngc 1754 field lies in the disk , suggesting that the inner regions of the lmc contain significantly more older stars than the outer regions . examining the residuals of the models and observations , we find that the old red giant branches of the models provide a poor fit to the observations , which suggests an error in the model isochrones . the effect of the disagreement appears to be to underestimate the contribution of the old population . |
the cmb has been a finest probe of cosmology .
a complete characterization of the cmb is made by quantifying the four stokes parameters @xcite associated to its unpolarized intensity , lp and cp . temperature and polarization anisotropies of the cmb have provided invaluable insight into the universe that we live in .
venture into the cmb studies began with its discovery in 1965 @xcite which showed the cmb radiation to be a black - body with a mean temperature of t @xmath1 2.7255 k. subsequently , the first detection of angular variation of the cmb mean temperature was measured with the cobe satellite @xcite and today , the temperature fluctuations have been measured with an unprecedented precision @xcite .
cmb is also linearly polarized due to thomson scattering at the surface of last scattering .
this polarization level is at @xmath1 5@xmath14 of the temperature anisotropies .
polarization fluctuations have been measured with a high precision and many future experiments are aimed to increase the precision of polarization measurements @xcite .
the planck experiment reached a sensitivity level , @xmath0 @xmath1 10@xmath2 for temperature and polarization , a much higher level of sensitivity compared to the wmap satellite .
together , the cmb temperature and lp data bring an overwhelming support for the cosmological standard model or the @xmath15cdm @xcite model .
it is only natural to wonder what information does the cp of the cmb store ?
the cp of the cmb has not been explored extensively and the current level of cmb cp measurement stands at a level @xmath16 @xmath1 10@xmath4 in the scales of 8@xmath9 and 24@xmath9 @xcite .
this upper limit is much higher than the level of cp signal expected from any cosmologically relevant sources .
previously , @xcite had made one of the first efforts to measure to the cp of the cmb .
recently , class @xcite and piper @xcite experiments proposed to measure the lp and cp of the cmb , however , the experiment is designed to focus on primarily the measurements of the lp of the cmb .
these experiments aim to detect primordial b - modes using variable - delay polarization ( vpm ) instruments where the observing strategy relies on the expectation of cp of the sky to be null to constrain systematic uncertainties . in this paper , we have provided galactic cp maps for frequencies relevant for both class at 40 ghz and piper at 220 ghz , showing that the galactic cp effects are important for frequencies below 50 ghz .
there are various cosmologically important sources which may induce a cp in the cmb via different mechanisms .
some of these cp production channels are intrinsic to the emission from a certain type of source , for example synchrotron emitting radio sources @xcite .
some are due to the effects of external magnetic fields @xcite or other birefringent effects , and finally some mechanisms propose cp generation in the cmb by models that stand on departures from the standard model of particle physics @xcite . among the sources which can induce a cp in the cmb due to presence of a magnetic field , are the so called pop iii stars , also known as the first stars .
see ( @xcite ) for a review of the pop iii stars .
@xcite describes how a cp in the cmb can be induced by the remnants of pop iii stars that went supernovae ( sne ) . these stars , residing in dark matter mini - halos ,
provide a window into the early structure formation which ended the cosmological dark ages and began the re - ionization along with metal enrichment of the inter - galactic medium .
pop iii stars are expected to exist based on the numerical simulations of primordial stars formation and fossil abundance of sne . however , there are no definite constraints on the properties of these stars @xcite
. these stars are generally @xmath17 expected to be directly detected by the most advanced future space telescopes like the wfirst and the jwst @xcite , except under certain conditions when these pop iii stars explode into pair instability sne @xcite or hyper nova @xcite releasing energies around @xmath1 10@xmath18 ergs .
these stars and their properties are speculated by numerical simulations @xcite , however , they are far from being verified by observations .
cp of the cmb provides an indirect and a much economical way of exploring into these pop iii stars . in this paper
we will primarily focus on s / n determination of the cp signal in the cmb due to the pop iii stars . however , there are other cosmologically important sources which induce cp in the cmb .
these sources include the primordial magnetic field ( pmf ) @xcite , different modifications and symmetry breaking mechanisms @xcite beyond the standard model particle physics and the galaxy clusters @xcite .
most of these sources induce a lower signal level in cp of the cmb when compared to the level of cp induced by the pop iii stars .
however , these sources could certainly be explored via the cp in the cmb once the instrumental sensitivity improves .
in addition to the cosmologically important sources of cp , the milky way ( mw ) galaxy produces synchrotron radiation which is intrinsically circularly polarized . circularly polarized synchrotron emission from the mw galaxy acts as a foreground towards the detection of the cosmological cp in the cmb .
currently , there is not enough observational data to accurately shed light on the level of cp from the gse . in this paper
, we generate numerical simulations of the galactic cp due to synchrotron emission using data - driven models of the gmf and cosmic ray electron energy distribution .
cp in the cmb could potentially detect the existence of the pop iii stars , symmetry breaking in the early universe or the existence of primordial magnetic field .
implementing direct detection of these sources will need a revolution on the instrumentation front , involve a long time - scale and a very high cost .
exploring some of these highly interesting sources indirectly via the cp in the cmb is possible within the current reach of instrumentation , achievable at a moderate timeline and cost . in this paper
, we will discuss the sources , foregrounds and detection prospects of the cosmological cp of the cmb .
[ sec : transfer ] discusses theoretical framework needed for the description of cp , sec .
[ sec : overview ] presents an overview of various sources and mechanisms which induce cp in the cmb .
[ sec : foregrounds]-[sec : maps ] discuss the galactic foregrounds in cp and simulations .
[ sec : detection ] represent the results on detection prospects .
finally , we discuss the future directions and implications of this work in sec .
[ sec : discussion ] .
a complete description of polarization of an electro - magnetic ( em ) wave is described by four stokes parameters , intensity i , linear polarizations ( lp ) q and u , and circular polarization ( cp ) v. together ( i , q , u , v ) represent the stokes vector associated to the em wave .
the evolution of different stokes vector components of an em wave propagating through a plasma are governed by the following polarization transfer equation @xcite . @xmath19 where the spatial derivatives on the left side of the equation indicate the change in stokes vector along the line of sight , taken to be inclined along the z axis .
the coefficients @xmath20 indicate emissivity and @xmath21 indicate absorption coefficients corresponding to the stokes vectors i , q , u and v. under an isotropic distribution of unperturbed particles in a plasma , the conversion coefficient , @xmath22 between q and v , vanishes due to dielectric symmetries @xcite .
( [ eq : transfer ] ) uses a coordinate frame where the sky projected magnetic field component is aligned along the y axis . under this geometry , stokes components + u is defined as lp aligned along an axis which makes an angle of 45@xmath9 , clockwise with respect to the y axis .
component + q(-q ) is defined as lp aligned along the y(x ) axis .
please see fig .
[ fig : qu ] for a schematic view of the coordinate system used .
stokes components i and v are invariants under coordinate transformations @xcite .
the ordinary rotation between stokes components q and u are driven by the faraday rotation ( fr ) coefficient , @xmath23 .
fr is simply the rotation of the plane of lp of an em wave propagating through a magneto - active plasma due to a local magnetic field along the line of sight . the generalized rotation , also known as the faraday conversion ( fc ) , described by the conversion coefficient , @xmath24 , controls the transfer between u and v components . to understand the fc effect , consider a linearly polarized em wave propagating through a plasma where an external magnetic field is aligned along the y axis .
charged particles moving along the y axis will experience a different lorentz force than those along the x axis .
let the lp of the incoming em wave be + u ( for example ) , which can be decomposed into two linear components , each along the y and the x axes .
since the particles along the x and y axes move differently due to the asymmetry introduced by the external magnetic field , there will now be a phase difference between the orthogonal lp components ( along x and y axes ) .
thus em wave will therefore have acquired a cp , depicted by the stokes component v. , between the mutually orthogonal e - field components is zero and represents a only a linearly polarized cmb prior to its passage through the pop iii remnant .
the magnetic field in the remnant creates an asymmetry due to the difference in lorentz forces between the charged particles moving along @xmath25 and @xmath26 .
this now results in a non - zero phase difference between the mutually orthogonal e - field components of the outgoing wave and hence creates a circularly polarized wave .
, scaledwidth=50.0% ] in the case of the cmb , under the standard cosmological model , @xmath27 vanishes @xcite . the coefficients @xmath28 are non - zero due to anisotropic thomson scattering .
effects of non - standard cosmology or new physics can induce a non - zero @xmath27 .
we provide a summary of such possibilities in sec .
[ sec : summary ] and in table .
( [ table : table1 ] ) .
conversion between u and q components of the cmb due to fr induced by the galactic magnetic field has been been studied by @xcite .
possibilities of conversion between u and v components of the cmb due to the fc effects induced by the pop iii remnants and galaxy clusters are discussed in sec .
[ sec : pop3 ] and sec .
[ sec : gc ] , respectively .
( [ eq : transfer ] ) also applies to the gse . under a coordinate system where the sky - projected magnetic field is inclined along the y - axis ,
@xmath29 and @xmath30 of the synchrotron emission are set to be zero . for the gse ,
@xmath31 is also set to be zero @xcite .
a detailed description of the generalized transfer equation in the context of synchrotron emission can be found in @xcite .
in this paper we adopt the symbols and conventions from @xcite to describe the polarization relevant quantities .
pop iii stars mark the transition of a simple homogeneous universe of h - he gas into a complex and structured universe after the cosmic dark ages .
pop iii stars are therefore also known as the first stars .
simple interpretation of optical depth observed by wmap and planck suggest star formation activity at z @xmath32 11 @xcite .
supernovae explosions of the pop iii stars was responsible for the metal enrichment of the inter - galactic medium @xcite .
these pop iii stars are predicted to form in dark matter minihalos of mass @xmath1 10@xmath33 around z @xmath1 20 - 30 @xcite .
a low - mass halo is needed to virial temperature below the threshold of t@xmath34=10@xmath35k to allow efficient atomic hydrogen cooling , necessary for collapse .
the pop iii stars are yet unobserved and there is significant lack of certainty of their properties , most of which are predicted from numerical simulations and therefore , are highly model dependent .
significance of understanding of the pop iii stars is enormous .
many of the implications drawn from the pop iii stars depend on the mass of these primordial stars .
if the pop iii stars are massive ( @xmath12 100 @xmath36 ) , they can be connected to several effects which can be tested in the distant future .
for example , sunyaev zeldovich effects of the cmb @xcite , gravity waves from black holes formed from pop iii remnants @xcite .
one of the most certain and significant effects of the pop iii stars is the cosmological heavy element production and the cosmic reionization @xcite . in @xcite ,
pop iii stars are established as a source of appreciable cp in the cmb .
intrinsic cp of such sources is small , however , the cmb acts as a back - light in this scenario .
as the cmb photons pass through the relativistic plasma of the pop iii remnants , a fraction of the cmb linear polarization is converted into cp , via the fc mechanism
. please refer to sec .
[ sec : transfer ] for a schematic explanation of the fc mechanism under which @xmath24 in eq .
( [ eq : transfer ] ) describes the transfer of the stokes u into the stokes v component in the cmb . using a simple analytical model of a sn remnant of a pop iii star
, @xcite evaluates @xmath24 described in eq .
( [ eq : transfer ] ) as @xmath37 where @xmath38 is the age of the sn remnant , @xmath39 is the energy of the explosion and @xmath40 and @xmath41 are the fractions of the explosion energy respectively into the relativistic electron energy and the magnetic fields in a sn remnant .
@xmath42 is the cmb observation frequency .
following the so - called halo model , the angular power spectrum of the cp due to the pop iii stars is evaluated .
angular power spectrum is effectively the square of the rms fluctuation in v , or @xmath43 . in @xcite ,
pop iii stars are assumed to only exist in halos with virial temperature , @xmath44 k where atomic hydrogen cooling is effective for the collapse .
since the signal of cp due to fc mechanism falls off with frequency , we set our normalization frequency to be 10 ghz in future equations in this paper .
a frequency much lower than 10 ghz calls for a full solution of the transfer equation , which will be addressed in future work . around @xmath45 ,
a simple formula corresponding to the brightness temperature associated to the fluctuation @xmath46v can be expressed as the following .
@xmath47 where n@xmath11 is the number of pop iii stars per halo .
@xmath48 is the spectral index of the electron energy distribution around the pop iii remnant , and @xmath49 @xmath1 2 @xcite . in fig .
[ fig : signal_all ] different pop iii associated cp signals are shown .
note that , @xmath50 falls off very sharply with frequency of cmb observation as @xmath51 and also with increasing age , @xmath38 of the pop iii remnants . in fig .
[ fig : signal_all ] we use @xmath52 which comes from the compton cooling timescale of the remnants .
the number of pop iii stars in a halo is uncertain and could be up to @xmath1 10@xmath53 @xcite .
@xmath50 increases with n@xmath5 as n@xmath54 . not explicit from eq .
( [ eq : pop3 ] ) , the quadratic dependence of the signal in @xmath46v on the local magnetic field , b is b@xmath55 .
this dependence can be easily seen in eq .
( [ eq : kc ] ) via linear dependence of @xmath24 on @xmath40 or effectively the magnetic energy ( @xmath56 ) .
all relevant cosmological and astrophysical parameters are also described in @xcite .
pop iii signal is produced via faraday conversion . the signal level shown depend on the number of pop iii stars per halo , age of the remnant , energy generated at collapse and very strongly on the frequency of the cmb observation @xcite .
a few cases for the age of the remnant of the pop iii explosion , with @xmath57 in years are shown .
the frequency of observation of the cmb has been chosen to be @xmath58 10 ghz and the number of pop iii stars is assumed to be 100 per halo .
also shown is the galaxy cluster signal due to fc at @xmath42=10 ghz ( taken from @xcite ) and the signal due to primordial magnetic field at @xmath42=10 ghz with @xmath59=1 ng ( taken from @xcite ) . , scaledwidth=45.0% ] [ table : table1 ] another source of cp in the cmb are the galaxy clusters which induce cp via the fc mechanism due to their magnetic field and relativistic electrons .
this scenario was explored in @xcite where a mean field of 10 @xmath13 g with a coherent scale of 1 mpc was adopted . in fig .
[ fig : signal_all ] we also represent the @xmath60 induced in the cmb due to fc in the galaxy clusters at @xmath61 ghz , a result which has been derived from @xcite .
fc coefficient in this scenario is given by , @xmath62 where rm is the rotation measure for the galaxy clusters .
temperature equivalent stokes v in the cmb due to galaxy clusters around @xmath63 can be simply expressed as @xmath64 note that the cp signal due to galaxy clusters is much smaller compared to that due to the pop iii stars .
primordial magnetic field ( pmf ) is speculated as a source of magnetic field in the universe @xcite .
current limits on the pmf is @xmath1 few ng today @xcite .
@xcite showed that cp is naturally generated in the cmb in presence of thomson scattering , primordial magnetic field , spatial curvature and adiabatic fluctuations .
this cp generation is intrinsic ( @xmath27 ) and does not require any pre - existing linear polarization , unlike fc . in fig .
[ fig : signal_all ] @xmath60 due to the presence of a primordial magnetic field of an equivalent field strength , b@xmath65=1 ng is shown , following @xcite .
@xmath66 for all cases in fig .
[ fig : signal_all ] , @xmath42 is set at 10 ghz . under the realms of the standard model
, the cmb does not have any significant intrinsic stokes v component .
however , an early universe symmetry - breaking , or the presence of new - physics , or scattering processes in presence of the primordial magnetic field could induce an intrinsic stokes v component in the cmb .
many of these possibilities are summarized in table .
( [ table : table1 ] ) along with references .
one of the most concerning source of non - cosmological cp is synchrotron emission in our galaxy .
gse is intrinsically circularly polarized and therefore poses as a foreground to cosmologically important cp sources in the cmb .
the gse is produced by the cosmic ray electrons in presence of the galactic magnetic field .
we will discuss the gse in much more detail in sec .
[ sec : foregroundsa ] and sec .
[ sec : foregroundsb ] . table .
( [ table : table1 ] ) also lists sources for fc driven and intrinsic cp generation . in each case ,
( [ table : table1 ] ) lists frequency and magnetic field dependence of the signal .
each case also projects an expected signal at @xmath42=10 ghz .
total brightness of the galactic radio sky is dominated by the diffused synchrotron emission due to the cosmic ray electrons and positrons gyrating in the gmf @xcite .
this emission is significant between frequencies of a few tens of mhz to a few tens of ghz .
the primary sources of cosmic ray electrons are supernovae remnants , pulsars in the galaxy .
the energy range of the cosmic ray electrons are between a few hundred mev to tens of gev .
the gmf strength @xmath1 a few @xmath13 g .
the ordered components of the gmf is coherent on kpc scales , however , the gmf also has a random component which varies in scales of a few hundred parsecs @xcite .
synchrotron emission due to the mw galaxy happens to be the strongest source of foreground to the cmb observations at low frequencies @xcite .
synchrotron emission is also naturally expected to have a rather high level of lp , and there has not be any significant measurements of its cp .
synchrotron emission coming from relativistic cosmic rays , is expected to be elliptically polarized even with an isotropic distribution of the electron velocity @xcite .
emissivity , @xmath27 or the intrinsic cp of the gse under a power - law electron energy distribution is given by @xmath67 where @xmath68 is the elemental length along the line of sight .
@xmath69 is the emissivity associated to the unpolarized intensity of the synchrotron emission .
@xmath70 is the gyro - frequency of the magnetic field component perpendicular to the line of sight .
@xmath71 is a function of the spectral index of the electron energy distribution and its detailed form is found in @xcite . in this paper , for simplicity , we set @xmath72 ( see eq .
( 31 - 35 ) of @xcite ) which is obtained using reasonable parameters for the electron energy distribution obtained from the observation of the crab nebula @xcite and low frequency observations from the planck satellite @xcite .
typically , f@xmath73 lies between 1 and 2 , if 1<@xmath74<3 , @xmath75 being the spectral index of cosmic ray electron energy distribution .
gse is a crucial and significant source of foreground towards cosmologically important sources of cp in the cmb . in this paper
, we only focus on the intrinsic emission of the galactic synchrotron as the most significant foreground towards the cosmic cp .
circularly polarized synchrotron emission in the mw galaxy is described by the term @xmath27 in eq .
( [ eq : transfer ] ) and its precise form is given in eq .
( [ eq : vsync ] ) .
another possible mechanism of galactic cp generation is the fc in the galaxy due to relativistic cosmic ray electrons gyrating in the gmf . in this section , using simple analytic models , we will estimate the magnitude of fc induced galactic cp .
we begin with the following cosmic ray electron density distribution @xcite .
@xmath76 where r is the galactic radius and z is the height .
synchrotron spectral index , @xmath77 @xmath1 3 , lorentz factor , @xmath78 lies between 100 and 300 for the galaxy and c@xmath79=4 @xmath7 @xmath80 @xmath81 .
the radial scale and disc height are respectively set by @xmath82 kpc and @xmath83=1 kpc .
we use @xmath84 0.5 kpc and @xmath85 0.5 kpc at @xmath42=10 ghz in eq .
( [ eq : electron ] ) to obtain relativistic electron density in the mw galaxy .
we then use appendix d. of @xcite to obtain fc coefficient in the galaxy , @xmath86kpc@xmath87 .
the amount of cp induced via the fc effect in the galaxy is @xmath1 @xmath88 , where @xmath89 is the total incoming lp due to the cmb and the gse at a given frequency @xmath42 .
@xmath90k , and therefore induces a cp due its passage through the galaxy as @xmath91k , which is much smaller than any other cosmologically important source of cp in the cmb .
gse also has a significant level of intrinsic lp @xcite and therefore is subjected to fc effects in the galaxy . in sec .
[ sec : psm ] , we show the expected level of lp in the gse ( see fig . [
fig : planck_vs_hammurabi ] ) at @xmath42=30 ghz .
we note that the highest level of intrinsic lp of the gse is @xmath1 a few tenths of k. this level does not change significantly with our frequencies of interest ( @xmath42@xmath92 5 ghz ) .
therefore , cp induced in the galaxy due fc of the gse is @xmath1 @xmath93 @xmath1 10@xmath94kpc@xmath87k .
in contrast , intrinsic cp of the gse at @xmath42=10 ghz is @xmath1 @xmath95 k , obtained using equations in appendix d. of @xcite with b @xmath1 10@xmath13 g .
this level is also supported by fig .
[ fig : hammu_map ] . therefore , fc induced cp of the gse is negligible compared to its intrinsic cp .
cp levels in the galaxy is a function of frequency and the relative importance of different channels to generate cp , depends on the frequency .
the @xmath96 @xmath97 of the ratio of intrinsically generated cp vs fc induced cp in the galaxy is given by @xmath98 where @xmath89 is the total incoming lp in the galaxy , composed of contributions from both the cmb and the gse .
@xmath99 is the angle between the magnetic field and the line of sight , @xmath100 is the minimum lorentz factor for the relativistic electrons in the galaxy and @xmath101 is the cyclotron frequency given by @xmath102 .
the ratio @xmath103 is @xmath104 at @xmath42=10 ghz , b=10@xmath13 g , @xmath105 and @xmath106 .
lp in the cmb is not a function of frequency and is given by @xmath107 k. the level of lp in the galactic synchrotron emission is significant compared to the unpolarized intensity .
@xmath108 k at @xmath42=10 ghz and eventually falls off with higher frequency . therefore , along a given line of sight , @xmath103 is a monotonically increasing function of frequency .
therefore , in our frequencies of interest ( 5 ghz < @xmath42 < 30 ghz ) for the cmb cp measurement , fc induced cp of the cmb or the gse is not an important effect .
to every emission there is an associated absorption .
this applies to both the unpolarized stokes intensity , i and circular polarization stokes intensity v. the intrinsic emission in stokes v will be extinct if the emission and absorption were perfectly balanced , or @xmath109 .
therefore , it is also important to consider the absorption of the circularly polarized emission in the galaxy .
the absorption is given by @xmath110 .
following appendix d of @xcite we obtain , @xmath111 where @xmath112 is a function of the spectral index , @xmath77 , of the relativistic electron energy distribution . for the galaxy , we use @xmath113 . for the gse , @xmath114 .
the ratio @xmath115 is @xmath116 at @xmath42=10 ghz , b @xmath1 10 @xmath13 g , @xmath117 k and is an increasing function of frequency .
this implies that absorption of the cp emission in cmb observation frequencies ( 1 ghz or above ) is not significant .
absorption of the synchrotron emission component is however significant in so - called self - absorbed synchrotron sources where @xmath118 .
this scenario is realized at much lower frequencies , @xmath119 @xmath120 10 mhz @xcite .
therefore , synchrotron self absorption of its circularly polarized emission is not a concern in our case .
synchrotron self - absorption of the unpolarized intensity , is also not important for the galaxy in the frequencies of interest ( @xmath42 @xmath92 1 ghz ) . in the case of very low frequencies
@xmath121 1 ghz , some extra - galactic sources could become self - absorbed or optically thick . in this low frequency regime ,
synchrotron flux from the sources decreases with decreasing frequency . on the contrary , at higher frequencies , flux emitted by the synchrotron sources decreases with increasing frequency
this turn - over in flux - frequency relation pollutes the smooth synchrotron frequency dependence , altering the spectral index of the synchrotron brightness temperature .
spectral smoothness is important in order to successfully remove foregrounds .
synchrotron flux from the galaxy is still high at frequencies @xmath32 1 ghz .
however , due to the smooth dependence of the synchrotron flux on the frequency , foreground removal via a polynomial fit , is easier .
this is especially relevant where the signal of interest ( for example , the cmb cp due to galaxy clusters ) is lower than the foregrounds .
unless the number of such sources is small enough , it is wiser to confine the search for the cosmic cp at frequencies @xmath32 5 ghz .
this is also the motivation for us to confine the cmb cp observation frequencies between 5 - 30 ghz .
below we summarize the conclusions from the current section .
* cp induced in the cmb due to the mw galaxy ( via the fc mechanism ) is much smaller than the levels of cp induced in the cmb due to cosmologically important sources ( see table . ( [ table : table1 ] ) ) . *
cp induced in the gse via the fc mechanism is much smaller than the intrinsic emission of circularly polarized synchrotron radiation in the galaxy . *
cp induced intrinsically ( via the @xmath27 term in eq .
( [ eq : transfer ] ) ) is higher than the fc induced cp in the galaxy at all frequencies of interest .
* self - absorption of the circularly polarized gse is not important in the frequencies of interest .
* self - absorption of unpolarized intensity of the gse may pollute the smoothness of synchrotron spectra at frequencies @xmath121 1 ghz . [
cols="^,^ " , ]
in this paper we have evaluated s / n for the detection of cosmologically important cp signals in the cmb .
frequency range of observation was chosen to be @xmath1 5 - 30 ghz
. the lower limit of @xmath1 5 ghz in the frequency range of interest is chosen to avoid irregularities in the synchrotron spectra due to extra - galactic self - absorbing sources . additionally ,
treatment of cp in lower frequencies need a full solution of the polarization transfer equation in many scenarios @xcite
. an upper limit of @xmath1 30 ghz is chosen due to sharp fall off of the cp signal with increasing frequency .
our work follows from @xcite who showed that pop iii stars could induce a @xmath122 cp signal in the cmb .
in addition to the pop iii signal , cp signal induced in the cmb due to the galaxy clusters and other primordial sources , such as the primordial magnetic field or symmetry breaking mechanisms , were also considered .
of all the cosmologically important sources of cp discussed in this paper , cp due to the pop iii stars dominate with a significantly higher signal in the frequency range of interest .
a detailed description of the mechanisms that produce cp in various sources are discussed in sec .
[ sec : overview ] and sec .
[ sec : transfer ] .
cp in the cmb is produced by the pop iii stars via the fc mechanism which transforms an incoming linear polarization into circular polarization in presence of an external magnetic field .
an important foreground to the cp observation is the galactic cp due to the gse , which is naturally ( intrinsically ) circularly polarized .
we evaluate the signal level of the cp generated from the gse using numerical simulations generated by the @xmath123 code where we have implemented the calculation of stokes v ( component corresponding to cp ) following eq .
( [ eq : vsync ] ) .
the goal frequency of observation is in the 5 - 30 ghz range which is chosen due to the reasons described earlier in this section .
there is not much known about the mass and other properties of the pop iii stars .
if the number of pop iii stars per halo is as high as 1000 , then observations at higher frequency ( of up to 50 ghz ) could lead to a s / n higher than unity .
the final result of our work is summarized in fig .
( [ fig : snr_grid1]-[fig : snr_grid1_dg ] ) .
a s / n significantly higher than unity is achievable if n@xmath11 @xmath12 100 .
generally , a s / n higher than unity is accessible if n@xmath11 @xmath12 10 , with an appropriately chosen frequency of observation and beam resolution . under the most optimistic scenario ( for example , n@xmath11=1000 ) ,
a s / n higher than unity is accessible at @xmath42=50 ghz with beam resolution of up to 40. under the least optimistic scenario , when n@xmath11=1 , we do not expect a s / n higher than unity
. limitations of our results come from the current status of the gmf models .
our results for the galactic foreground in cp is based on numerical simulations which are partially driven by galactic synchrotron and low frequency cmb data .
the gmf models considered in this paper are compared with haslam data ( 408 mhz map ) , low frequency cmb data from planck and wmap satellites .
however , the resolution of these datasets are at best 40-1@xmath9 , limiting our knowledge of the gmf at smaller scales . maps of stokes v presented in this paper depend on the gmf models we have adopted , which need to be improved to produce the accurate maps of the gse which match the observations more closely .
this will also improve the stokes v maps .
finally , there is no observed maps of the galactic cp .
an all - sky map of the observed galactic cp will reveal the nature of the gmf in greater detail , verify the theoretical predictions on cp from the polarization transfer equations and yield foregrounds levels to the measurements of the cosmic cp .
a low frequency measurement of the cosmic cp in the cmb can reveal information about the pop iii stars , early universe symmetry breaking or new physics , galaxy clusters or even the primordial magnetic field .
the signal from the pop iii stars are particularly interesting due to their expected high signal level and consequently , significantly high s / n .
observing cp in the cmb will be an indirect probe of the pop iii stars which are highly significant as the first formed structures since the cosmic dark ages and seeds of reionization of the universe .
not much is known about the pop iii stars currently .
direct observation of these high redshift objects ( z @xmath12 15 ) are generally beyond the reach of future telescopes like the jwst . observing the cp induced in the cmb provides a much economical way of learning about the existence and nature of the pop iii stars , which can even be realized in the immediate future . with the current status of instrumentation ,
observing cp of the cmb provides the highest promise in probing the pop iii stars .
if the instrumentation is improved in the near future , new physics related signal can also be explored which currently remains far out of the reach of the highly expensive modern day accelerators .
finally , an immediate practical significance of this work is towards the cosmic b - mode exploration .
this applies to the telescopes that are now being built and propose to explore primordial b - modes using vpm techniques .
the observing strategy relies on using cp as a systematics rejections channel .
this work points out that the galactic cp effects are important in large scales , especially for frequencies below 50 ghz .
we thank hiroyuki tashiro for important discussions and suggestions as well for a previous collaboration which inspired the current work .
we thank tess jaffe for her help with the @xmath123 code .
we thank ari kaplan for discussions on planck sky models simulations used to compare with some of the @xmath123 results .
we thank the anonymous referee for an insightful and encouraging review . | the cosmic microwave background ( cmb ) is one of the finest probes of cosmology .
its all - sky temperature and linear polarization ( lp ) fluctuations have been measured precisely at a level of @xmath0 @xmath1 10@xmath2 . in comparison ,
circular polarization ( cp ) of the cmb , however , has not been precisely explored .
current upper limit on the cp of the cmb is at a level of @xmath3 @xmath1 10@xmath4 and is limited on large scales . some of the cosmologically important sources which can induce a cp in the cmb include early universe symmetry breaking , primordial magnetic field , galaxy clusters and pop iii stars ( also known as the first stars ) . among these sources ,
pop iii stars are expected to induce the strongest signal with levels strongly dependent on the frequency of observation and on the number , n@xmath5 , of the pop iii stars per halo .
optimistically , a cp signal in the cmb due to the pop iii stars could be at a level of @xmath6 2 @xmath7 10@xmath8 in scales of 1@xmath9 at 10 ghz , which is much smaller than the currently existing upper limits on the cp measurements .
primary foregrounds in the cosmological cp detection will come from the galactic synchrotron emission ( gse ) , which is naturally ( intrinsically ) circularly polarized .
we use data - driven models of the galactic magnetic field ( gmf ) , thermal electron density and relativistic electron density to simulate all - sky maps of the galactic cp .
this work also points out that the galactic cp levels are important below 50 ghz and is an important factor for telescopes aiming to detect primordial b - modes using cp as a systematics rejection channel . in this paper , we focus on s / n evaluation for the detectability of the pop iii induced cp signal in the cmb .
we find that a s / n higher than unity is achievable , for example , with a 10 m telescope and an observation time of 20 months at 10 ghz , if n@xmath10 100 .
we also find that , if frequency of observation and resolution of the beam is appropriately chosen , a s / n higher than unity is possible with n@xmath11 @xmath12 10 and resolution - per - pixel @xmath1 1 @xmath13k at an observation time of 60 months .
a summary of different sources which can induce a cp in the cmb is summarized in table .
[ table : table1 ] .
final results related to the detectability of cosmologically important cp are summarized in fig .
( [ fig : snr_grid1]-[fig : snr_grid1_dg ] ) . |
let @xmath0 be the graph with set of vertices @xmath1 and set of ( unoriented ) bonds @xmath2 , where @xmath3 denote the vectors in the canonical basis of @xmath4 .
let @xmath5 be a sequence in the interval @xmath6 $ ] and consider a bernoulli bond percolation model where each bond @xmath7 is open with probability @xmath8 , where @xmath9 denotes the @xmath10 distance between the two endpoints of @xmath11 .
that is , take @xmath12 , where @xmath13 , @xmath14 is the canonical product @xmath15-algebra , and @xmath16 , where @xmath17 .
an element @xmath18 is called a percolation configuration . as usual
, the set @xmath19 denotes the set of configurations such that the origin is connected to infinitely many vertices by paths of open bonds ( bonds where @xmath20 ) .
our principal assumption concerning the sequence @xmath21 will be @xmath22 so that , by borel - cantelli lemma , we have @xmath23 .
we now consider a truncation of the sequence @xmath21 at some finite range @xmath24 .
more precisely , for each @xmath25 consider the truncated sequence @xmath26 , defined by @xmath27{l}p_i,\mbox { if } i{\leqslant}k,\\ 0,\ \mbox { if } i > k .
\end{array}\right.\label{eq : truncation}\ ] ] and the measure @xmath28 , where @xmath29 .
then , the * truncation question * is : does @xmath30 hold for @xmath24 large enough ? the works @xcite , @xcite , @xcite , @xcite and @xcite ( in chronological order ) give an affirmative answer to the truncation question under different sets of assumptions on the dimension @xmath31 and the sequence @xmath32 .
in particular , @xcite gives an affirmative answer for @xmath33 and no assumption on @xmath32 other than ; moreover , this work shows how the analogous question for the long - range potts model can be studied via a long range percolation model .
we would like to mention that the general truncation question for @xmath34 is still open and it is not difficult to see that for @xmath35 the answer is negative . in the nonsummable situation ,
the positive answer to the truncation question ( in dimensions more than 1 ) appears to be more robust than in the summable case .
indeed , the presence of first - order transitions in the occupation density , or in a temperature - like parameter for summable infinite - range models , causes the truncation question to have a negative answer , as observed in @xcite .
although continuity of the transition is known for ising models , and their associated random - cluster models , in considerable generality ( see for example the recent work @xcite ) , this is not the case for independent percolation , where even in @xmath36 it is a famous open question in the nearest - neighbor model , while for @xmath37-state potts models first - order transitions are quite common for @xmath38 ( see the references @xcite , @xcite and @xcite ) . in this paper , we consider the truncation question in an oriented graph .
let @xmath39 be the oriented graph defined as follows .
the vertex set is @xmath40 , where @xmath41 ; elements of @xmath42 will be denoted @xmath43 , where @xmath44 and @xmath45 .
the set @xmath46 of oriented bonds is @xmath47 again we are given a sequence @xmath5 satisfying and we assume each bond @xmath48 is open with probability @xmath49 independently of each other . again denoting by @xmath50 the probability measure corresponding to this percolation configuration and by @xmath51 the event that there exists an infinite open oriented path starting from @xmath52 , borel - cantelli gives @xmath53 . for each @xmath25
, we then consider the truncated sequence given in and the corresponding measure @xmath54 and ask the truncation question , that is , whether @xmath55 .
we prove : [ thm : main_cor ] for any @xmath56 , if the sequence @xmath21 satisfies , the truncation question has an affirmative answer for the graph @xmath57 .
moreover , @xmath58 this theorem is proved in the next section . in section [
s : other ] , we will treat a related question for the contact process and also for a different oriented graph .
we obtain theorem [ thm : main_cor ] as an immediate consequence of a stronger result , which we now describe .
we fix @xmath59 and consider @xmath60 defined as above , with vertex set @xmath61 and set of oriented bonds given in .
we take two sequences @xmath32 , @xmath62 and now prescribe that bonds of the form @xmath63 are open with probability @xmath64 and bonds of the form @xmath65 are open with probability @xmath66 . the truncated measure @xmath54
is obtained by truncating both sequences @xmath21 and @xmath67 at range @xmath24 .
[ prop : anis ] if @xmath5 satisfies and @xmath68 is not identically zero , then @xmath69 . by assumption
, we can fix @xmath70 such that @xmath71
. we will define certain _ bifurcation events _ which will imply that a point @xmath43 is connected to two new points @xmath72 and @xmath73 . for each @xmath74 , define the _ bifurcation event _
@xmath75 we have @xmath76 which can be made arbitrarily close to 1 by increasing @xmath24 , by . also note that @xmath77 finally , under @xmath50 and @xmath54 , @xmath78 and @xmath79 are independent and identically distributed as soon as either @xmath80 or @xmath81 .
the next step is to prove that , if @xmath24 is large enough , a certain projection of the @xmath24-truncated process dominates an oriented supercritical bernoulli percolation on @xmath82 .
define the following order in @xmath82 : given @xmath83 we say that @xmath84 if and only if @xmath85 or @xmath86 . given @xmath87 , we define the exterior boundary of @xmath88 as the set @xmath89 we define the vertex @xmath90 as _ red _ if and only if the following event occurs : @xmath91 . to avoid confusion ,
let us emphasize that , if a vertex in @xmath82 has coordinates @xmath92 , then this vertex is defined as red through an event in the original lattice @xmath93 ; this event involves a bifurcation with some starting point in the line @xmath94 . in particular , in figure 1 , one horizontal unit and one vertical unit in the lattice depicted on the left correspond respectively to @xmath95 units and @xmath96 units in the lattice on the right .
we will construct a red cluster dynamically , defining inductively two sequences @xmath97 and @xmath98 of subsets of @xmath82 .
set @xmath99 and @xmath100 .
assuming @xmath101 and @xmath102 have been defined for @xmath103 , we let @xmath104 now , if @xmath105 we stop our recursive definition .
otherwise we let @xmath106 be the minimal point of @xmath107 with respect to the order @xmath108 defined above , and continue the recursion .
regardless of whether or not the recursion ever ends , we let @xmath109 be the union of all sets @xmath110 that have been defined .
it follows from that @xmath111 .
now , observe that @xmath112 this implies that @xmath109 stochastically dominates the cluster of the origin in bernoulli oriented site percolation on @xmath82 with parameter @xmath113 ( see lemma 1 of @xcite ) .
as noted earlier , @xmath113 can be made arbitrarily close to 1 ; this proves that @xmath114 .
here we will give a counterpart of theorem [ thm : main_cor ] for the contact process obtained from truncating an infinite set of rates .
let us define precisely the model that we have in mind .
we are given a sequence of non - negative real numbers , @xmath115 .
we take a family of independent poisson point processes on @xmath116 : * a process @xmath117 of rate 1 for each @xmath44 ; * a process @xmath118 of rate @xmath119 for each ordered pair @xmath120 with @xmath44 and @xmath121 with @xmath122 and @xmath123 .
we view each of these processes as a random discrete subset of @xmath116 and write , for @xmath124 , @xmath125 } = d^{\vec{x } } \cap [ a , b]$ ] and @xmath126 } = b^{(\vec{x},\vec{y } ) } \cap [ a , b]$ ] .
fix @xmath127 . given @xmath128 and @xmath129
, we say @xmath130 and @xmath131 are @xmath24-connected , and write @xmath132 , if there exists a function @xmath133 \to \mathbb{z}^d$ ] that is right - continuous , constant between jumps and satisfies : @xmath134 , & \gamma(r ) \notin d^{\gamma(r)},\\ & r \in b^{(\gamma(r-),\gamma(r ) ) } \text { if } \gamma(r ) \neq \gamma(r-),\\ & |\gamma(r ) - \gamma(r-)| { \leqslant}k .\end{array } \end{aligned}\ ] ] we then define @xmath135 @xmath136 is then a markov process on the space @xmath137 for which the configuration that is identically equal to 0 ( denoted here by @xmath138 ) is absorbing . in case
@xmath139 only for @xmath140 , @xmath141 is the contact process of harris ( @xcite ) . for all @xmath142 , if @xmath143 , then @xmath144 it is enough to prove the case @xmath34 .
fix @xmath145 and @xmath146 .
let @xmath147 , for @xmath148 .
fix @xmath149 such that @xmath150 . for @xmath151 and @xmath45 ,
let @xmath152 be the event @xmath153 } = \varnothing\}\cap \bigcup_{a \in \mathbb{z } } \left\{\begin{array}{l } d^{\vec{x } + a \vec{e}_1}_{[t_n , t_{n+1 } ] } = d^{\vec{x } + a \vec{e}_1 + b \vec{e}_2}_{[t_n , t_{n+1 } ] } = \varnothing,\\[.4 cm ] b^{(\vec{x } , \vec{x } + a \vec{e}_1)}_{[t_n , t_n + \delta/2]}\neq \varnothing,\ ; b^{(\vec{x}+a\vec{e}_1 , \vec{x } + a \vec{e}_1 + b\vec{e}_2)}_{[t_n+\delta/2 , t_{n+1 } ] } \neq \varnothing \end{array } \right\}.\ ] ] then , @xmath154 by first taking @xmath155 small and then taking @xmath24 large , the probability of these events can be made arbitrarily close to 1 .
moreover , @xmath156 the proof is then completed with a comparison with oriented percolation almost identical to the one that established proposition [ prop : anis ] . in this section
we consider a graph @xmath157 .
once more , the vertex set is @xmath158 .
the set of bonds @xmath159 consists of two disjoint subsets ; one of them , denoted @xmath160 , only contains oriented bonds , and the other , @xmath161 , only unoriented bonds .
these subsets are given by @xmath162 that is , we are considering the hypercubic lattice where there are nearest neighbour , oriented bonds along the vertical direction and long range , unoriented bonds parallel to all other coordinate axes .
we consider an anisotropic oriented bernoulli percolation on this graph . given @xmath163 and a sequence @xmath5 in the interval @xmath6
$ ] , each bond @xmath164 is open with probability @xmath165 or @xmath8 , if @xmath166 or @xmath167 , respectively
. given two vertices @xmath168 and @xmath169 with @xmath170 , we say that @xmath43 and @xmath171 are connected if there exists a path @xmath172 such that @xmath173 or ( @xmath174 and @xmath175 ) for all @xmath176 , and the bonds @xmath177 are open for all @xmath176 .
that is , all allowed paths use vertical bonds only in the upward direction .
we use the notation @xmath178 to denote the set of configurations in which there is an infinite open path starting at @xmath52 .
we use also the notations @xmath50 and @xmath54 to denote the non - truncated and the truncated ( in the range @xmath24 ) probability measures , respectively . [ hex ] for any @xmath142 , any @xmath179 and any sequence @xmath5 such that @xmath180
, we have @xmath181 .
a weaker result was proven in @xcite ( see theorem 6 therein ) in the context of non - oriented and isotropic percolation .
the proof of theorem [ hex ] is inspirated by the proof thereof ( @xcite ) .
it is sufficient to prove the theorem for @xmath59 .
let @xmath182 be the function satisfying @xmath183 define the events @xmath184 clearly , @xmath185 for all @xmath186 .
also note that , if @xmath187 , then the line that contains @xmath188 and @xmath189 does not share any bonds of @xmath161 with the line that contains @xmath190 and @xmath191 .
hence , the events @xmath192 defined above are independent .
moreover , we have @xmath193 ( a proof of this can be found in the first few lines of the proof of theorem 6 in @xcite ) .
now , fix @xmath194 and @xmath145 .
let @xmath195 be an integer satisfying @xmath196 . then , using , choose @xmath197 such that @xmath198
. then let @xmath199 for each @xmath200 , let @xmath201 be the indicator function of the event @xmath202 @xmath203 then , the elements of the sequence of random variables @xmath204 are independent and , by the choice of @xmath195 , each of them is equal to 1 with probability @xmath205 .
now note that an infinite sequence @xmath206 such that @xmath207 , @xmath208 and @xmath209 for each @xmath210 necessarily corresponds to an infinite open path in @xmath211 .
moreover , the probability of existence of such a sequence can be taken arbitrarily close to 1 since @xmath155 is arbitrary .
this work was done during b.n.b.l.s sabbatical stay at impa ; he would like to thank rijksuniversiteit groningen and impa for their hospitality .
the research of b.n.b.l . was supported in part by cnpq and fapemig ( programa pesquisador mineiro ) . | we consider different problems within the general theme of long - range percolation on oriented graphs . our aim is to settle the so - called truncation question , described as follows .
we are given probabilities that certain long - range oriented bonds are open ; assuming that the sum of these probabilities is infinite , we ask if the probability of percolation is positive when we truncate the graph , disallowing bonds of range above a possibly large but finite threshold .
we give some conditions in which the answer is affirmative .
we also translate some of our results on oriented percolation to the context of a long - range contact process .
keywords : contact processes ; oriented percolation ; long - range percolation + msc numbers : 60k35 , 82b43 |
j/@xmath0 suppression in relativistic heavy ion collisions was suggested as a potential signal of the quark - gluon plasma @xcite . in order to study the j/@xmath0 suppression quantitatively ,
we need to understand how the j/@xmath0 mesons are produced , and the reliability of qcd calculations for j/@xmath0 production . in order to produce a j/@xmath0 meson in hadronic collisions ,
the energy exchanges in the collisions have to be larger than the j/@xmath0 mass ( @xmath2 ) or the invariant mass of the produced quark pair ( @xmath3 gev ) .
the @xmath4 pairs should be produced at a distance scale @xmath5 fm .
since this is much smaller than the size of a j/@xmath0 and the energy exchanges are much larger than the nonperturbative momentum scales in the j/@xmath0 wave functions , the j/@xmath0 is unlikely to be formed at the collision point . instead
, the j/@xmath0 meson should be formed after some resonance interactions ( or coherent soft gluon interactions between the charm and anticharm quarks ) .
thus , the transformation from a pre - j/@xmath0 @xmath4 pair to a physical j/@xmath0 occurs over several fermi @xcite . during this transformation , the produced @xmath4 pairs can also radiate gluons and have interactions with spectators .
due to gluon radiation and the spectator interactions , it is possible for the @xmath4 states , even with different quantum numbers , to become a j/@xmath0 meson .
however , because of the large energy exchange , the spectators roles in the hard collisions are suppressed by an extra factor @xmath6 $ ] with the medium density @xmath7 and medium length @xmath8 .
if the transverse momentum @xmath9 is large enough , the spectator interactions are strongly suppressed , and the cross sections for hadronic j/@xmath0 production can be factorized @xcite , @xmath10 where @xmath11 run over all parton flavors and the @xmath12 are the normal parton distributions .
the @xmath13 represents the cross section for two partons to produce a j/@xmath0 . without the nuclear medium
, @xmath8 is of the order of the hadron radius ( @xmath14 fm ) . because the production of the pre - j/@xmath0 @xmath4 states and the transformation to the j/@xmath0 take place at very different distance scales , quantum interference between these two stages is suppressed by the ratio of these two distance scales .
therefore , the partonic cross sections for the j/@xmath0 production in eq .
( [ eq1 ] ) can be further factorized into two stages : ( 1 ) short - distance production of the pre - j/@xmath0 partonic states , and ( 2 ) long - distance transition from the partonic states to the physical j/@xmath0 mesons @xcite , @xmath15 } \int \frac{dz}{z^2}\int dm_{c\bar{c}}^2 \left ( \frac{d\hat{\sigma}_{ab\rightarrow [ c\bar{c } ] } } { dm_{c\bar{c}}^2 dy d^2p_{c\bar{c}_t } } \right ) d_{[c\bar{c}]\rightarrow{\rm j/}\psi}(z;m_{c\bar{c}}^2 ) \label{eq2}\ ] ] where @xmath16}$ ] sums over all @xmath4 states , @xmath17 and latexmath:[$m_{c\bar{c}}^2 \ll transition probabilities from pre - j/@xmath0 @xmath4 states to j/@xmath0 mesons , and are defined by matrix elements of non - local operators @xcite .
the differences between our approach and other models of j/@xmath0 production can be cast as differences in @xmath19}$ ] . in qcd perturbation theory ,
the @xmath19}$ ] include all soft resonance interactions and gluon radiations .
similar to the parton - to - hadron fragmentation functions , the @xmath19}$ ] depend on nonperturbative physics .
thus , the predictive power of the factorized formula relies on the universality of these transition distributions @xcite .
the non - relativistic qcd ( nrqcd ) model @xcite and the color evaporation ( ce ) model @xcite correspond to two diverse assumptions for the @xmath19}$ ] .
the nrqcd model assumes that the @xmath19}$ ] are steeply falling functions of the invariant masses of the states , and sensitive to the pairs quantum numbers ( such as color and spin ) .
we can represent the leading contributions of this model by expanding the partonic cross section @xmath20}$ ] in eq .
( [ eq2 ] ) at @xmath21 and @xmath22 . in the nrqcd model ,
the produced @xmath4 pairs with large @xmath23 have very small probabilities to become j/@xmath0 mesons . on the other hand
, the ce model assumes that the @xmath19}$ ] are independent of the pairs invariant masses ( provided they are below the open charm threshold ) , and are not sensitive to the pairs quantum numbers .
we can represent the ce model by taking @xmath19}=f_{{\rm j/}\psi } \delta(1-z ) \theta(4m_d^2-m_{c\bar{c}}^2)$ ] in eq .
( [ eq2 ] ) with a fitting constant @xmath24 independent of the @xmath4 pairs quantum numbers .
since the nrqcd and the ce models are both consistent with the existing data on _ inclusive _ j/@xmath0 production , the new qcd factorized formula should also be consistent with the data @xcite . however , both models have difficulties explaining the recent cdf data from fermilab on j/@xmath0 polarization at large transverse momentum @xcite .
-0.35 in at large transverse momentum , j/@xmath0 are produced mainly from the fragmentation contribution , as shown in fig . 1 , where a virtual gluon is produced and decays into a @xmath4 pair which transmutes into a physical j/@xmath0 .
the polarization of the produced j/@xmath0 depends on the polarization of the @xmath4 pair and the details of the transition distributions @xmath19}$ ] . in order to use the polarization measurements to discern between the production models ,
we need independent tests of the reliability of the qcd formalism for calculating the @xmath4 polarization at large transverse momentum . as illustrated in fig . 1 , the low mass drell - yan lepton - pair s angular distributions at large transverse momentum have a lot in common with the @xmath4 polarization @xcite .
if the qcd calculations are consistent with the virtual photon polarization in drell - yan massive lepton - pair production , we can conclude that the qcd formalism for calculating the polarization of the @xmath4 at large transverse momentum is reliable , and therefore , measurements of j/@xmath0 polarization provide decisive tests for models of j/@xmath0 production .
in addition to polarization measurements , the nuclear medium dependence of j/@xmath0 production is sensitive to the j/@xmath0 formation process @xcite .
if the j/@xmath0 mesons are formed immediately after the production of the @xmath4 pair , the medium dependence is dominated by the hadronic scattering between a color - singlet j/@xmath0 and the nuclear matter . on the other hand ,
if the j/@xmath0 mesons are formed much later , the observed medium dependence is a consequence of the dynamical interactions between the pre - j/@xmath0 @xmath4 states and the nuclear medium .
therefore , the nuclear dependence can help select the correct model for j/@xmath0 production . in proton - nucleus or nucleus - nucleus collisions ,
the produced @xmath4 pairs are likely to interact with the nuclear medium before they exit .
if we assume each interaction between the @xmath4 pair and the nuclear medium is about the _ same _ and can be treated _ independently _ , we naturally derive the glauber formula : @xmath25 $ ] with effective absorption cross section @xmath26 to change a @xmath4 pair to a pair of open charms , and effective length @xmath27 of the medium in a+b collisions .
the characteristic feature of the glauber formula is a straight line on a semi - log plot of @xmath28 vs. @xmath27 , which is not consistent with the strong j/@xmath0 suppression observed in pb - pb collisions @xcite . our model of j/@xmath0 suppression
is based on the following points : ( 1 ) it is the @xmath4 pair , not the j/@xmath0 meson , that interacts through most of the nuclear medium ; ( 2 ) multiple interactions of the @xmath29 and/or @xmath30 with soft gluons in the medium increase the relative momentum between the @xmath29 and @xmath30 ; and ( 3 ) as the relative momentum increases , the phase space to form a j/@xmath0 meson decreases @xcite . since multiple soft - scatterings lead to a larger relative momentum between the @xmath29 and @xmath30 , the effective @xmath26 of the pair increases as the @xmath4 passes through the medium leading to a stronger suppression than given by the simple glauber formula which applies to single particle propagation @xcite .
integrating over j/@xmath0 s momentum in eq .
( [ eq1 ] ) , we obtain a factorized expression for the j/@xmath0 total cross section @xmath31 } \int dx \phi_{a / a}(x)\int dx ' \phi_{b / b}(x ' ) \int dm_{c\bar{c}}^2 \frac{d\hat{\sigma}_{ab\rightarrow [ c\bar{c } ] } } { dm_{c\bar{c}}^2 } f_{[c\bar{c}]\rightarrow{\rm j/}\psi}(m_{c\bar{c}}^2 ) \label{eq3}\ ] ] where @xmath32\rightarrow{\rm j/}\psi}$ ] represents an inclusive transition probability for a @xmath4 pair of invariant mass @xmath23 to become a j/@xmath0 meson . without the large transverse momentum , the corrections to eq .
( [ eq3 ] ) from the spectator interactions , which are suppressed by a factor @xmath33 , become more important because of the large medium length @xmath27 . just like the random walk of two particles ,
the soft spectator interactions effectively increase the invariant masses of the @xmath4 pairs , and therefore , reduce the phase space to form j/@xmath0 mesons .
we calculate the j/@xmath0 suppression in the nuclear medium by including the multiple spectator interactions to the pre - j/@xmath0 @xmath4 pairs , and plot our results in figs . 2 and 3 .
the data are from ref .
the dashed lines are predictions of the glauber formula , and the solid lines are the results of our calculations . with only one parameter @xmath34 , which is defined as average gain of invariant mass per unit medium length and is fixed by the total suppression in fig . 2 , our calculations of the j/@xmath0 suppression are consistent with all existing data , except a couple of points at the highest @xmath1 bins of the new na50 data @xcite .
we conclude that the nrqcd model and the ce model are closely connected to the new qcd factorization formula for j/@xmath0 production .
the measurements of j/@xmath0 polarization could provide a decisive test of the production mechanism .
our model of j/@xmath0 suppression is consistent with nearly all existing data on j/@xmath0 total cross section . | in terms of a new qcd factorization formula for j/@xmath0 production , we calculate the j/@xmath0 suppression in nuclear collisions by including the multiple scattering between the pre - j/@xmath0 partonic states and the nuclear medium .
we find agreement with all data on j/@xmath0 suppression in hadron - nucleus and nucleus - nucleus collisions , except a couple of points ( the `` second drop '' ) at the highest @xmath1 bins of the new na50 data . |
as well known , the evidence for atmospheric @xmath0 disappearance observed in the super - kamiokande ( sk ) experiment @xcite can be explained by two - family oscillations with large mixing in the @xmath2 channel @xcite .
when this solution , known as twofold maximal mixing , is assumed , the relevant flavor oscillation probabilities @xmath4 read : @xmath5 where @xmath6 and @xmath7 are the neutrino pathlength and energy , respectively , and the experimentally inferred value of the neutrino squared mass difference is @xcite @xmath8 within a factor of about two .
oscillations in the @xmath2 channel are not affected by matter effects , and are consistent with the negative results of the chooz @xmath1 disappearance experiment @xcite , which requires @xmath9 for @xmath10 ev@xmath11 .
conversely , two - family oscillations in the @xmath3 channel are heavily affected by matter effects , do not provide a good fit to sk data @xcite , and are independently excluded by chooz @xcite .
going beyond pure @xmath2 oscillations , one should consider the possibility that the flavor eigenstates @xmath12 are linear combinations of three states @xmath13 with definite masses @xmath14 through a unitary matrix @xmath15 : @xmath16 where @xmath17 are the mixing angles in the standard parametrization , and a possible cp violating phase has been neglected .
the parameter @xmath18 should then be taken equal to one of the following squared mass differences : @xmath19 assuming that @xmath20 .
different choices for @xmath18 have different implications for three - flavor scenarios .
in particular , if @xmath18 is taken within chooz bounds ( i.e. , @xmath21 ev@xmath11 ) , then large @xmath0 disappearance in sk can be generated only by dominant @xmath2 oscillations , while @xmath3 transitions must be small ( or absent ) .
this is the scenario favored by the quantitative analyses @xcite . on the other hand , for @xmath22 ev@xmath11 the chooz bounds are not operative and , in principle , sizable @xmath3 oscillations can occur .
even though it seems difficult to reconcile sk data with large @xmath1 mixing and relatively small @xmath18 @xcite , nevertheless , this possibility has been recently claimed in some phenomenological models , including : a ) `` threefold maximal mixing '' of solar and atmospheric neutrinos @xcite ; b ) three - flavor fits to solar , atmospheric , and accelerator evidence for neutrino oscillations @xcite .
such models with large @xmath1 mixing are representative of the two possible identifications for @xmath18 , which is taken equal to @xmath23 in model ( a ) and to @xmath24 in model ( b ) . since both models
( a ) and ( b ) have received considerable and continued attention in the neutrino physics literature , it seems useful to assess clearly their phenomenological status through a quantitative study of the super - kamiokande observations . in this work , we perform such an analysis and show that model ( a ) is disfavored ( although not yet excluded ) , while model ( b ) is definitely ruled out .
we also try to trace the origin of different claims by other authors .
threefold maximal mixing is defined by a democratic mixing matrix : @xmath25 @xmath26 , so that all oscillation channels are open .
the parameters @xmath23 and @xmath24 are assumed to drive atmospheric and solar neutrino oscillations , respectively , so that @xmath20 can be assumed .
the phenomenological implications of such model have been intensively studied by harrison , perkins , and scott ( see @xcite and references therein ) . as far as atmospheric neutrinos are concerned
, one can take @xmath27 and @xmath28 ( one mass scale dominance @xcite ) .
since the mixing matrix elements ( squared ) are fixed to the common value 1/3 , the only parameter probed by sk data is @xmath23 . in the following ,
we discuss how matter effects and fits to the sk and chooz data constrain the value of @xmath23 for threefold maximal mixing , and distinguish this scenario from twofold maximal mixing .
the three - flavor oscillation probabilities for the case of @xmath29 and generic mixing can be found , e.g. , in appendix c of ref .
the specific case of threefold maximal mixing , obtained by taking @xmath30 and @xmath31 , leads to the well - known results for vacuum oscillations : @xmath32 where @xmath33 . in the threefold maximal mixing scenario ( as well as in any model with large @xmath1 mixing ) atmospheric neutrino oscillations below the horizon
are heavily affected by earth matter effects @xcite , as emphasized , e.g. , in @xcite , and recently realized also in @xcite .
since the majority of upgoing leptons in sk are produced by neutrinos traversing only the earth mantle , the bulk of matter effects can be understood by assuming a constant electron density ( @xmath34 mol/@xmath35 ) along the neutrino trajectory . in this case , the flavor oscillation probabilities @xmath36 depend on the ( constant ) matter - induced squared mass term @xmath37 : @xmath38\ , \ ] ] and can be computed analytically .
explicit expressions for @xmath39 can be found in @xcite .
two limit cases are particularly illuminating for understanding matter effects in threefold maximal mixing : low energy events ( @xmath40 ) and high energy events ( @xmath41 ) .
the regime @xmath40 corresponds ( for @xmath42 ev@xmath11 ) to @xmath43 gev , and is thus relevant for the so - called sub - gev ( sg ) atmospheric neutrino events . in this limit
it turns out that @xcite @xmath44 where the term @xmath45 , for threefold maximal mixing , is given by : @xmath46 in the above equations , @xmath47 is the neutrino zenith angle ( @xmath48 ) , and the accidental equality @xmath49 for @xmath50 mol/@xmath35 has been used .
it thus appears that , in the sub - gev sample , no significant matter effects can be expected for electron events , while the muon event rate should be further suppressed via the energy - independent term @xmath51 .
this term is generated by the effective splitting in matter of the two vacuum - degenerate eigenstates @xmath52 @xcite .
the regime @xmath41 corresponds ( for @xmath42 ev@xmath11 ) to @xmath53 gev , and is thus relevant for the so - called upward through - going muon events ( up@xmath54 ) . in this limit
it turns out that @xmath55 namely , threefold maximal mixing in matter becomes equivalent to twofold maximal mixing in vacuum @xcite , but with an effective squared mass difference decreased by a factor 2/3 : @xmath56 @xcite .
the above approximations are useful for a qualitative understanding of oscillations in matter , but can not be applied in the regime of intermediate energies [ @xmath57 gev ) ] , typical of the so - called multi - gev ( mg ) events .
therefore , for numerical calculations we prefer to solve exactly , at any energy , the neutrino propagation equations along the earth density profile .
details of our computation technique can be found in our previous works @xcite ; see , in particular , @xcite for sk observables .
representative results of matter effects for threefold maximal mixing are shown in fig . 1 .
figure 1 shows the sk distributions of sub - gev electrons and muons ( sg@xmath58 and sg@xmath54 ) , of multi - gev electrons and muons ( mg@xmath58 and mg@xmath54 ) and of upward - going muons ( up@xmath54 ) , as a function of the lepton zenith angle @xmath59 , for @xmath60 ev@xmath11 . in each bin ,
the lepton rates are normalized to their expectations in the absence of oscillations .
the sk data ( dots with error bars ) refer to 45 kty @xcite .
the solid histograms ( blue in color ) represent the predictions of threefold maximal mixing including matter effects ; the dashed histograms ( red in color ) are obtained by assuming pure vacuum oscillations .
as expected from the previous discussion , matter effects are negligible for the sg@xmath58 distribution , while they generate an additional suppression for the sg@xmath54 distribution .
the up@xmath54 distribution is less `` tilted '' in the presence of matter oscillations , as a result of the smaller effective squared mass at high energy : @xmath61 . at `` intermediate energies '' ( mg events ) , both the @xmath58 and @xmath54 distributions are affected by oscillations in matter .
it can be seen that matter effects help the fit to sg and mg data , but slightly worsen the fit to up@xmath54 data , as compared to pure vacuum oscillations .
it has recently been claimed that , including matter effects , threefold maximal mixing with @xmath42 represents a good fit to all the sk data @xcite .
although matter effects certainly help to fit the data in such model , the statements in @xcite seem too optimistic as compared with our detailed analysis of either 33 kty @xcite or 45 kty @xcite sk data , which disfavors scenarios with large values of @xmath62 , including threefold maximal mixing ( @xmath63 ) . in this section ,
we clarify and further corroborate our previous results , and argue about the differences with the more optimistic claims in @xcite .
figure 2 shows the sk distributions for threefold maximal mixing ( @xmath60 ev@xmath11 , matter effects included ) , as compared with the case of twofold maximal mixing @xmath2 ( at @xmath64 and @xmath65 ev@xmath11 ) .
the differences among @xmath66 and @xmath67 distributions are relatively small for sg@xmath58 , sg@xmath54 , and mg@xmath58 events , while they are more significant for mg@xmath54 and up@xmath54 events .
let us thus focus on the latter two samples .
for mg@xmath54 , threefold maximal mixing overestimates the bin rate around the horizon @xmath68)$ ] , as compared to twofold maximal mixing .
this effect has two components : ( i ) the relatively low value of @xmath23 implied by chooz for @xmath67 maximal mixing ( @xmath69 ev@xmath11 ) , with respect to the best - fit value for @xmath66 maximal mixing @xmath70 ev@xmath11 ; and ( ii ) the further `` matter '' suppression of @xmath23 by a factor 2/3 for the highest - energy part of the mg@xmath54 sample ( partially contained events ) .
both components tend to increase the typical oscillation wavelength for @xmath67 maximal mixing ( as compared to the @xmath66 case ) , so that longer pathlengths are required to get effective muon suppression .
data from sk rapidly drop at the horizon , and show no evidence ( within errors ) for a wavelength significantly longer than for the pure @xmath66 best - fit case at @xmath71 ev@xmath11 ; see fig . 3 in @xcite .
] concerning up@xmath54 s one faces similar problems .
threefold maximal mixing is unable to fit the slope suggested by the data pattern , both because @xmath23 is necessarily low , and because of the further effective 2/3 suppression of @xmath23 in matter . for up@xmath54 events ,
threefold maximal mixing with @xmath60 ev@xmath11 is phenomenologically equivalent to twofold maximal mixing with @xmath72 ev@xmath11a value which is definitely below the 99% c.l .
range allowed by up@xmath54 data alone ( see fig . 9 in @xcite ) .
summarizing , fig .
2 shows that @xmath66 and @xmath67 maximal mixing are discriminated mainly by mg@xmath54 and up@xmath54 data ; the low value of @xmath23 allowed by chooz for @xmath67 maximal mixing ( @xmath69 ev@xmath11 ) prevents a good fit to the mg@xmath54 rate around the horizon and to the slope of up@xmath54 s , in contrast to the very good fit provided with @xmath66 maximal mixing at @xmath42 ev@xmath11 .
figure 3 shows @xmath73-fits to all the sk data ( sg+mg+up ) for twofold and threefold maximal mixing , as a function of @xmath23 .
details of our statistical analysis can be found in @xcite .
the dot - dashed line ( black in color ) refers to the twofold maximal mixing , characterized by @xmath74 and @xmath75 ev@xmath11 at best fit .
the allowed range of @xmath23 is in very good agreement with the sk official analysis ( compare with fig . 10 in @xcite ) .
the dashed line ( red in color ) refers to threefold maximal mixing ( sk data only ) , which in principle could provide a good fit for @xmath76 ev@xmath11 , if such value were not excluded by chooz . adding the chooz constraint ( solid line , blue in color ) , the best fit is pushed to lower @xmath23 values and the value of @xmath77 increases up to @xmath78 .
although this is not high enough to be ruled out , it is definitely worse than in the two - flavor case since , as observed in fig . 2
, @xmath67 maximal mixing fails to fit the up@xmath54 and horizontal mg@xmath54 data
. the situation would be even worse if matter effects were hypothetically switched off ( dotted line , green in color ) .
the quantitative results of fig . 3 indicate that the two - flavor and three - flavor scenarios with maximal mixing differ by @xmath79 ; therefore , if @xmath67 mixing is left unconstrained , the minimum will fall close to the @xmath66 case with zero @xmath1 mixing , while @xmath67 cases with large @xmath1 mixing will be disfavored .
this is indeed the pattern found in the @xmath67 parameter space analysis of @xcite .
however , it should also be said that the @xmath67 maximal mixing fit to the sk data is not terribly bad , so it is wiser to wait for higher statistic in the mg@xmath54 and up@xmath54 samples , or for more stringent reactor bounds on @xmath23 , before ( dis)proving it definitely .
we conclude by examining possible reasons for the more optimistic results found in @xcite .
the work @xcite did not include up@xmath54 data , which contribute to worsen the fit for threefold maximal mixing . the work @xcite , moreover , considered only the up - down asymmetry for sg and mg data , thus excluding also the important information given by nearly horizontal mg@xmath54 .
the comparison with @xcite is more delicate since , in principle , a data set similar to ours is analyzed .
however , the lack of details and of explicit comparison with sk calculations in @xcite prevents us to trace the source of the differences . nevertheless , a comparison of our fig . 3 with the corresponding results in @xcite
( see their fig .
7 ) gives us some hints .
it is quite evident that , for the two - flavor oscillation case , the preferred range for @xmath23 in @xcite is biased towards low values , as compared with our range in fig . 3 and with the sk official analysis ( fig . 10 in @xcite ) .
this is surprising , since the authors of @xcite include also the old kamiokande data , which should rather pull the fit towards slightly higher @xmath23 values .
a bias towards low @xmath23 s artificially increases the chance that threefold maximal mixing can survive below chooz bounds . in conclusion ,
we find that threefold maximal mixing is significantly disfavored with respect to twofold maximal mixing @xmath80 .
the most discriminating data samples are mg@xmath54 and up@xmath54
. however , further reactor data , and atmospheric neutrino data in the intermediate and high energy range , are needed in order to ( dis)prove the model with higher confidence .
the super - kamiokande collaboration itself has now all the tools to investigate quantitatively generic three - flavor scenarios including matter effects @xcite , so as to settle definitely some differences existing among present independent phenomenological analyses .
the sources of evidence for neutrino oscillations coming from solar and atmospheric neutrinos , as well as from the liquid scintillator neutrino detector experiment ( lsnd ) @xcite suggest three widely different neutrino mass square differences .
therefore , they can not be reconciled in a single three - flavor scenario , unless some data are `` sacrificed '' @xcite .
one possibility @xcite , indicated as `` case 3c '' in table vi of ref .
@xcite , has been recently revived in several works @xcite which claim that it can provide a good fit to the sk atmospheric neutrino data , and also explain most of the solar neutrino deficit and the lsnd data . in this section ,
we briefly discuss such scenario , and show quantitatively that it is ruled out by the sk atmospheric @xmath81 data .
the starting hypothesis is that @xmath82 , so that : ( i ) concerning solar neutrinos , both @xmath24 and @xmath23 drive energy - averaged oscillations , which can explain the bulk of the solar neutrino deficit but not possible distortions in the energy spectrum ; ( ii ) for atmospheric neutrinos , @xmath24 drives the energy - dependent oscillations while @xmath23 gives an energy - averaged contribution ; and ( iii ) for lsnd , @xmath24 can be taken effectively equal to zero , and oscillations are driven by @xmath83 . in order to get the desired @xmath84 ,
the heavy state @xmath85 is taken close to @xmath86 ( @xmath87 ) , so that @xmath52 are basically linear combinations of @xmath88 . by choosing nearly maximal mixing between @xmath89 and @xmath90 ,
one then hopes to solve both the atmospheric neutrino anomaly and the solar neutrino deficit through dominant @xmath91 oscillations driven by @xmath92 .
the chooz results then constrain @xmath24 to range below @xmath93 ev@xmath11 .
the mixing matrix for such scenario is rather strongly constrained by the requirement to fit solar , atmospheric , and lsnd data .
indeed , the phenomenological matrices found in different works @xcite are similar to each other , with small differences which are not important for our discussion . for definiteness , we report the matrix @xmath15 found in the paper @xcite by thun and mckee : @xmath94 in the same work @xcite , the best - fit value of @xmath23 is 0.4 ev@xmath11 . such choice for the mass - mixing parameters will be dubbed `` the thun - mckee scenario . ''
figure 4 shows the expected zenith distributions of lepton events in sk for the thun - mckee scenario at @xmath95 ev@xmath11 , with and without matter effects .
matter effects play an important role in suppressing oscillations : when the @xmath88 mixing is ( nearly ) maximal in vacuum , it can only be smaller in matter . a similar pattern , in fact ,
is found for pure two - flavor @xmath3 oscillations ( see fig .
9 of @xcite ) .
it can be seen in fig . 4 that the thun - mckee scenario is ruled out by the sk data for several reasons : ( i ) the sg@xmath58 and mg@xmath58 upgoing event rates are overestimated ; ( ii ) the sg@xmath54 and mg@xmath54 up - down asymmetries are largely underestimated ; ( iii ) the predicted up@xmath54 rate is too low ; and ( iv ) matter effects play an important role in worsening the fit
. ratio of contained events .
] the situation is not improved by lowering the value of @xmath23 , as it can be seen in fig .
we find that the thun - mckee scenario , in any case , gives @xmath96 , and is thus definitely ruled out by the sk data _ alone_. in addition , the thun - mckee scenario does not provide a good fit to the current reactor and accelerator oscillation data , including lsnd , as already observed in @xcite . in fact , the quantitative three - flavor analysis of laboratory data performed in @xcite selects only two possible solutions at @xmath97 c.l . , namely , @xmath98 or @xmath99 , while @xmath100 is highly disfavored . in conclusion , we find that the thun - mckee scenario is not a viable explanation of the solar+atmospheric+lsnd data .
the sk data are sufficient to rule it out definitely .
the much more optimistic claims in @xcite are not substantiated by quantitative calculations of electron and muon event rates in sk .
we have made a detailed comparison of the sk atmospheric @xmath81 data with the predictions of two popular models characterized by large @xmath3 transitions : ( a ) threefold maximal mixing @xcite ; b ) three - flavor fits to solar , atmospheric , and lsnd data @xcite .
we have found that model ( a ) @xcite is disfavored ( although not yet excluded ) , while model ( b ) @xcite is definitely ruled out .
the origin of different claims by other authors has been elucidated .
glf and el thank the organizers of the neutrino oscillation workshop ( amsterdam ) and of the school of physics `` bruno pontecorvo '' ( capri ) , respectively , where preliminary results of this work were presented .
y. totsuka , in the proceedings of the 1998 nobel symposium `` particle physics and the universe , '' ( enkoping , sweden , 1998 ) , edited by l. bergstrom , p. carlson , and c. fransson ( physica scripta , 1999 , to appear ) report available at http://www-sk.icrr.u-tokyo.ac.jp/doc/sk/pub
. g.l .
fogli , e. lisi , a. marrone , and g. scioscia , report no .
hep - ph/9904465 , to appear in the proceedings of _
17th international workshop on weak interactions and neutrinos ( cape town , south africa , 1999 ) .
l. wolfenstein , phys .
d * 17 * , 2369 ( 1978 ) ; s.p .
mikheev and a.yu .
smirnov , yad .
* 42 * , 1441 ( 1985 ) [ sov .
j. nucl .
phys . * 42 * , 913 ( 1986 ) ] ; v. barger , k. whisnant , s. pakvasa , and r.j.n .
phillips , phys .
d * 22 * , 2718 ( 1980 ) . p.f .
harrison , d.h .
perkins , and w.g .
scott , phys .
b * 374 * , 111 ( 1996 ) .
bilenky and c. giunti , phys .
b * 380 * , 331 ( 1996 ) ; g.l .
fogli , e. lisi , and d. montanino , phys .
d * 54 * , 2048 ( 1996 ) . fig . 1 .
threefold maximal mixing scenario at @xmath60 ev@xmath11 . from left to right : zenith distributions of sub - gev ( sg ) electrons ( @xmath58 ) and muons @xmath101 , of multi - gev ( mg ) @xmath58 and @xmath54 , and of upward going muons ( up@xmath54 ) , normalized in each bin to the standard expectations ( no oscillation ) .
solid line : predictions including matter effects .
dashed lines : predictions without matter effects .
dots with error bars : super - kamiokande data . fig . 2 .
comparison of twofold ( 2@xmath81 ) and threefold ( 3@xmath81 ) maximal mixing predictions .
fit to twofold and threefold maximal mixing scenarios for variable @xmath23 .
thun - mckee @xmath67 mixing scenario @xcite at @xmath95 ev@xmath11 , with and without matter effects .
thun - mckee @xmath67 mixing scenarios at @xmath95 and @xmath102 ev@xmath11 ( matter effects included ) . | the evidence for @xmath0 disappearance in the super - kamiokande atmospheric neutrino experiment and the negative searches for @xmath1 disappearance in the chooz reactor experiment can be easily reconciled by assuming oscillations with large amplitude in the @xmath2 channel and small ( or null ) amplitude in the @xmath3 channel .
it has been claimed , however , that some oscillation scenarios with large @xmath3 mixing can also be constructed in agreement with the present data .
we investigate quantitatively two such scenarios : ( a ) threefold maximal mixing ; and ( b ) attempts to fit all sources of evidence for oscillations ( solar , atmospheric , and accelerator ) with three neutrinos . by using mainly super - kamiokande data ,
we show that the case ( a ) is disfavored , and that the case ( b ) is definitely ruled out . |
figures [ figallqso ] and [ figallqsosub ] tell the main story of this paper .
we urge the reader to look at these beautiful @xmath1 images before continuing with the text and the quantitative details .
we summarize in this paper the results of our analysis of @xmath1-wfpc2 observations of a representative sample of 20 of the most luminous ( m@xmath2 ) nearby ( @xmath3 ) quasars .
the goal of these observations was to help understand the quasar phenomenon by determining the environment in which quasars occur .
the main result of this paper is that there is not one type of environment , but instead a wide range of environments in which the most luminous quasars appear to be embedded .
the @xmath1 images also contain a number of extraordinary phenomena and some surprises , including : host galaxies that appear normal and show no evidence of strong gravitational interactions , very close galactic companions , host ellipticals for radio quiet quasars , spiral hosts with well developed arms and prominent h ii regions , galaxies caught in the act of merging , apparently faint galactic hosts , and very extended emission .
some partial results of this study have been reported previously : bahcall et al .
( 1994 , 1995 , 1995a , b , c , 1996 ) .
for those aspects of the work that depend upon the subtraction of a stellar point spread function ( psf ) , there are small , quantitative differences between the results described in this paper and previous results we have reported . in the previous work
, we used a stellar psf determined from a red standard star , f141 . in the present work ,
we have used stellar psfs that were obtained for four separate blue stars ( see discussion in section [ aftersub ] of the psfs constructed by krist and burrows 1996 ) .
the visual appearance of the hosts in the subtracted images is , in a few cases , significantly improved by using the psfs of the blue stars .
other @xmath1 studies of quasar imaging ( although many of the objects do not satisfy our luminosity criterion ) include hutchings & morris ( 1995 ) , hutchings et al . , ( 1994 ) , and disney et al .
( 1995 ) .
the subject of quasar environments has a long and distinguished history .
some representative papers that report on ground - based observations are : kristian ( 1973 ) , wyckoff et al .
( 1980 ) , wyckoff , wehinger , & gehren ( 1981 ) , tyson , baum , and kreidl ( 1982 ) hutchings et al .
( 1982 ) , gehren , et al .
( 1984 ) , heckman et al . ( 1984 ) , malkan ( 1984 ) , malkan , margon , & chanan ( 1984 ) , boroson , persson , & oke ( 1985 ) , smith et al . ( 1986 ) , hutchings ( 1987 ) , stockton & mackenty ( 1987 ) , yee ( 1987 ) , hutchings , janson , & neff ( 1989 ) , romanishin & hintzen ( 1989 ) , vron - cetty & woltjer ( 1990 ) , hutchings & neff ( 1992 ) , dunlop et al . ( 1993 ) , and mcleod & rieke ( 1994a , b ) .
the analyses described in these pioneering ground - based studies are made difficult because of atmospheric seeing ; the light from the bright central ( nuclear ) sources may be a few magnitudes brighter than the total emission from the host galaxies .
nevertheless , there is agreement ( within typically one magnitude or better ) between our @xmath1 observations and previous ground - based estimates of the total apparent magnitudes of the host emission , although @xmath1 reveals details not previously accessible and corrects some important conjectures that are not supported by the higher - resolution observations .
the paper is organized as follows . in section [ obssection ]
we describe the sample selection and observations ; in section [ subunproc ] we present the unprocessed data ; in section [ aftersub ] we describe the method of removing the light due to the quasar ( point spread function subtraction ) and present the data after a stellar psf is subtracted ; in section [ anasection ] we describe the methods of analysis of the data ; in section [ secsize ] we report on the measurements of the host galaxies ; in section [ companion ] we discuss the presence of companion galaxies ; in section [ groundsection ] we compare our measurements with results from some ground - based observations ; in section [ comments ] we comment on the environment and host galaxy of each quasar , and in section [ discuss ] we summarize and discuss our results .
we assume in this paper that @xmath4 and @xmath5 .
we describe in this section how the sample was selected ( [ subsample ] ) and how the observations were performed ( [ subobs ] ) .
a sample of 14 quasars was selected solely on the basis of luminosity ( @xmath6 ) , redshift ( @xmath7 ) , and galactic latitude ( @xmath8 ) .
all the quasars in the vron - cetty & vron ( 1991 ) catalog that satisfied the redshift , luminosity , and galactic latitude criteria were included in this sample ; no distinction was made on the basis of radio or other secondary properties .
one may choose to call this set of quasars a `` complete sample '' within the context of the vron - cetty & vron ( 1991 ) catalog .
the 14 quasars with @xmath9 have an average ( median ) absolute magnitude @xmath10 and an average redshift @xmath11 .
only 3 radio loud quasars are present in the original sample of 14 objects . in the vron - cetty and vron ( 1996 ) catalog , there are an additional 4 quasars with @xmath12 that satisfy our luminosity and galactic latitudes requirements ; all are radio quiet . by combining the time available from gto and go programs , an additional 6 quasars with redshift between @xmath13 were added to the original sample of 14 objects ; these additional objects satisfied the same luminosity and galactic latitude constraints as the original sample .
the additional objects contain 3 radio loud quasars ; the total sample of 20 contains 6 quasars that are classified as radio loud , i.e. with @xmath14 ( kellermann et al .
the additional 6 objects are slightly brighter in the optical on the average ( median ) , @xmath15 , and have slightly larger redshifts , @xmath16 .
the range of absolute magnitudes in the 6 added objects , @xmath17 , is included within the range of absolute magnitudes spanned by the original sample of 14 objects , i.e. , @xmath18 . in what follows
, we shall refer to the total sample of 20 quasars as a `` representative sample '' with @xmath19 and @xmath3 .
the average ( median ) absolute magnitude for the full sample is @xmath20 ; the average redshift @xmath21 .
nearly all ( 18 ) of the quasars have @xmath22 , but two are much brighter in the optical band : 3c 273 ( @xmath23 ) and he 1029@xmath24140 ( @xmath25 ) . a journal of the observations is given in table [ journal ] , which lists the following quantities for each object : the date observed , the longest exposure time of a single image , the detected number of @xmath26 in the sky , the quasar redshift , the distance in kiloparsecs that corresponds to an angular separation of 1 as seen from earth , the apparent @xmath27-magnitude ( from vron - cetty & vron 1996 ) , the absolute @xmath27-magnitude ( without k - correction ) , the radio properties ( an `` x '' identifies the radio loud quasars ) , and in the last column the existence of @xmath1 spectroscopy ( a `` k '' indicates that the @xmath1/fos observations were taken as part of the @xmath1 quasar absorption line key project , and some of the results are reported by bahcall et al .
( 1993 ) and jannuzi et al .
( 1997 ) ; an `` o '' indicates that other fos observations exist ) .
the quasars were observed with the wide field / planetary camera-2 ( wfpc2 ) through the f606w filter , which is similar to the @xmath27 bandpass but is slightly redder ; the mean wavelength and fwhm of the f606w system response are 5935 and 1500 , respectively .
the f606w filter was chosen because of its high throughput . in all of the quasars discussed here ,
redshifted @xmath28 and [ o iii ] are included in the bandpass . at a given angular radius ,
the scattered light in the wide field camera-2 ( wf ) is about five times less than the scattered light in the planetary camera ( see krist & burrows 1994 ) .
we chose to use the wf instead of the pc in the original formulation of this program because of the likelihood that the host galaxies would have low surface brightnesses that extended over areas large compared to the wf resolution ( @xmath29 or about 0.2 kpc ) .
the results reported here support the original choices , since the observed galaxy extensions are indeed large compared to the wf resolution .
the center of light of all quasars was placed within @xmath30 from the center of the wide - field camera ccd 3 ( wf3 ) .
three exposures were taken of each quasar .
the integration times for 14 objects were 1400 s , 500 s , and 200 s ; the exposures for the remaining six objects were 1100 s , 600 s , and 100 s. the size of wf3 is @xmath31 pixels ( exposed area @xmath32 pixels ) and its image scale is 00996 pixel@xmath33 .
we report measured f606w magnitudes on the @xmath1 photometric scale established by holtzman et al .
( 1995b ) .
the adopted photometric zero - point for 1 @xmath34 is 24.94 mag for the f606w filter . for further information about the wfpc2 ,
see burrows ( 1994 ) , trauger et al .
( 1994 ) , and holtzman et al .
( 1995a , b ) .
additional details of the observational procedures are given in paper i and paper ii .
the innermost regions of the quasar images are saturated in all of the longest exposures out to a radius @xmath35 , except for the the two optically brightest quasars in our sample , he 1029@xmath24140 and 3c 273 , in which cases the images were saturated out to @xmath36 and @xmath37 , respectively .
the number of saturated pixels in the central region of the quasar images were typically 30 pixels for the longest exposures .
in addition , @xmath38 saturated pixels were present due to the `` vertical bleeding '' .
for he 1029@xmath24140 and 3c 273 , the number of saturated pixels was approximately 105 and 190 , respectively , plus 90 and 390 from the vertical bleeding .
the initial data processing ( bias frame removal and flat - field calibration ) was performed at the space telescope science institute with their standard software package .
the individual images of each quasar were aligned to better than 0.3 pixel ; this made it easy to identify and eliminate cosmic ray events .
cosmic rays were identified by a pixel - by - pixel comparison of pairs of images ; the intensity of a pixel containing a cosmic ray was replaced by the scaled value of the intensity of the pixel in the other image .
the flat - field corrections were based upon pre - flight calibrations ; these calibrations remove the small - scale ( few pixel ) sensitivity variations . the typical signal and rms of the noise of the sky in the long exposures ( in detected photons per pixel ) are 157 and 14 , respectively .
the sky level corresponds to an average surface brightness of approximately @xmath39 mag arcsec@xmath40 .
the limiting surface brightness at which objects can be detected is typically between @xmath41 and @xmath42 ( f606w ) , cf .
table [ size ] .
we present in this section the images as received from stsci , without further processing except for the removal of cosmic rays .
figure [ figallqso ] shows a @xmath43 wf image of each of the 20 quasars , plus the image of a blue star ( upper left - hand portion ) .
the images displayed are the longest individual exposures we have .
many , but not all , of the quasars are noticeably non - stellar , and host galaxies are visible on the unprocessed @xmath1 images .
the exposure time of the star image shown in the top panel of figure [ figallqso ] is 20 s , which yields a total number of counts similar to that obtained in the quasar images .
the star is mmj 6490 in the m67 cluster .
its apparent @xmath27 magnitude is 10.99 and its @xmath44 ( montgomery , marschall & janes 1993 ) .
some features of host galaxies are obvious in figure [ figallqso ] .
normal spiral galaxies , with prominent h ii regions , envelop pg 0052@xmath45251 and pg 1402@xmath45261 .
host elliptical galaxies are clearly seen in the images of phl 909 , he 1029@xmath24140 , pg 1116@xmath45215 , and 3c 273 .
there are three obvious cases of current gravitational interaction : 0316@xmath24346 , pg 1012@xmath45008 , and pks 2349@xmath24014 .
some of the host galaxies are seen even in our shortest exposures ( 200 s , see bahcall et al .
1996 for short exposure images of pg 0052@xmath45251 and phl 909 ) . in some cases ,
like nab 0205@xmath4502 , pg 0953@xmath45414 , and pg 1307@xmath45085 , it is difficult to distinguish the quasar image from the star even in our longest exposures .
the major challenge in the analysis of these @xmath1 data is the removal of the light produced by the quasar , which is presumed to be a point source .
we have adopted an empirical approach .
we have used images of stars observed at the same location on the detector as the quasars to determine the point - spread function , psf , of the stellar quasar .
for about half of the quasars considered here , the principal observational results concerning the host environment can be obtained without the psf subtraction .
examples of exceptions are the hosts of nab 0205@xmath4502 , pg 0953 + 414 , pg 1444@xmath45407 , and 3c 323.1 , which are more apparent in figure [ figallqsosub ] than in figure [ figallqso ] .
close companions for pks 1302@xmath24102 , pks 2135@xmath24147 , and pks 2349@xmath24014 are also more clearly visible after the subtraction of a stellar psf .
the stellar psf was measured in cycle 5 ( hst program 5849 ) by obtaining a set of 13 images for each of four blue stars in the m67 cluster .
the calibration stars ( and their @xmath46 colors ) are : mmj 6481(@xmath47 ) , mmj 6490(@xmath48 ) , mmj 6504(@xmath49 ) , and mmj 6511(@xmath50 ) .
the apparent magnitudes range from @xmath51 to @xmath52 . for each star
, a series of four images were used by krist and burrows ( 1996 ) to produce a psf that samples the full dynamic range of the star image and covers the saturation range found in the quasar images . in these calibration images ,
the radius of the saturated region varies from @xmath53 to @xmath54 ; the exposure times range from @xmath55 s to @xmath56 s. the psf data are publicly available at http://www.sns.ias.edu/jnb .
there is detailed documentation regarding the psf data and their use at this site ( go to hst images ) by j. krist and c. burrows describing how the psfs were constructed .
whenever psf subtractions were required , we used all four psfs determined by krist and burrows . for each case , we selected the result that gave the cleanest subtraction ( i.e. , the fewest artifacts produced )
. we fit a stellar point - spread function to each quasar image and subtracted a multiple of the normalized psf to search for underlying diffuse light from hosts .
the best fit was determined by minimizing the differences between the quasar and the psf using a @xmath57-routine calculated in two distinct areas : azimuthal averages and narrow regions centered on the diffraction spikes .
the two methods gave essentially same results ( see paper ii ) , differing in inferred host galaxy magnitude by @xmath58 mag .
the quasar images with the psf subtracted presented in this paper were obtained by minimizing the @xmath57 in azimuthal averages .
we have estimated some of the likely systematic uncertainties in the subtraction process by subtracting a best - fit psf of the standard star , mmj 6504 , from the image of another standard star , mmj 6490 .
we have cycle 5 observations of mmj 6490 observed at the same position in wf3 as the sample quasars .
figure [ figallqsosub]a ( first panel ) shows the image of the best fit of mmj 6504 subtracted from mmj 6490 .
the psf subtraction , star from star , is very good , although some faint diffuse residuals are still present . if we scale the intensity of mmj 6490 so that the central brightness corresponds to the apparent magnitude of one of our typical quasars , i. e. , @xmath59 , then the residual `` nebulosity '' left over when the mmj 6504 psf is subtracted is about 21.0 mag .
figure [ figallqsosub ] shows the images for all the quasars in our sample after a best - fit stellar point - spread function ( psf ) was subtracted from the original images shown in figure [ figallqso ] .
table [ size ] lists the major diameter of the host galaxies in arcseconds and in kpc , the surface brightness of the isophote for which the size was measured , and a tentative morphological classification . for each quasar
, the magnitude of the host galaxy was calculated using three different methods , aperture photometry ( see [ subap ] ) , and one - dimensional and two - dimensional galaxy modeling ( see [ subrp ] and [ subtwo ] ) .
we determined the size of the quasar hosts by examining the psf subtracted images and measuring the faintest isophotes of the host galaxy .
the morphological classification was done by visual inspection of the images and following as closely as possible the classification given in _
the hubble atlas of galaxies _ by sandage ( 1961 ) . in a number of cases indicated by ( ? ) in table [ size ] , we have denoted as `` en '' featureless , smooth hosts that do not show any obvious discontinuities or morphological features .
in this section , we describe the different methods of analysis that we have used to determine the properties of the quasar hosts .
the measured quantities are presented in tables [ size][veron ] and discussed in [ secsize][companion ] .
we describe in
[ subap ] to [ subtwo ] different methods for estimating the magnitude of the host galaxies .
we discuss in [ subap ] our results for aperture photometry , in
[ subrp ] the radial profile fits , and in [ subtwo ] the two - dimensional fits .
some of the host environments are complex , including merging or tidally interacting galaxies and close companions .
the smooth de vaucouleurs or exponential disk models used in [ subrp ] and [ subtwo ] are obviously not realistic descriptions of the light distribution for complex environments . in order to provide a common basis of comparison with ground based observations ,
we provide the results of fits with smooth models for all of the host environments , complex or not .
we performed aperture photometry in circular annuli centered on the quasars , after a best - fit psf was subtracted .
an inner radius of @xmath60 was used for all quasars , except for he 1029@xmath24140 and 3c 273 . in most observations ,
the region @xmath61 is contaminated by artifacts left by the psf subtraction .
the saturated areas in the images of he 1029@xmath24140 and 3c 273 are larger ; for those two cases , the inner radii used were 1.5 and 2.0 , respectively . the outer radii chosen in general represent how far we could see the host galaxy ( see table [ result ] ) .
the aperture magnitudes calculated using an inner radius @xmath60 are typically @xmath62 mag fainter than the total magnitude for the galaxy that was estimated by fitting a model ( see [ subrp][subtwo ] ) to the measured surface brightness .
a difference between aperture and model magnitudes is expected because the aperture magnitudes do not include the area within 1 of the quasar ; all of the models we considered have surface brightnesses that increase monotonically toward the center .
the one - dimensional azimuthally - averaged surface brightness profiles of the host galaxies were constructed from the @xmath1 data after subtraction of a best - fit stellar psf .
regions affected by saturation , diffraction spikes , or residual artifacts from the psf subtraction were not included in the azimuthally averaged profiles .
for each galaxy , we obtained a best - fitting exponential disk ( henceforth disk ) and a de vaucouleurs ( 1948 , henceforth gdv ) profile that fits the observed data in the region @xmath63 .
the total magnitudes obtained this way are systematically brighter than the ones obtained by aperture photometry ( [ subap ] ) which excludes the innermost , saturated regions of the profiles .
the @xmath1 imaging provides greater detail than has been available previously in ground - based images of luminous quasars .
traditionally , the properties of host galaxies have been determined in ground - based studies by making model fits to azimuthally - averaged radial profiles ( see [ subrp ] ) . to take advantage of the @xmath1 resolution ,
we have developed software to fit a two - dimensional model to the psf - subtracted quasar images . for each quasar
, we fit an analytic galaxy model ( exponential disk or de vaucouleurs profile ) to the data and calculated the @xmath57 .
the area used for the fit was approximately an annular region , centered on the quasar , that excluded the central area ( @xmath64 ) , and the remnants of the diffraction spikes or other artifacts clearly due to improper psf subtraction .
we fit four parameters : the ( x , y ) pixel position of the center , the total number of counts , and the radius ( effective radius or scale length ) in the galaxy model .
we begin the iteration by giving the software the position of the quasar nucleus , and calculate only the total number of counts .
the number of counts found in this step is entered as the initial guess for the galaxy brightness , and the program then fits the counts and radius , keeping the position fixed . finally , we supply the software with the previously calculated counts and radius and ask the software to fit all four parameters .
we tested the software for the two - dimensional fits in different ways .
initially we created two model galaxies : a disk and an elliptical galaxy .
we checked how well the two - dimensional software reproduced the position of the center of the model galaxies , the scale length or effective radius , and the total number of counts .
we made extensive checks by varying the input parameters and the order in which the parameters were fit .
the best results were achieved when we followed the three step iterative process in the order described above .
we fit each model galaxy with an exponential disk and a de vaucouleurs model - the smallest @xmath65 residuals were obtained when the disk galaxy was fit by an exponential disk model , and the elliptical was fit by a de vaucouleurs model . for the model galaxies ,
the discrepancy in position and size were @xmath66 pixel .
the discrepancy in the total number of counts was @xmath67 .
the accuracy achieved in the simulations is greater than can be achieved with the real data , because of the imperfect subtraction of the psf for the @xmath1 images .
the host galaxies of pg 0052@xmath45251 and phl 909 were also used to test the two - dimensional software .
we compared the software output position with our measured position for the center of the galaxy ( agreement better than 0.2 ) , compared the total number of counts with the value we estimated from aperture photometry ( agreement turned out to be 0.1 mag for pg 0052 + 251 , and 0.5 mag for phl 909 ) , and verified that the output scale length or effective radius were plausible .
in this section , we report on the measurements of the magnitude and position of the host galaxies . in table
[ result ] we list the magnitudes for the host galaxies measured using aperture photometry and for a few cases the surface brightness method .
the inner and outer radii used in performing aperture photometry are listed as well .
the outer radius indicates how far from the quasar the host galaxy was clearly visible .
we transformed the measured f606w aperture magnitudes to @xmath27 applying k - corrections calculated by fukugita et al .
given the redshift of the quasar and the morphological type of the host , we used the fukugita et al .
table 6 and figure 14b to obtain the ( @xmath68 ) color . for the cases in which we are uncertain about the morphological type of the host
, we assumed an average of the ( @xmath69 ) color for ellipticals and sab galaxies .
table [ model ] lists the results for the one - dimensional and two - dimensional galaxy model fitting .
the two - dimensional models are somewhat fainter than the corresponding one dimensional fits .
specially , we find : @xmath70 the magnitudes obtained by model fits are brighter than those obtained by aperture photometry since the models include estimated contributions from the inner ( saturated ) regions of the images . on average @xmath71 as equation ( 3 ) shows , the magnitudes of the host galaxies estimated by fitting a gdv model to the azimuthal averaged radial profile of the residual light are on average 1.0 mag brighter than the magnitudes obtained from aperture photometry ; fitting an exponential disk model gives magnitudes that are on average 0.5 brighter than the results from aperture photometry .
for the two - dimensional model fits , we have on average @xmath72 table [ bestmodel ] lists our best - estimate absolute @xmath27 magnitudes . in computing the entries in table [ bestmodel ] , we used the k - correction of fukugita et al .
( 1995 ) that are listed in the next - to - last column of table [ result ] . in computing the absolute magnitudes
, we selected the best - fit model based on the morphology of the host galaxy , unless the morphology is uncertain , and for those cases we list the model that gives the smallest @xmath65 residuals . the results from the two - dimensional fits indicate that the host galaxies are , on average , centered on the quasar with : @xmath73 the host galaxies are typically centered within 400 pc of the location of the quasar point source .
if we eliminate from the comparison the most extreme examples of especially complex environments ( 0316@xmath24346 , pg 1012@xmath45008 , and pks 2349@xmath24014 ) the results are , @xmath74 the 2-d ( 1-d ) gdv model gives magnitudes for the host that are on average 0.6 mag ( 0.7 mag ) brighter than the exponential disk estimates .
inspection of the @xmath1 images of the quasar fields reveals a number of companion galaxies projected close to the quasars . the relatively long exposures ( 1100 or 1400 seconds ) , combined with the excellent angular resolution , allowed galaxies to be identified down to limiting magnitude @xmath75 and as close as 1 or 2from the central quasar .
we performed aperture photometry on the galaxies in the quasar fields using circular apertures with radii of @xmath76 to @xmath77 , as appropriate .
we counted the number of companion galaxies brighter than a specified limiting absolute magnitude that were found to have a metric separation from one of the quasars of less than or equal to some predetermined distance .
we choose _ a priori _ a limiting absolute magnitude of @xmath78 ( four magnitudes fainter than @xmath79 ) and a maximum separation of 25 kpc ( see paper ii ) .
table [ comps ] lists all galaxies found around the quasars that satisfy these specifications .
this table gives for each quasar the number of companion galaxies that are at least as bright as @xmath80 ( if they have the same redshift as the quasar ) and that are projected within 25 kpc of the center of light of the quasar .
the separations both in arcseconds and in kpc are also given in table [ comps ] , along with the brightnesses of the companion galaxies , tabulated in both apparent and absolute magnitude .
the density of companion galaxies brighter than @xmath81 , within 25 kpc of the quasars , may be higher around quasars with elliptical hosts
. there are 2 companions for 4 spiral hosts and 13 companions for 12 elliptical hosts .
fisher et al . (
1996 ) examined the clustering of galaxies around all the quasars in this sample and found a significant enhancement of galaxies within a projected separation of @xmath82 @xmath83 of the quasars .
modeling the quasar / galaxy correlation function as a power law with a slope given by the galaxy / galaxy correlation function , fisher et al.find that the ratio of the quasar / galaxy to galaxy / galaxy correlation functions is @xmath84 .
the galaxy counts within @xmath85 @xmath86 of the quasars are too high for the density profile to have an appreciable core radius ( @xmath87 @xmath83 ) .
these results provide further support for the idea that low redshift quasars are located preferentially in groups of 1020 galaxies rather than in rich clusters .
fisher et al.do not detect a significant difference in the clustering amplitudes derived from radio loud and radio quiet subsamples .
in this section we compare the results of @xmath1-based images of luminous quasars with previously - published representative analyses of ground - based observations . in
[ subsecveron ] , we compare our @xmath1 measurements with vron - cetty & woltjer ( 1990 ) @xmath88-band results , in [ subsecdunlop ] , with dunlop et al . near - infrared images , and in [ subsecmcleod ] , with mcleod & rieke ( 1994b ) @xmath89-band observations .
vron - cetty & woltjer ( 1990 ) suggested that the apparent magnitudes of host galaxies should be measured in a fixed metric annulus that is well removed from the quasar nucleus .
they proposed an annular region of @xmath90 kpc to @xmath91 kpc for @xmath92 and @xmath93 km s@xmath33mpc@xmath33 .
the vron - cetty & woltjer proposal compares directly measurements of the same quantity made by separate groups using different techniques . in this way
, measurement uncertainties can be separated from differences caused by the variety of choices in the models used to fit to the observations .
we have three objects in common with vron - cetty & woltjer : pks 1302@xmath24102 , pks 2135@xmath24147 , and pks 2349@xmath24014 .
the specified annular regions are 2.17 and 4.35 , 2.82 and 5.63 , and 3.15 and 6.30 for the three quasars , respectively .
all three quasars have close companion galaxies in the regions specified by vron - cetty & woltjer ( see table [ comps ] ) .
the close companions of pks 1302@xmath24102 and pks 2349@xmath24014 were not noticed on the ground - based images .
table [ veron ] summarizes the aperture photometry that was performed in the same annular regions as vron - cetty & woltjer .
we list the @xmath88-band annular magnitudes obtained by vron - cetty & woltjer and their estimated absolute @xmath27-magnitude , converted to the cosmological parameters used in this paper , the absolute magnitude we measured in the @xmath1 images ( excluding the light of the companions ) with the f606w , and our estimated absolute @xmath27-magnitude .
we also include in the table the total absolute @xmath27-magnitude vron - cetty & woltjer obtained for the host galaxies ( fitting a spheroidal model ) and our estimated 2-d model @xmath94band host galaxy magnitude . the agreement between our results and the ground - based observations of vron - cetty & woltjer for the aperture photometry between 12.5 kpc and 25 kpc is satisfactory , but not as precise as we would have hoped .
the average difference between our estimated @xmath95 and that of vron - cetty & woltjer is @xmath96 this discrepancy can not be attributed to the contribution of companion galaxies . in the case of pks 1302@xmath24102 , the companion at 2lies inside the annular region studied ( 12 kpc to 25 kpc ) ; if we include its light , our estimated brightness for the host increases 0.2 mag .
part of the companion galaxy 5.5 from pks 2135@xmath24147 lies in the annular region considered , but vron - cetty & woltjer ( 1990 ) also subtracted its contribution from their measurements . the compact companion at 2 of pks 2349@xmath24014 lies outside the annular region considered .
the vron - cetty & woltjer magnitudes for the host that were estimated by fitting a spheroidal model are typically about one mag brighter than our 2-d model magnitudes , i.e. , @xmath97 table [ metric ] lists the annulus measurements of the whole sample . for the range of redshifts in our sample
, the designated vron - cetty & woltjer annular region 12.5 kpc to 25 kpc ( @xmath98 , @xmath99 ) corresponds approximately to 6 kpc to 12 kpc with our chosen cosmological parameters . in table [ metric ]
we list for each quasar the inner and outer radii in arcsec , the apparent and absolute @xmath100 aperture magnitude , and the absolute @xmath27 magnitude in the annulus ( see adopted values for @xmath101 values in table [ result ] ) . as stressed by vron - cetty and woltjer ( 1990 ) ,
these annular measurements can be compared to future measurements obtained by other techniques .
dunlop et al . ( 1993 ) obtained deep ground - based near - infrared images in the @xmath102 band for a sample of nearby @xmath103 radio loud and radio quiet quasars .
they built a library of infrared psfs by observing many bright stars .
the nuclear component was removed by selecting from the library the psf that produced the best match to the quasar psf .
the stellar psf was scaled to the same height as the central peak of the quasar .
dunlop et al .
suggest that their procedure will cause the luminosities of the hosts to be overestimated , but in practice the sign of the error could depend on whether there was a positive or a negative fluctuation in the measured light in the central peak .
an aperture diameter of 12 was used by dunlop et al . to measure the magnitudes of the hosts .
we have eight quasars in common .
table [ dunlop ] compares the @xmath1 and the dunlop et al.results .
we list the @xmath102 magnitude they obtained for the host galaxies using an aperture of diameter 12 , the @xmath102 absolute magnitude for the host , using our comoslogical parameters , the color @xmath104 for an elliptical galaxy obtained from bruzual & charlot ( 1993 ) , the corresponding @xmath27 absolute magnitude expected , the 2-d model @xmath27 absolute magnitude estimated from the @xmath1-f606w images ( see next - to - last column of table [ bestmodel ] ) , and the difference between the @xmath27 absolute magnitude derived from both bands , @xmath105 .
the average discrepancy is @xmath106 for seven of the eight cases , our estimated magnitudes are brighter than obtained by dunlop et al .
( 1993 ) .
mcleod & rieke ( 1994a , b ) obtained ground - based images of luminous quasars in the @xmath89-band . for most cases , they determined a one - dimensional profile for the quasar , subtracted a stellar psf , and then fit the resulting profile with an analytic galaxy model .
we have fourteen quasars in common . to compare the results , we transform both the @xmath89-band and the f606w magnitudes to the @xmath27 band .
we used the k - corrections and the relative sensitivities of the different bands calculated by fukugita et al .
( 1995 ) to convert our f606w measurements to @xmath27 ( see [ secsize ] ) . to transform the @xmath89-band magnitudes to @xmath27
, we assumed @xmath107 for normal galaxies plus a k - correction given by mcleod and rieke ( 1995 ) . for objects not in their table , we interpolated in redshift to obtain the expected @xmath108 for a normal galaxy .
table [ mcleod ] lists the following information : column 1 : quasar ; column 2 : mcleod - rieke @xmath89-band magnitude for the host galaxy ; column 3 : color @xmath108 for normal galaxies including k - correction ; column 4 : absolute @xmath27-magnitude based on the @xmath89-band measurements and calculated assuming @xmath4 , and @xmath109 ; column 5 : absolute 2-d model @xmath27-magnitude estimated from f606w images ; and column 6 : difference between absolute @xmath27-magnitude estimated from @xmath89-band and f606w images .
the average discrepancy is @xmath110 for 8 of the 14 cases , the mcleod & rieke luminosities are brighter than the @xmath1 luminosities .
in some cases , the difference is clearly due to the mcleod & rieke magnitudes also including the companion galaxies ( see pg 1012@xmath45008 in figures [ figallqso ] and [ figallqsosub ] , and the discussion in [ comments ] ) .
in this section we will discuss the images of each of the quasars .
table [ size ] summarizes the morphological information obtained from the @xmath1 images and tables [ result]-[bestmodel ] give the inferred luminosities of the host galaxies .
for comparisons , the absolute magnitudes reported by other authors have been converted to our cosmological parameters .
information about close galaxy companions is summarized in table [ comps ] .
measurements of the surface brightness along the major axes of the relatively bright elliptical host galaxies of phl 909 , pg 0923 + 201 , pks 1004@xmath45130 , he 1029@xmath24140 , pg 1116@xmath45215 , 3c 273 , and pks 2135@xmath24147 , do not show evidence of discontinuity in the light distribution . thus all the apparently elliptical hosts discussed below for which we could make detailed photometry satisfy the criterion of having smooth light distributions .
the @xmath1 images reveal spiral host galaxies with h ii regions for three quasars , pg 0052 + 251 , pg 1309@xmath45355 , and pg 1402 + 261
. it should be feasible to obtain spectra of the brightest h ii regions .
the spectra may reveal the composition of the material which makes up the quasar hosts and perhaps provide further clues to the quasar phenomenon . *
pg 0052@xmath45251 : * the host is a beautiful spiral galaxy ( see bahcall et al 1996 for a detailed discussion ) .
the spiral host is evident even in our 200 s exposure .
the southern spiral arm extends the furtherest from the quasar ( in the direction of the companion ) .
there is good agreement between the absolute @xmath27 magnitude estimated from the mcleod and rieke ( 1994b ) @xmath89-band image measurements and our 2-d model estimate based on the @xmath1 images .
miller ( 1996 , private communication ) has recently obtained spectra for some of the bright h ii regions identified ( bahcall et al.1996 ) in the @xmath1 images .
the h ii regions observed have the same redshift as the quasar . the second and third quasar nuclei suggested by hutchings , janson & neff ( 1989 ;
the second nucleus was confirmed by dunlop et al .
1993 ) , are seen on the @xmath1 images to be , respectively , a bright h ii region in the spiral arm of the host galaxy , and a foreground star in the galaxy . * phl 909 ( 0054@xmath45144 ) : * the quasar host is a normal elliptical e4 galaxy ( see bahcall et al 1996 for a more extensive discussion and a variety of images ) .
this radio quiet quasar does not occur in a spiral galaxy , as the conventional view had suggested before the @xmath1 observations .
we did not detect extended emission towards the western companion galaxy , as suggested by dunlop et al ( 1993 ) .
the elliptical host is apparent in our 200 s image .
* nab 0205@xmath4502 : * the appearance of this quasar in the @xmath1 images ( see fig .
[ figallqso ] ) resembles , on visual inspection , that of a bright star .
indeed , it is hard to distinguish the quasar from a star by just looking at the images ( see figure [ figallqso ] ) .
after psf subtraction , we detect a small disk - like host galaxy ( see figure [ figallqsosub ] ) , with size and inclination similar to the companion 8.3 , position angle @xmath111 .
the scale length of the host is only @xmath112 kpc and the absolute visual magnitude is @xmath113 .
figure [ bluevsred ] compares the host image after subtraction of a blue standard star , mmj 6481 ( see figure [ bluevsred]a ) , and after the subtraction of a red standard star , f141 ( see figure [ bluevsred]b ) .
the image of the host galaxy obtained after the subtraction of the psf for the blue standard star is significantly clearer than the image obtained using the red - star psf .
however , there is very little difference quantitatively , @xmath114 mag , between the host galaxy magnitudes we determined from the two subtracted images .
stockton , ridgway , & kellogg ( 1996 in preparation ) recently obtained ground - based images in two line - free continuum bands that show a definite host galaxy elongated roughly nw - se as well as some extended low - surface - brightness material to the west . in agreement with our results
, they report that the host galaxy has about the same luminosity and orientation as the companion in the shorter wavelength image , but the host has somewhat redder colors than the companion .
nab 0205@xmath4502 was not resolved in the deep images taken by smith et al ( 1986 ) .
figure [ nabobj ] shows a fascinating object , first noticed by stockton and mackenty ( 1987 ) .
the object , which is 12 to the east of the quasar , is visible in their [ o iii ] image , but absent from their continuum image .
it appears as a point - source with a bright jet - like structure to the south and a much fainter and curved extension to the north .
the width of the jet and of the curved extension is less than 0.3 kpc .
the total apparent f606w magnitude of the object is 22.6 , and the extension of the jet - like structure and curved tail are 0.7 ( 1.2 kpc ) and 1.1 ( 1.9 kpc ) , respectively .
the offset of this object from the quasar position is and @xmath115 . *
0316@xmath24346 : * the morphology of the host environment is complex , probably the result of gravitational interactions with a neighboring galaxy
. figures [ figallqso ] and [ figallqsosub ] show clearly what appear to be tidal tails that extend about 20 kpc west of the quasar .
there are also bright diffuse clumps within @xmath116 kpc of the quasar , which may contain h ii regions .
we do not see unambiguous signs of a compact remnant close to the center of light of the quasar .
the radial profile is reasonably described by an exponential disk from radius @xmath117 1 to 6 . there is a relatively bright peculiar spiral galaxy ( @xmath118 mag ) projected 26.7of the quasar , @xmath119 .
* pg 0923@xmath45201 : * the host of this radio quiet quasar is an e2 elliptical , a member of a small group of galaxies . the two galaxies at 11.0 , @xmath120 , and 15.1 , @xmath121 , have redshifts similar to the quasar ( heckman et al 1984 ) ; the companion at 15se is not shown in figures 1 and 2 .
the @xmath27 magnitude estimated from mcleod & rieke ( 1994b ) @xmath89-band images is in good agreement ( 0.1 mag ) with the magnitude determined from our 2-d model .
* pg 0953@xmath45414 : * low surface brightness fuzz was detected in the early @xmath1 images ( bahcall et al .
1994 , 1995a ) and confirmed to be real by longer ( cycle 5 ) exposures .
we are unable to establish the morphology of the fuzz , which is faint and extended .
hutchings et al ( 1989 ) suggested , on the basis of ground - based images , that the host of the quasar has spiral structure and is possibly interacting .
the hst images do not provide convincing evidence for or against this conjecture .
spectroscopy of the fuzz around the quasar obtained by boroson et al ( 1985 ) , shows that the off - nuclear spectrum is dominated by a red continuum with @xmath122 and possibly [ s ii ] emission ; mg ib absorption might also be present .
the absolute @xmath27 magnitude derived from mcleod and rieke ( 1994b ) @xmath89-band observations is 0.5 mag brighter than our 2-d model estimate .
* pks 1004@xmath45130 : * the host is a bright elliptical , about as bright as the brightest cluster galaxies .
there seems to be some structure in the inner region of the host ( @xmath123 ) .
the absolute @xmath27 magnitude derived from mcleod and rieke ( 1994b ) @xmath89-band observations is 0.8 mag fainter than our 2-d model estimate .
stockton and mackenty ( 1987 ) presented narrow band [ o iii ] images of this quasar , but did not find any significant extended luminosity in the region between @xmath124 to 28 kpc ( for @xmath125 and @xmath126 km s@xmath33mpc@xmath33 ) .
stockton ( 1978 ) obtained spectroscopic observations of the galaxies around the quasar and found that two of them have redshifts similar to the quasar redshift .
one of the galaxies is separated by 33.4 ( 79.8 kpc , @xmath127 ; not shown in figures 1 and 2 ) from the quasar and has @xmath128 ; the other is not in our @xmath1 image field of view .
* pg 1012@xmath45008 : * this quasar is `` caught in the act '' of merging .
the host of this quasar is seen in figures [ figallqso ] and [ figallqsosub ] to be merging with a bright galaxy .
the distance between the two galactic nuclei is 6.7 kpc ( 3.3 ) .
there is another compact galaxy 6.8 ( 12.4 kpc ) north of the quasar , probably taking part in the interaction as well .
both interacting companions are also visible in our 200 s exposure .
figure [ pg1012 ] shows an expanded view of the pg 1012@xmath45008 image .
heckman et al . (
1984 ) obtained spectra for both galaxies and found the expected absorption lines ( @xmath28 , mg i , fe ii and na i ) at the same redshift as the quasar .
the absolute @xmath27 magnitude based on the @xmath89-band measurement by mcleod & rieke ( 1994b ) is 0.9 mag brighter than our 2-d model estimated value ; we believe that part of this discrepancy is caused by the light of the merging galaxy being included in their model fit .
* he 1029@xmath24140 : * this radio quiet quasar has a bright elliptical host galaxy .
some faint structures resembling shells are seen at @xmath129 and 19 from the quasar .
there is a compact galaxy 4.1north from the quasar ; spectroscopy is required to determine if it is associated with the quasar .
wisotzki et al .
( 1991 ) obtained @xmath130-band images and , based on spheroidal model fits , reported the host to be a giant elliptical with @xmath131 mag and @xmath132 mag .
assuming @xmath133 for elliptical galaxies at @xmath134 ( fukugita et al .
1995 ) , their absolute @xmath27 magnitude for the host is @xmath135 mag .
we obtained @xmath136 by fitting a gdv model to the azimuthally averaged profile of the host .
wisotzki ( 1994 ) measured redshifts of galaxies in the field of he 1029@xmath24140 and found that four galaxies have redshifts similar to the quasar .
wisotzki did not detect the close compact object at 4 projected separation . the closest galaxy with a redshift known to be similar to the quasar lies at 134 kpc ( = 109.4 ) @xmath137 and was detected in wf2 ( not shown in figures [ figallqso ] and [ figallqsosub ] ) .
the @xmath1 images show that this wisotzki companion galaxy has @xmath138 ( @xmath139 ) and is highly disturbed , probably an advanced stage of merger between two disk galaxies .
two possibly galactic nuclei , separated by 1.4 , are visible , as well as h ii regions .
there is a @xmath140 mag galaxy projected at 22.9 , @xmath141 , of the quasar , which wisotzki showed is a background galaxy with @xmath142 .
wisotzki classified this galaxy as an elliptical , but the @xmath1 images show that it is a spiral galaxy ( type @xmath117sab ) .
* pg 1116@xmath45215 : * the host of this radio quiet quasar is probably an elliptical galaxy . the bright central region is surrounded by a faint smooth fuzz .
some arc - like structures are seen in the center , but are almost certainly psf artifacts .
the one - dimensional profile is reasonably described by a de vaucouleurs model .
the @xmath27 magnitude estimated from mcleod & rieke ( 1994b ) @xmath89-band images is in good agreement with the magnitude determined from our 2-d model .
mcleod and rieke ( 1995 ) applied the techniques they developed to study their @xmath89-band ground - based images , to reanalyse the @xmath1 archive images of pg 1116@xmath45215 .
they subtracted just enough of the psf of the red standard star , f141 ( @xmath143 ) , from the azimuthally - averaged quasar 1-d profile to make the residual profile almost turn over in the center . because the @xmath1 images were saturated in the center
, mcleod & rieke normalized the light profile just outside the saturation region .
we note , however , that the region inside @xmath144 is heavily contaminated by psf artifacts when the psf of the red star f141 is used for the subtraction .
if f141 is used for the psf subtraction , there are artifical radial spikes - which are not visible in their published @xmath1 image of this quasar .
mcleod and rieke state that the host of pg 1116@xmath45215 looks nearly identical " to a field galaxy detected in wf4 .
an @xmath1 image of this field galaxy was published in bahcall et al .
( 1995a , figure 5c ) .
the field galaxy is rich in morphological details : it is a barred spiral with internal and external rings ( rsbb ) .
if it is at the same distance as the quasar , its absolute @xmath27 magnitude is @xmath145 . although the host and the field galaxy have apparently similar luminosities , they are morphologically quite different .
* pg 1202@xmath45281 : * the host of this radio quiet quasar , also known as gq com , is a small elliptical e1 galaxy .
stockton and mackenty ( 1987 ) showed that the compact galaxy @xmath146 from the quasar , @xmath147 , is at the same redshift as the quasar .
boroson et al . (
1985 ) reported that the off - nuclear spectra are dominated by a red continuum , with [ o iii ] lines and possibly @xmath122 in emission , and a possible mg ib absorption feature .
the @xmath27 magnitude estimated from the mcleod & rieke ( 1994b ) @xmath89-band images is in good agreement with the magnitude determined by our 2-d model . *
3c 273 ( pg 1226@xmath45023 ) : * the host galaxy is an elliptical .
the @xmath27 magnitude estimated from the mcleod & rieke ( 1994b ) @xmath89-band images is in good agreement with the magnitude determined by our 2-d model . using deep ground - based ccd images , tyson et al .
( 1982 ) obtained @xmath148 for the host , which is in good agreement with our best 2-d model magnitude of @xmath149 .
the host is somewhat brighter than the brightest galaxy in a rich cluster .
stockton ( 1980 ) measured redshifts for galaxies in the 3c 273 field and found that four of them have redshifts similar to the quasar , in agreement with the suggestion of bahcall & bahcall ( 1970 ) .
one of those galaxies was detected ( a spiral galaxy ) in wf4 : it lies at 75 east of the quasar ( @xmath150 kpc ) and its redshift is @xmath151 .
wyckoff et al . ( 1981 ) obtained @xmath152 for the host , which is equivalent to @xmath153 mag .
the inner part of the jet is barely visible in the psf subtracted image in figure [ figallqsosub ] .
@xmath1 and merlin observations of the 3c 273 jet are reported in bahcall et al .
( 1995 ) .
* pks 1302@xmath24102 : * the @xmath1 images show there are two small compact galaxies at 1 and 2 from the quasar , which are expected to spiral into the quasar in a time short compared to the hubble time ( bahcall et al.1995a ) .
the presence of these very close companions can be seen most clearly in the expanded image , figure 8 , of bahcall et al .
( 1995a ) .
hutchings and neff ( 1992 ) performed optical imaging with 0.5 resolution and reported structures at the positions of those galaxies ; they suggest the host is a mildly disturbed elliptical galaxy .
the psf - subtracted residual image ( see figure [ figallqsosub ] ) appears to the eye to be similar to an elliptical galaxy , but an exponential disk fits the data slightly better . the absolute @xmath27 magnitude estimate for the host galaxy is 0.4 mag fainter than the value determined from the mcleod & rieke ( 1994b ) @xmath89-band observations , and is 0.9 mag fainter than the value estimated from @xmath88-band images ( vron - cetty & woltjer 1990 ) .
wyckoff et al . (
1981 ) obtained @xmath154 for the host ; using the fukugita et al .
( 1995 ) transformations of galaxy colors , this corresponds to @xmath155 . *
pg 1307@xmath45085 : * the host appears to be a small early - type galaxy .
the absolute @xmath27 magnitude estimated from our 2-d model of the @xmath1 image is 0.2 mag fainter than the value derived from the mcleod & rieke @xmath89-band images .
* pg 1309@xmath45355 : * figure [ pg1309 ] shows an expanded view of the image for pg 1309@xmath45355 with lower contrast than shown in figures [ figallqso ] and [ figallqsosub ] .
the details near the center of pg 1309@xmath45355 are shown more clearly in figure [ pg1309 ] , after a best - fit stellar psf has been subtracted .
this quasar has a bright early - type host galaxy , probably an sab .
tightly wrapped spiral arms are clearly seen in the inner region surrounding the center of the quasar .
overall , the gdv model describes the radial profile better than the exponential disk model .
the absolute @xmath27 magnitude estimated from our 2-d model of the @xmath1 image is 0.1 mag brighter than the value derived from the mcleod & rieke ( 1994b ) @xmath89-band images .
* pg 1402@xmath45261 : * figure [ pg1402 ] shows an expanded view of the image of this quasar at two different contrast levels .
the @xmath1 images show that the host is a beautiful spiral . after the psf subtraction
a bar and a possible inner ring are visible .
the morphological type is approximately sbb(r ? ) . h ii regions are also visible along one of the spiral arms ; they are less prominent than the h ii regions seen in the spiral host of pg 0052@xmath45251 . the relative positions and magnitudes of the brightest h ii regions
are marked for identification in the upper panel in figure [ pg1402 ] .
table [ hiitab ] lists for each h ii region the aperture magnitude , the distance to the quasar nucleus , and the offset in right ascension and declination between the h ii region and the quasar nucleus .
the magnitudes were measured using apertures of @xmath156 .
pg 1402@xmath45261 is a relatively isolated quasar .
stockton and mackenty ( 1987 ) did not find any significant extended [ o iii ] emission around the quasar .
mcleod and rieke ( 1994b ) obtained @xmath89-band images ; the absolute @xmath27 magnitude based on their measurement is 0.5 mag brighter than the value estimated with our 2-d model of the @xmath1 images .
* pg 1444@xmath45407 : * the host galaxy of this quasar is smooth and is elongated in the ns direction . after the psf subtraction , the nuclear region
is seen as an extended structure running ne - sw , as shown in figure [ figallqsosub ] .
a bar may be present .
the exponential disk model describes the radial light distribution a little better than does the gdv model .
the overall appearance is most similar to an e2 galaxy .
hutchings and neff ( 1992 ) originally suggested the possibility of a bar in this host , which they proposed might be in an advanced stage of the merger of galaxies of very different masses .
the absolute @xmath27 magnitude estimated from our 2-d model of the @xmath1 image is 0.5 mag fainter than the value derived from the mcleod & rieke @xmath89-band images . *
3c 323.1 ( pg 1545@xmath45210 ) : * a low surface - brightness elliptical host galaxy appears in the @xmath1 images , as well a neighboring companion at a projected distance of @xmath124 kpc , ( 2.7 , @xmath157 ) , ( see figures [ figallqso ] and [ figallqsosub ] ) .
the @xmath27 magnitude of the host galaxy determined from the @xmath89-band images by mcleod and rieke ( 1994b ) is 0.4 mag brighter than the value derived from our 2-d model of the @xmath1 images .
more recently neugebauer et al . ( 1995 ) obtained @xmath89 and @xmath102-band images of 3c 323.1 ; their estimated model fit @xmath89-band magnitude for the host agrees with mcleod & rieke ( 1994b ) results .
neugebauer et al .
measurements suggest that if the host galaxy is a normal elliptical , its expected apparent @xmath27 magnitude is @xmath158 , and if it is an spiral galaxy , @xmath159 .
these values are brighter than the @xmath27-band apparent magnitudes expected using a color transformation without reddening of our @xmath1 images ( see table [ model ] ) , but are consistent with the host galaxy being a reddened elliptical , as pointed out by neugebauer et al .
the gdv one - dimensional model fits better the surface brightness radial profile of the quasar host in our @xmath1 image than does the exponential disk model .
stockton ( 1982 ) took spectra of the companion galaxy at 2.7 from the quasar , and apparently detected [ o iii ] at the same redshift as the quasar .
complex , asymmetric extended [ o iii ] emission is seen surrounding the quasar in [ o iii ] images obtained later by stockton and mackenty ( 1987 ) . however , there is no excess of emission at the companion galaxy position in the [ o iii ] images .
stockton and mackenty conclude that the previously reported detection of [ o iii ] emission in the spectrum of the companion galaxy was fortuitous , since the emission is due to the general distribution of ionized gas surrounding the quasar .
neugebauer et al
. found that the @xmath89 and @xmath102 model fit magnitudes obtained for the companion galaxy , when combined with the @xmath1 measurements by bahcall et al .
( 1995a ) , gives a color @xmath160 consistent with the companion galaxy being a faint elliptical at the redshift of the quasar .
* pks 2135@xmath24147 : * the quasar host is an elliptical galaxy ; the envelope is not featureless like phl 909 , but contains some faint clumps .
the image suggests the presence of a jet at @xmath161 , with visual extension of @xmath162 and width @xmath163 .
the surface brightness of the jet - like feature is @xmath164 ( f606w ) , at 1.7 from the quasar center of light .
many galaxies are seen in the field ; the closest companions to the quasar are at 1.9 , @xmath165 , and 5.5 , @xmath166 , seen in figures [ figallqso ] and [ figallqsosub ] .
stockton ( 1978 , 1982 ) obtained spectra of the field galaxies and found that four are at the same distance as the quasar , including the closest companions at projected separations of 1.9 and 5.5 .
somewhat later , stockton & mackenty ( 1987 ) obtained [ o iii ] images .
they noted that the companion 2 southeast did not appear enhanced in the [ o iii ] images , indicating that the emission lines seen in the combined spectrum are due to extended emission around the quasar that is not confined to the companion .
hickson and hutchings ( 1987 ) reported mgb in the spectrum of the secondary nucleus ( the galaxy at 2 se ) corresponding to a galaxy at the redshift of the quasar .
pks 2135@xmath24147 has been included in many imaging programs .
for example , dunlop et al . (
1993 ) measured @xmath167 , which gives @xmath168 .
our 2-d model analysis of @xmath1 images gives @xmath169 .
vron - cetty & woltjer ( 1990 ) obtained @xmath88-band images and measured aperture photometry in an annulus with radii 12.5 kpc and 25 kpc .
they estimated @xmath170 .
we measured the absolute magnitude of the host in our @xmath1 images within the same annulus and obtained @xmath171 .
smith et al .
( 1986 ) measured the absolute blue magnitude to be @xmath172 , which converts to @xmath173 ( if @xmath174 for an elliptical galaxy at @xmath175 ) .
wyckoff et al . ( 1981 ) obtained @xmath176 , which corresponds to @xmath177 .
* pks 2349@xmath24014 : * the quasar host galaxy is undergoing gravitational interaction , which is evidenced by apparently tidal arms and possibly also by a huge ( @xmath178 kpc ) diffuse off - center nebulosity .
a compact companion galaxy at 2 e is detected in the @xmath1 images .
bahcall et al.(1995b , 1995c ) provide an extensive discussion of this object and include a number of different images ( the tidal wisps and extended nebulosity are seen most clearly in their figures [ figallqso ] and [ figallqsosub ] ; the close companion galaxy is featured in their figure 3 ) .
dunlop et al .
( 1993 ) suggests that the quasar is interacting with the three closest galaxies se of the quasar .
the @xmath1 images do not show evidence for this interaction ; the three galaxies do not seem to be morphologically disturbed and no obvious link between them and the quasar is seen .
dunlop et al .
measured @xmath179 , which gives @xmath180 .
our 2-d model fits to the @xmath1 image of the host gives @xmath181 , similar to the light from a brightest cluster galaxy . the radial profile of the interacting system pks 2349@xmath24014
is well described by a de vaucouleurs model . as pointed out by toomre ( 1995 ) and bahcall et al .
( 1995b ) , there are some morphological resemblances between pks 2349@xmath24014 and ngc 3921 .
schweizer ( 1996 ) has carefully and extensively studied ngc 3921 , suggesting that it is the result of a merger between two disk galaxies and is now a protoelliptical . in this case also the mean light distribution of the system is well described by a @xmath182 law .
the images shown in figures 1 and 2 establish the two main conclusions of this paper : 1 ) the most luminous nearby quasars exist in a variety of environments ; 2 ) @xmath1 observations provide unique information about the circumstances of the quasar phenomenon .
many of the most important results obtained in this paper are visible on the unprocessed images shown in figure [ figallqso ] , which can be compared with the images with a stellar psf subtracted shown in figure [ figallqsosub ] .
the subtraction of a stellar psf is important in about half of the cases .
our results are based upon a representative sample of 20 of the most luminous known ( @xmath19 ) and nearby ( @xmath3 ) quasars .
the characteristics of the sample are summarized in table [ journal ] and section [ subsample ] . in separate subsections ,
we summarize and discuss below our results on host morphologies , host luminosities , comparisons with our earliest analyses , close companions , ground - based studies , and future work . figures [ figallqso ] and [ figallqsosub ] show three hosts that apparently are normal spirals with h ii regions ( pg 0052@xmath45251 , pg 1309@xmath45355 , and pg 1402@xmath45261 ) , seven ellipticals ( phl 909 , pg 0923@xmath45201 , pks 1004@xmath45130 , he 1029@xmath24140 , pg 1116@xmath45215 , 3c 273 , and pks 2135@xmath24147 ) , as well as three obvious cases of current gravitational interaction ( 0316@xmath24346 ,
pg 1012@xmath45008 , and pks 2349@xmath24014 ) .
there are also five other hosts that appear to be elliptical galaxies and are listed as en ( ? ) in table [ size ] .
the hosts for two of the quasars ( nab 0205@xmath4502 and pg 0953@xmath45414 ) are faint and difficult to classify .
seven of the 14 radio quiet quasars in our sample occur in hosts that are classified as elliptical galaxies in table [ size ] .
two particularly beautiful examples of elliptical hosts for radio quiet quasars are shown in figures 1 and 2 for phl 909 and he 1029@xmath24140 .
we have presented a more extensive discussion of the host of phl 909 in bahcall et al .
five of the six radio loud quasars in our sample appear to lie in elliptical galaxies .
the sixth radio loud quasar , pks 2349@xmath24014 , is in a complex interacting system containing a close companion , apparent tidal tails , and a large off - center nebulosity .
the fact that about half of the radio quiet quasars in our sample have elliptical hosts contradicts the conventional wisdom that radio quiet quasars occur in spiral galaxies .
however , we confirm the expectation that most radio loud quasars are in elliptical galaxies .
three of the quasars in our sample , 0316@xmath24346 , pg 1012 + 008 , and pks 2349@xmath24014 , have been `` caught in the act '' , i. e. , the @xmath1 images of these quasars show dramatic evidence of currently intense gravitational interactions .
two of the three quasars caught in the act are radio quiet . in all three cases , the unprocessed images ( see figure [ figallqso ] ) are sufficient to reveal extended curved features that look like the tidal arms generated in numerical simulations of interacting galaxies . for 0316@xmath24346 and pks 2349@xmath24014 ,
there is no clear evidence for a normal host galaxy centered on the quasar .
pg 1012 + 008 appears to be an example of a merger currently going on between two comparable galaxies . for a more extensive analysis of pks 2349@xmath24014 ,
the reader is referred to bahcall et al .
( 1995b , c ) , in which the close companion , the tidal arms , the very extended off - center nebulosity , and the possible host galaxy are all discussed in some detail .
the hst images provide the best available data base to search for optical jets in nearby luminous quasars .
information on the existence or non - existence of small scale optical jets can constrain theories of the origin of radio and optical jets .
we examined carefully the images of all 20 of the quasars for the existence of optical jets . with the exception of the well - known jet in 3c 273 , the only other quasar for which we have found evidence for linear features is pks 2135@xmath24147 .
sensitive radio searches should be undertaken to test whether pks 2135@xmath24147 has a radio jet .
for the other quasars , we can rule out the existence of a narrow optical feature with a surface brightness in excess of 24.5 mag @xmath184 extending more than 3 kpc beyond an inner region beginning at about 6 kpc .
we list in table [ summary ] our best estimates for the magnitudes of the host galaxies in our sample ; these magnitudes were obtained ( see [ subtwo ] ) by fitting a two - dimensional analytic galaxy model to the data .
we also list the effective radius or exponential scale length , and give the morphology of the host based on visual inspection of the images ( see table [ size ] and [ subunproc ] ) .
the different measurements of the brightnesses of the host environments are summarized and compared in tables [ result ] and [ model ] and in section [ secsize ] .
the average best - fit 2-d model magnitudes for the hosts of the 14 radio quiet quasars is the fact that in our sample the radio loud quasars are , on the average , about a magnitude brighter than the radio quiet quasars can not be explained by a selection effect resulting from the fact that the radio loud quasars have a slightly larger average redshift .
in fact , the 3 radio loud quasars with @xmath7 have @xmath187 mag and the 3 radio loud quasars with @xmath188 have @xmath189 mag .
figure [ mvmv ] shows the 2-d model absolute visual magnitudes ( from table [ summary ] ) of the host galaxies ( and other nebular material ) versus the absolute visual magnitudes of the quasars . for two of the quasars ,
phl 909 and 3c 323.1 , the symbols overlap at @xmath190 and @xmath191 . in order to be detectable , the host must have a luminosity that is not too small when compared with the luminosity of the quasar .
the minimum detectable host brightness depends strongly upon the assumed morphology of the host galaxy .
we have shown by a series of numerical experiments , described in table 3 of paper ii , that host galaxies are , on the average , visible on our images down to about 4.2 mag fainter than the quasar luminosity .
galaxies that are smooth ellipticals are the most difficult to detect ( see rows 5d and 5e of table 3 of paper ii ) . for the eight quasars discussed in paper ii , smooth elliptical hosts are , on average , visible on our images down to @xmath192 mag fainter than the quasar .
the limiting brightnesses were determined by visually inspecting simulated galaxies placed in the actual hst quasar observations and are therefore somewhat subjective .
the diagonal line in figure [ mvmv ] represents the detection limit for smooth ellipticals in an idealized sample in which the limiting host magnitude is determined entirely by the quasar luminosity . in calculating the limiting absolute visual magnitudes for the hosts
, we have included an average k - correction ( see fukugita et al .
1995 ) for elipticals at @xmath193 , as well as the average magnitude difference , @xmath194 mag , between the quasar and the faintest detectable host .
thus the form of equation ( [ linearrelation ] ) reflects the fact that the limiting host luminosity must increase linearly with the luminosity of the quasar , since by assumption the noise introduced by the quasar signal determines how faint a host can be detected .
smooth ellipticals fainter than @xmath196 mag of the diagonal line in figure [ mvmv ] would not have been expected to be detected if this idealization of the problem is correct .
for the three intrinsically faintest quasars in our sample , the noise introduced by the quasar , the host galaxy , and the background light are all similar . in practice , for our sample one might expect some flattening of the detection limit at the lowest quasar luminosities if photon noise is more important than systematic uncertainties in the subtraction of the quasar light .
mcleod and rieke ( 1995 ) have suggested that there is a linear relation between the quasar absolute magnitude and the minimum host galaxy absolute magnitude .
they interpret this linear relation , shown in their figure 2 , as indicating that a more luminous host galaxy is required to fuel a more luminous quasar .
the linear relation that they find between @xmath197 and @xmath198 is essentially identical to our minimum detection limit for smooth ellipticals that is shown in figure [ mvmv ] ( for @xmath199 ) . as pointed out by mcleod(1996 ) , the relation described by mcleod and rieke can not be an artifcat produced by detection limits if all of their detections are real detections .
an artifical correlation would be introduced only if true non - detections were interpreted as marginal detections .
there is not convincing evidence in figure [ mvmv ] for a significant dependence of host luminosity upon the luminosity of the quasar .
the apparent correlation that is suggested to the eye is due in large part to the fact that the single most luminous quasar in our sample , 3c 273 , has the most luminous host .
there is a hint in figure [ mvmv ] that the average luminosity of elliptical hosts is somewhat higher than for spiral hosts .
most of this difference , however , is due to the fact that , even for the same objects ( cf .
table [ model ] ) , the de vaucouleurs fitting formula yields estimated luminosities that are @xmath200 mag brighter than the disk fitting function .
the de vaucouleurs formula introduces a luminosity peak in the unmeasured region that is not present in the disk formula .
our results are inconsistent with the hosts having a schechter luminosity function .
the average absolute magnitude for a field galaxy is about 1.8 mag fainter than @xmath201 , ( for an assumed minimum luminosity of @xmath202 ; see , e. g. , efstathiou , ellis , and peterson 1988 for a discussion of the field galaxy luminosity function ) . in our sample ( see table [ summary ] and figure [ mvmv ] ) the average host is @xmath203 . moreover , about half by number of the field galaxies would be expected , in a volume limited sample , to lie within a magnitude of the lower limit cutoff of the schechter luminosity function ( which may be fainter than @xmath202 ) .
thus , if the host galaxies were distributed with a schechter luminosity function , we would have expected that about half of the hosts in our sample would be fainter than @xmath204 and therefore undetectable on the hst images .
this is clearly not the case . by comparing the 2-d model magnitudes of table [ summary ] and
figure [ mvmv ] with the results expected from a schechter luminosity function , we conclude that , on average , the host galaxies of the luminous quasars in our sample are about @xmath0 magnitudes more luminous than typical field galaxies .
the conclusions presented in this summary paper are different in emphasis from the conclusions in our first two studies ( see papers i and ii ) . in our earlier work , we discussed images of eight quasars ( all are included in the sample in the current work ) , and reported the definitive detection of three host galaxies .
we also presented limits on the brightnesses of the hosts for the other five quasars .
the images presented in this paper show that more than half of the entire sample of 20 quasars has obvious hosts , and there is solid evidence that most , if not all , of the remaining quasars also have host galaxies .
the initial caution that we expressed regarding the detection of host galaxies was due to a combination of the unlucky observing sequence and our conservatism about interpreting the complex @xmath1 images .
the unprocessed @xmath1 data ( see figure [ figallqso ] ) are sufficient to show , even to the untrained eye , that at least nine of the 20 quasars in our sample have obvious hosts or diffuse environments .
these obvious examples include the three spiral hosts ( pg 0052 + 251 , 1309 + 355 , and pg 1402 + 261 ) , the three quasars `` caught in the act '' ( 0316@xmath24346 , pg 1012 + 008 , and pks 2349@xmath24014 ) , and three prominent ellipticals ( phl 909 , he 1029@xmath24140 , and 3c 273 ) .
none of these quasars were among the first four objects ( pg 0953 + 414 , pg 1116 + 215 , pg 1202 + 281 , and pg 1307 + 085 ) observed ( paper i ) , and only one ( 3c 273 ,
whose host we described in paper ii together with the original observations ) belonged to the sample of eight quasars ( paper ii ) . with a probability of @xmath205 per observation of observing an obvious host , it was simply bad luck ( about a 5% chance ) that only one of the initial eight quasars studied in paper i and paper ii had an obvious host
. given the repaired , but still complex and temporally variable psf of the @xmath1 , we presented our results ( see table 3 of paper ii ) for the non - detections as a morphology - dependent limit on the brightness of the host galaxy .
spiral galaxies , with their azimuthal variation in brightness and regions of high surface brightness , could be seen to considerably fainter total brightnesses ( more than a magnitude ) than large , extended ellipticals with their smooth , regular profiles . as we have gained experience with the data , we have become more confident of our ability to judge the reality of low surface brightness features .
most importantly , during @xmath1 observation cycle 5 , we obtained additional images of pg 0953 + 414 , a quasar analyzed in paper i which showed very faint , extended nebulosity that was not centered on the quasar .
the cycle 5 observations were obtained at a different roll angle than those described in paper i ; the new observations confirmed the reality of the diffuse features ( the nebulosity remained fixed in the sky when the telescope was rotated ) .
the observations of pg 0953 + 414 suggested that some of the faint features were real that we had initially worried could be psf features .
in retrospect , it appears that the vast majority of our initial observations consisted of quasars whose hosts had smooth , regular profiles .
comparing our adopted brightnesses of the hosts ( table [ summary ] ) with the brightness limits set in our earlier work by simulations ( table 3 in paper ii ) , we find that the appropriate brightnesses limits ( those for smooth ellipticals like 5d and 5e of figure 5 of paper ii ) were reasonably accurate .
the detected brightnesses reported in table [ summary ] of this paper range from considerably fainter than the paper ii limit ( _ e.g. , _ pg 0953 + 414 ) to slightly brighter than the limit ( _ e.g. , _ 3c 323.1 ) .
since the morphology of the hosts of the initial quasars was biased towards one type of galaxy ( the least favorable type as far as detectability ) , it was not accurate , as we did , to quote a detection limit that was the average of all the galaxy types .
the @xmath1 images frequently reveal companion galaxies that are projected very close to the quasar , in some cases as close as @xmath206 or @xmath207 .
table [ comps ] shows 20 galaxy companions that are projected closer than @xmath41 kpc to the center - of - light of a quasar and brighter than @xmath208 .
altogether , 13 of the 20 quasars in our sample have close companions that satisfy the requirements for inclusion in table [ comps ] .
the amplitude for the quasar - galaxy correlation function determined from our data is @xmath209 times larger than the galaxy - galaxy correlation function ( fisher et al .
1996 ) .
our results for individual objects are compared in section [ groundsection ] with the results from previous ground - based observations . in general , the agreement with ground - based observations is satisfactory , but not as precise as we would have hoped .
the most straightforward comparison is with the results of vron - cetty & woltjer ( 1990 ) in annular regions well separated from the quasar . even in this case ,
our @xmath1 magnitudes are , on average , 0.4 mag fainter than the vron - cetty & woltjer values .
our 2-d model estimates for the total luminosities are , on average , 0.8 magnitude fainter than their 1-d model fits .
the average discrepancy between our results and dunlop et al . ( 1993 ) is @xmath210 mag ( our results are generally brighter than dunlop et al . ) , and the average discrepancy between our results and mcleod & rieke is @xmath211 mag . some pre - hst theoretical analyses ( e. g.,by falle , perry , and dyson 1981 , weymann et al . 1982 , begelman 1985 , and chang , schiano , and wolfe 1987 ) have suggested that luminous quasars may have dramatic effects on their environments via the radiation or hot winds that the quasar emits .
the continuum images shown in this paper do not provide obvious evidence for the effects of the quasar on the host environment .
in fact , the three spiral host galaxies and several of the host ellipticals appear remarkably normal . broad band colors and spectroscopic observations are required in order to determine more sensitively whether the host galaxy is really oblivious to the presence of the luminous quasar in its center .
the theoretical modeling can now be made more specific and compared with the results of the hst observations for individual host galaxies .
these studies will be important in constraining the time scale of the quasar phenomenon and the isotropy of the quasar emissions .
if a quasar shines brightly for only a short period of time or if the emission is highly anisotropic , then the lack of a dramatic effect of the bright agn on the surrounding medium may be more easily understood .
one of the key results that is apparent in figures [ figallqso ] and [ figallqsosub ] is the detailed evidence for gravitational interactions among the systems `` caught in the act '' ( 0316@xmath24346 , pg 1012 + 008 , and pks 2349@xmath24014 ) .
dynamical modeling of these systems could provide insights into the processes involved in the formation and fueling of quasars .
@xmath1 images provide detailed quantitative information about the environments in which the quasar phenomenon occurs .
we hope to increase our sample in the future and to obtain color information about the objects discussed here
. it should be feasible to obtain spectra of the brightest h ii regions in the spiral hosts of pg 0052 + 251 , pg 1309 + 355 , and pg 1402 + 261 .
detailed comparisons between the @xmath1 and ground - based images will be very informative .
we are grateful to p. crane , m. fall , o. gnedin , k. kellermann , k. mcleod , j. miller , g. neugebauer , m. rees , g. rieke , and m. strauss for valuable discussions , comments , and suggestions .
j. krist and c. burrows constructed the four stellar psfs used in this work ; we are grateful to them for their skill and generosity in providing these data .
we would like to thank digital equipment corporation for providing the dec4000 axp model 610 system used for the computationally intensive parts of this project .
this work was supported in part by nasa contract nag5 - 1618 , nag5 - 3259 , nasa grant number nagw-4452 and grant number go-5343 from the space telescope science institute , which is operated by the association of universities for research in astronomy , incorporated , under nasa contract nas5 - 26555 .
we have used the nasa / ipac extragalactic database ( ned ) , operated by the jet propulsion laboratory , caltech , under contract with nasa , and nasa s astrophysics data system abstract service ( ads ) .
bahcall , n. , & bahcall , j.n .
1970 , , 82 , 1276 bahcall , j. n. , et al .
1993 , , 87 , 1 bahcall , j. n. , kirhakos , s. , & schneider , d. p. 1994
, , 435 , l11 ( paper i ) bahcall , j. n. , kirhakos , s. , & schneider , d. p. 1995a
, , 450 , 486 ( paper ii ) bahcall , j. n. , kirhakos , s. , & schneider , d. p. 1995b
, , 447 , l1 ( paper iii ) bahcall , j. n. , kirhakos , s. , & schneider , d. p. 1995c
, , 454 , l175 ( erratum ) bahcall , j. n. , kirhakos , s. , & schneider , d. p. 1996
, , 457 , 557 bahcall , j. n. , kirhakos , s. , schneider , d. p. , davis , r.j . , muxlow , t. w. b. , garrington , s. t. , conway , r. g. , & unwin , s.c .
1995 , , 452 , l91 ( paper iv ) begelman , m. c. 1985 , , 297 .
492 boroson , t. a. , person , s. e. , & oke , j. b. 1985 , , 293 , 120 bruzual a. , g. , & charlot , s. 1993 , , 405 , 538 burrows , c. j. 1994 , hubble space telescope wide field and planetary camera 2 instrument handbook , version 2.0 ( baltimore : space telescope science institute ) chang , c. a. , schiano , a. v. r. , and wolfe , a. m. 1987 , , 322 , 180 de vaucouleurs , g. 1948 , ann .
dap . , 11 , 247 disney , m.j . ,
boyce , p.j . , blades , j.c . ,
boksenberg , a. , crane , p .
, deharveng , j.m . , macchetto , f. , mackay , c.d . , sparks , w.b . , & phillips , s. 1995 , nature , 376 , 150 dunlop , j. s. , taylor , g. l. , hughes , d. h. , & robson , e. i. 1993 , , 264 , 455 efstathiou , g. , ellis , r. s. , & peterson , b. a. 1988 , , 232 , 431 falle , s. a. e. g. , perry , j. j. , and dyson , j. e. 1981 , , 195 , 397 fisher , k. b. , bahcall , j. n. , kirhakos , s. , & schneider , d. p. 1996
, , in press gehren , t. , fried , j. , wehinger , p.a . , & wyckoff , s. 1984 , , 278 , 11 fukugita , m. , shimasaku , k. , & ichikawa , t. 1995 , , 107 , 945 heckman , t. m. , bothun , g. d. , balick , b. , & smith , e. p. , 1984 , 89 ,
958 hickson , p. , & hutchings , j. b. 1987 , , 312 , 518 hoessel , j. g. , & schneider , d. p. 1985
, , 90 , 1648 holtzman , j. a. , et al . 1995a , , 107 , 156 holtzman , j. a. , et al .
1995b , , 107 , 1065 hutchings , j.b .
1987 , , 320 , 122 hutchings , j.b . ,
crampton , d. , campbell , b. , gower , a.c .
, & morris , s.c .
1982 , , 262 , 48 hutchings , j. b. , janson , t. , & neff , s. g. 1989 , , 342 , 660 hutchings , j. b. & neff , s. g. 1992 , , 104 , 1 hutchings , j. b. & morris , s. 1995 , , 109 , 1541 hutchings , j. b. holtzman , j. , sparks , w.b . ,
morris , s.c . ,
hanisch , r.j .
, & mo , j. 1994 , , 429 , l1 jannuzi , b.t . , et al .
1997 , in preparation kellermann , k.i . ,
sramek , r.a . ,
schmidt , m. , green , r.f . , & shaffer , d.b .
, 1994 , , 108 , 1163 .
kirshner , r. f. , oemler , a. , schechter , p. l. , & shectman , s. a. 1983 , aj , 88 , 1285 krist , j. , & burrows , c. 1994 , wfpc2 instrument science report 94 - 01 ( baltimore : space telescope science institute ) krist , j. , & burrows , c. 1996 , http://www.sns.ias.edu/jnb ( go to hst images ) .
kristian , j. 1973 , , 179 , l61 malkan , m.a . ,
margon , b . , & chanan , g.a .
1984 , , 280 , 66 mcleod , k. k. , 1996 , private communication .
mcleod , k. k. , & rieke , g. h. 1994a , , 420 , 58 mcleod , k. k. , & rieke , g. h. 1994b , , 431 , 137 mcleod , k. k. , & rieke , g. h. 1995 , , 454 , l77 montgomery , k. a. , marschall , l. a. , & janes , k. a. 1993 , , 106 , 181 neugebauer , g. , matthews , k. , & armus , l. 1995 , , 455,l123 postman , m. , & lauer , t. r. 1995 , , 440,28 romanishin , w. , & hintzen , p. 1989
, , 341 , 41 sandage , a. 1961 , the hubble atlas of galaxies , publication 618 , carnegie institution of washington , washington , d.c .
schechter , p. 1976
, , 203 , 297 schweizer , f. 1996 , , 111 , 109 smith , e. p. , heckman , t. m. , bothun , g. d. , romanishin , w. , & balick , b. 1986 , , 306 , 64 stockton , a. 1978 , , 223 , 747 stockton , a. 1980 iau simposium 92 , page 89 stockton , a. 1982 , , 257 , 33 stockton , a. , & mackenty , j. w. 1987 , , 316 , 584 stockton , a. , ridgway , s. , & kellogg , m. 1996 , in preparation trauger , j. t. , et al . 1994 , , 435 , l3 toomre , a. , 1995 ( private communication ) tyson , j. a. , baum , w. a. , & kreidl , t. , 1982 , 257 , l1 vron - cetty , m. p. & woltjer , j. 1990 , , 236,69 , vron - cetty , m. p. , & vron , p. 1991
, a catalogue of quasars and active nuclei ( fifth edition ) , eso scientific report no .
10 , european southern observatory , garching bei munchen , germany vron - cetty , m. p. , & vron , p. 1996 , a catalogue of quasars and active nuclei ( seventh edition ) , eso scientific report no .
17 , european southern observatory , garching bei munchen , germany weymann , r. j. , scott , j. s. , schiano , a. v. r. , and christiansen , w. a. , 262 , 497 wisotzki l. 1994 , , 292 , 45 wisotzki l. , hagen , h. , hopp u. , & reimer d. 1991 , physics of active galactic nuclei , eds .
duschl , s.j .
wagner , springer - verlag , p.438 wyckoff , s. wehinger , p. a. , & gehren t. , 1981 , 247 , 750 yee , h.k.c .
1987 , , 94 , 1461 lcccccccclc pg 0052@xmath45251 & 05 dec 94 & 1400 & 0.114 & 0.155 & 1.75 & 15.4 & @xmath2423.0 & & & phl 909 & 17 oct 94 & 1400 & 0.136 & 0.171 & 1.88 & 15.7 & @xmath2422.9 & & & nab 0205 + 02 & 26 oct 94 & 1400 & 0.139 & 0.155 & 1.75 & 15.4 & @xmath2423.0 & & & 0316@xmath24346 & 20 nov 94 & 1400 & 0.075 & 0.265 & 2.55 & 15.1 & @xmath2424.5 & & & pg 0923@xmath45201 & 23 mar 95 & 1400 & 0.136 & 0.190 & 2.04 & 15.8 & @xmath2423.1 & & & pg 0953@xmath45414 & 03 feb 94 & 1100 & 0.110 & 0.239 & 2.38 & 15.3 & @xmath2424.1 & & k & pks 1004@xmath45130 & 26 feb 95 & 1400 & 0.150 & 0.240 & 2.39 & 15.2 & @xmath2424.2&x & & pg 1012@xmath45008 & 25 feb 95 & 1400 & 0.129 & 0.185 & 2.00 & 15.6 & @xmath2423.2 & & & he 1029@xmath24140 & 06 feb 95 & 1400 & 0.100 & 0.086 & 1.22 & 13.9 & @xmath2423.2 & & & pg 1116@xmath45215 & 08 feb 94 & 1100 & 0.128 & 0.177 & 1.93 & 14.7 & @xmath2424.0 & & k , o & pg 1202@xmath45281 & 08 feb 94 & 1100 & 0.116 & 0.165 & 1.83 & 15.6 & @xmath2423.0 & & k & 3c 273 & 05 jun 94 & 1100 & 0.151 & 0.158 & 1.78 & 12.9 & @xmath2425.6&x&k , o & pks 1302@xmath24102 & 09 jun 94 & 1100 & 0.136 & 0.286 & 2.67 & 15.2 & @xmath2424.6&x&k & pg 1307@xmath45085 & 05 apr 94 & 1400 & 0.129 & 0.155 & 1.75 & 15.1 & @xmath2423.3 & & & pg 1309@xmath45355 & 26 mar 95 & 1400 & 0.088 & 0.184 & 1.99 & 15.6 & @xmath2423.2 & & o & pg 1402@xmath45261 & 07 mar 95 & 1400 & 0.089 & 0.164 & 2.13 & 15.5 & @xmath2423.0 & & & pg 1444@xmath45407 & 27 jun 94 & 1100 & 0.071 & 0.267 & 2.56 & 15.7 & @xmath2423.9 & & k , o & 3c 323.1 & 09 jun 94 & 1100 & 0.082 & 0.266 & 2.55 & 16.7 & @xmath2422.9&x & & pks 2135@xmath24147 & 15 aug 94 & 1400 & 0.174 & 0.200 & 2.11 & 15.5 & @xmath2423.5&x&o & pks 2349@xmath24014 & 18 sep 94 & 1400 & 0.161 & 0.173 & 1.90 & 15.3 & @xmath2423.4&x & & lrcclc pg 0052@xmath45251 & 20 & 35 & 25.1 & sb & phl 909 & 19 & 36 & 24.7 & e4 & nab 0205@xmath4502 & 9 & 16 & 25.0 & s0 ? & 0316@xmath24346 & 17 & 43 & 25.6 & complex , interaction & pg 0923@xmath45201 & 13 & 27 & 24.8 & e1 & pg 0953@xmath45414 & 11 & 26 & 25.8 & faint , tail ? & pks 1004@xmath45130 & 15 & 36 & 25.0 & e2 & pg 1012@xmath45008 & 18 & 36 & 25.0 & interacting galaxies & he 1029@xmath24140 & 40 & 49 & 26.0 & e1 & pg 1116@xmath45215 & 14 & 27 & 25.1 & e2 & pg 1202@xmath45281 & 10 & 18 & 24.7 & e1 , bright companion & 3c 273 & 29 & 52 & 25.4 & e4 & pks 1302@xmath24102 & 15 & 40 & 25.2 & e4 ( ? ) two close companions & pg 1307@xmath45085 & 9 & 16 & 24.6 & faint e1 ( ? ) & pg 1309@xmath45355 & 18 & 36 & 25.7 & sab & pg 1402@xmath45261 & 14 & 30 & 25.6 & sbb & pg 1444@xmath45407 & 10 & 26 & 25.2 & e1 ( ? ) & 3c 323.1 & 11 & 28 & 25.1 & e3 ( ? )
( bright companion ) & pks 2135@xmath24147 & 15 & 32 & 24.8 & e1 ( companions ) pks 2349@xmath24014 & 21 & 40 & 24.5 & complex , interacting & lrrrcccc pg 0052@xmath45251 & 1.0 & 10.0 & 17.1 & @xmath2421.3 & @xmath240.31 & @xmath2421.0 phl 909 & 1.0 & 10.0 & 17.5 & @xmath2421.1 & @xmath240.41 & @xmath2420.7 nab 0205@xmath4502 & 1.0 & 4.5 & 19.5 & @xmath2418.9 & @xmath240.31 & @xmath2418.6 0316@xmath24346 & 1.0 & 11.5 & 18.2 & @xmath2421.4 & @xmath240.47 & @xmath2420.9 pg 0923@xmath45201 & 1.0 & 6.5 & 18.3 & @xmath2420.6 & @xmath240.42 & @xmath2420.2 pg 0953@xmath45414 & 1.0 & 5.5 & 19.1 & @xmath2420.3 & @xmath240.38 & @xmath2419.9 pks 1004@xmath45130 & 1.0 & 7.5 & 17.7 & @xmath2421.7 & @xmath240.48 & @xmath2421.2 pg 1012@xmath45008 & 1.0 & 4.5 & 17.8 & @xmath2421.0 & @xmath240.39 & @xmath2420.6 he 1029@xmath24140 & 1.5 & 20.0 & 16.5 & @xmath2420.6 & @xmath240.35 & @xmath2420.2 pg 1116@xmath45215 & 1.0 & 8.0 & 17.7 & @xmath2421.0 & @xmath240.41 & @xmath2420.6 pg 1202@xmath45281 & 1.0 & 5.0 & 18.6 & @xmath2420.0 & @xmath240.40 & @xmath2419.6 3c 273 & 2.0 & 15.0 & 16.6 & @xmath2421.9 & @xmath240.40 & @xmath2421.5 pks 1302@xmath24102 & 1.0 & 7.5 & 18.4 & @xmath2421.4 & @xmath240.50 & @xmath2420.9 pg 1307@xmath45085 & 1.0 & 4.5 & 18.7 & @xmath2419.7 & @xmath240.37 & @xmath2419.3 pg 1309@xmath45355 & 1.0 & 9.0 & 17.4 & @xmath2421.4 & @xmath240.35 & @xmath2421.0 pg 1402@xmath45261 & 1.0 & 7.5 & 18.1 & @xmath2420.4 & @xmath240.30 & @xmath2420.1 pg 1444@xmath45407 & 1.0 & 5.0 & 18.8 & @xmath2420.8 & @xmath240.47 & @xmath2420.3 3c 323.1 & 1.0 & 5.5 & 18.9 & @xmath2420.7 & @xmath240.47 & @xmath2420.2 pks 2135@xmath24147 & 1.0 & 7.5 & 18.1 & @xmath2420.9 & @xmath240.45 & @xmath2420.4 pks 2349@xmath24014 & 1.0 & 12.0 & 16.7 & @xmath2422.0 & @xmath240.41 & @xmath2421.6 lrcccccccccc pg 0052@xmath45251 & 16.1 & 4.7 & 16.8 & 1.4 & & 16.7 & 1.8 & 17.2 & 1.3 phl 909 & 16.7 & 2.5 & 17.4 & 1.0 & & 17.2 & 2.3 & 17.6 & 1.5 nab 0205@xmath4502 & 18.0 & 0.6 & 18.7 & 0.6 & & 18.4 & 0.7 & 19.0 & 0.7 0316@xmath24346 & 17.2 & 3.4 & 18.0 & 1.1 & & 17.8 & 2.1 & 18.3 & 1.2 pg 0923@xmath45201 & 17.3 & 2.5 & 18.0 & 1.0 & & 17.5 & 2.9 & 18.2 & 1.3 pg 0953@xmath45414 & 17.9 & 2.3 & 18.5 & 1.1 & & 18.2 & 1.8 & 18.8 & 1.1 pks 1004@xmath45130 & 16.7 & 1.6 & 17.3 & 0.9 & & 16.9 & 1.2 & 17.5 & 1.0 pg 1012@xmath45008 & 16.3 & 6.2 & 17.3 & 1.4 & & 17.0 & 3.4 & 17.7 & 1.6 he 1029@xmath24140 & 15.9 & 2.8 & 16.4 & 1.5 & & 16.2 & 3.2 & 16.7 & 1.9 pg 1116@xmath45215 & 16.6 & 1.9 & 17.3 & 1.0 & & 16.9 & 1.4 & 17.5 & 1.2 pg 1202@xmath45281 & 17.4 & 1.5 & 18.1 & 0.9 & & 17.7 & 1.4 & 18.3 & 1.0 3c 273 & 15.6 & 2.3 & 16.2 & 1.3 & & 16.0 & 3.7 & 16.7 & 1.6 pks 1302@xmath24102 & 17.1 & 2.6 & 17.8 & 1.1 & & 17.7 & 1.4 & 18.2 & 1.1 pg 1307@xmath45085 & 17.4 & 1.8 & 18.1 & 0.9 & & 17.8 & 1.3 & 18.4 & 1.0 pg 1309@xmath45355 & 16.4 & 2.8 & 17.1 & 1.1 & & 16.8 & 2.0 & 17.3 & 1.2 pg 1402@xmath45261 & 16.9 & 2.2 & 17.6 & 1.0 & & 17.6 & 1.5 & 18.3 & 1.6 pg 1444@xmath45407 & 17.6 & 1.3 & 18.2 & 1.0 & & 17.8 & 1.3 & 18.4 & 1.0 3c 323.1 & 17.8 & 1.4 & 18.4 & 0.9 & & 18.1 & 1.6 & 18.7 & 1.0 pks 2135@xmath24147 & 17.2 & 2.0 & 17.8 & 1.1 & & 17.4 & 2.6 & 18.0 & 1.3 pks 2349@xmath24014 & 15.9 & 5.6 & 16.6 & 1.6 & & 16.2 & 4.8 & 16.8 & 2.5 lcccccc pg 0052@xmath45251 & @xmath2421.3 & disk & & @xmath2420.9 & disk & phl 909 & @xmath2421.5 & gdv & & @xmath2421.0 & gdv & nab 0205@xmath4502 & @xmath2419.4 & disk & & @xmath2419.1 & disk & 0316@xmath24346 & @xmath2421.1 & disk & & @xmath2420.8 & disk & pg 0923@xmath45201 & @xmath2421.2 & gdv & & @xmath2421.0 & gdv & pg 0953@xmath45414 & @xmath2420.5 & disk & & @xmath2420.2 & disk & pks 1004@xmath45130 & @xmath2422.5 & gdv & & @xmath2422.0 & gdv & pg 1012@xmath45008 & @xmath2422.1 & gdv & & @xmath2420.7 & disk & he 1029@xmath24140 & @xmath2420.8 & gdv & & @xmath2420.5 & gdv & pg 1116@xmath45215 & @xmath2421.7 & gdv & & @xmath2421.4 & gdv & pg 1202@xmath45281 & @xmath2420.8 & gdv & & @xmath2420.5 & gdv & 3c 273 & @xmath2422.5 & gdv & & @xmath2422.1 & gdv & pks 1302@xmath24102 & @xmath2421.5 & disk & & @xmath2421.1 & disk & pg 1307@xmath45085 & @xmath2420.6 & gdv & & @xmath2420.2 & disk & pg 1309@xmath45355 & @xmath2421.3 & disk & & @xmath2421.1 & disk &
pg 1402@xmath45261 & @xmath2420.6 & disk & & @xmath2419.9 & disk & pg 1444@xmath45407 & @xmath2420.9 & disk & & @xmath2420.5 & disk & 3c 323.1 & @xmath2421.3 & gdv & & @xmath2421.0 & gdv & pks 2135@xmath24147 & @xmath2421.3 & gdv & & @xmath2421.1 & gdv & pks 2349@xmath24014 & @xmath2422.4 & gdv & & @xmath2422.1 & gdv & 5inlcrrcrr & number of & & & + & companions & & & & & + pg 0052@xmath45251 & 1 & 14.1 & 24.6 & & 18.8 & @xmath212 + phl 909 & 1 & 12.5 & 23.5 & & 21.4 & @xmath213 + nab 0205@xmath4502 & 1 & 8.3 & 14.5 & & 20.0 & @xmath214 + pg 0923@xmath45201 & 2 & 10.9 & 22.2 & & 19.5 & @xmath215 + & & 11.0 & 22.5 & & 18.0 & @xmath216 + pg 0953@xmath45414 & 1 & 8.2 & 19.6 & & 22.7 & @xmath217 + pg 1012@xmath4500 & 2 & 3.3 & 6.7 & & 17.6 & @xmath218 + & & 6.8 & 13.7 & & 19.0 & @xmath219 + pg 1116@xmath45215 & 1 & 12.3 & 23.8 & & 19.3 & @xmath215 + pg 1202@xmath45281 & 3 & 5.2 & 9.5 & & 18.9 & @xmath220 + & & 8.4 & 15.3 & & 21.5 & @xmath221 + & & 9.6 & 17.5 & & 21.3 & @xmath222 + pks 1302@xmath24102 & 2 & 1.1 & 2.9 & & 20.3 & @xmath223 + & & 2.3 & 6.2 & & 21.5 & @xmath224 + he 1029@xmath225 & 1 & 4.1 & 5.0 & & 20.7 & @xmath226 + 3c 323.1 & 1 & 2.7 & 6.9 & & 20.6 & @xmath227 + pks 2135@xmath24147 & 2 & 1.9 & 3.9 & & 19.5 & @xmath223 + & & 5.5 & 11.7 & & 19.7 & @xmath228 + pks 2349@xmath24014 & 2 & 1.9 & 3.5 & & 20.8 & @xmath229 + & & 11.6 & 22.0 & & 21.4 & @xmath222 + lrcccccc pks 1302@xmath24102 & 19.5 & @xmath2420.3 & @xmath2420.4 & @xmath2420.0 & @xmath2422.0 & @xmath2421.1 & pks 2135@xmath24147 & 18.9 & @xmath2419.9 & @xmath2419.9 & @xmath2419.4 & @xmath2422.2 & @xmath2421.1 & pks 2349@xmath24014 & 17.6 & @xmath2420.7 & @xmath2420.8 & @xmath2420.4 & @xmath2422.5 & @xmath2422.1 & lrrccc pg 0052@xmath45251 & 3.43 & 6.86 & 18.3&@xmath2420.1 & @xmath24 19.8 phl 909 & 3.19 & 6.38 & 18.8&@xmath2419.8 & @xmath24 19.4 nab 0205@xmath4502 & 3.43 & 6.86 & 20.7&@xmath2417.7 & @xmath24 17.4 0316@xmath24346 & 2.80 & 4.71 & 19.8&@xmath2419.8 & @xmath24 19.3 pg 0923@xmath45201 & 2.96 & 5.91 & 19.3&@xmath2419.6 & @xmath24 19.2 pg 0953@xmath45414 & 2.52 & 5.04 & 20.1&@xmath2419.3 & @xmath24 18.9 pks 1004@xmath45130 & 2.51 & 5.02 & 18.9&@xmath2420.5 & @xmath24 20.0
pg 1012@xmath45008 & 3.00 & 6.00 & 18.1&@xmath2420.7 & @xmath24 20.3 he 1029@xmath24140 & 5.56 & 11.11 & 18.0&@xmath2419.1 & @xmath24 18.7 pg 1116@xmath45215 & 3.11 & 6.22 & 19.0&@xmath2419.7 & @xmath24 19.3 pg 1202@xmath45281 & 3.28 & 6.56 & 19.9&@xmath2418.8 & @xmath24 18.4 3c 273 & 3.37 & 6.74 & 17.8&@xmath2420.7 & @xmath24 20.3 pks 1302@xmath24102 & 2.25 & 4.49 & 19.4&@xmath2420.4 & @xmath24 19.9
pg 1307@xmath45085 & 3.43 & 6.86 & 19.9&@xmath2418.5 & @xmath24 18.1 pg 1309@xmath45355 & 3.02 & 6.03 & 18.7&@xmath2420.1 & @xmath24 19.7 pg 1402@xmath45261 & 3.28 & 6.56 & 19.5&@xmath2419.0 & @xmath24 18.7 pg 1444@xmath45407 & 2.34 & 4.69 & 19.7&@xmath2419.9 & @xmath24 19.4 3c 323.1 & 2.35 & 4.71 & 19.9&@xmath2419.7 & @xmath24 19.2 pks 2135@xmath24147 & 2.84 & 5.69 & 19.1&@xmath2419.9 & @xmath24 19.4 pks 2349@xmath24014 & 3.16 & 6.32 & 17.9&@xmath2420.8 & @xmath24 20.4 lcclccr pg 0052@xmath45251 & 15.14 & @xmath2423.3 & 3.90 & @xmath2419.4 & @xmath2420.9 & 1.5 phl 909 & 14.40 & @xmath2424.2 & 3.95 & @xmath2420.3 & @xmath2421.0 & 0.7 pg 0923@xmath45201 & 14.95 & @xmath2423.9 & 4.00 & @xmath2419.9 & @xmath2421.0 & 1.1 pg 0953@xmath45414 & 15.28 & @xmath2424.1 & 4.20 & @xmath2419.9 & @xmath2420.2 & 0.3 pks 1004@xmath4513 & 15.12 & @xmath2424.3 & 4.20 & @xmath2420.1 & @xmath2422.0 & 1.9 pg 1012@xmath4500 & 13.94 & @xmath2424.9 & 4.00 & @xmath2420.9 & @xmath2420.7 & @xmath240.2 pks 2135@xmath24147 & 14.75 & @xmath2424.2 & 4.10 & @xmath2420.1 & @xmath2421.1 & 1.0 pks 2349@xmath24014 & 13.98 & @xmath2424.7 & 3.95 & @xmath2420.8 & @xmath2422.1 & 1.3 lrcccr pg 0052@xmath45251 & 14.46 & 3.16 & @xmath2420.8 & @xmath2420.9 & @xmath2300.1 pg 0923@xmath45201 & 14.86 & 3.22 & @xmath2420.8 & @xmath2421.0 & @xmath2300.2 pg 0953@xmath45414 & 15.38 & 3.32 & @xmath2420.7 & @xmath2420.2 & @xmath240.5 pks 1004@xmath45130 & 14.86 & 3.32 & @xmath2421.2 & @xmath2422.0 & @xmath2300.8 pg 1012@xmath45008 & 14.02 & 3.21 & @xmath2421.6 & @xmath2420.7 & @xmath240.9 pg 1116@xmath45215 & 13.97 & 3.20 & @xmath2421.6 & @xmath2421.4 & @xmath240.2 pg 1202@xmath45281 & 15.07 & 3.18 & @xmath2420.3 & @xmath2420.5 & @xmath2300.2 3c 273 & 13.01 & 3.17 & @xmath2422.3 & @xmath2422.1 & @xmath240.2 pks 1302@xmath24102 & 14.79 & 3.47 & @xmath2421.5 & @xmath2421.1 & @xmath240.4
pg 1307@xmath45085 & 15.24 & 3.16 & @xmath2420.0 & @xmath2420.2 & @xmath2300.2 pg 1309@xmath45355 & 14.55 & 3.21 & @xmath2421.0 & @xmath2421.1 & @xmath2300.1 pg 1402@xmath45261 & 14.95 & 3.18 & @xmath2420.4 & @xmath2419.9 & @xmath240.5 pg 1444@xmath45407 & 15.19 & 3.42 & @xmath2421.0 & @xmath2420.5 & @xmath240.5 3c 323.1 & 14.80 & 3.42 & @xmath2421.4 & @xmath2421.0 & @xmath240.4 ccccc a&23.7&2.8&0.3 & 2.8 b&24.3&2.9&1.5 & 2.5 c&25.3&2.9&2.1 & 2.0 d&25.6&3.5&3.3 & 1.3 e&25.5&4.6&4.5&0.9 f&26.2&4.8&4.5&1.8 g&26.2&5.0&4.2&2.8 h&25.9&5.3&3.8&3.7 i&25.6&4.1&1.1 & 4.0 j&25.2&4.0&2.7 & 3.0 lcccclc pg 0052@xmath45251 & 0.155 & 17.2 & @xmath2420.9 & 1.3 & sb & phl 909 & 0.171 & 17.2 & @xmath2421.0 & 2.3 & e4 & nab 0205@xmath4502 & 0.155 & 19.0 & @xmath2419.1 & 0.7 & s0 ? & 0316@xmath24346 & 0.265 & 18.3 & @xmath2420.8 & 1.2 & inter . &
pg 0923@xmath45201 & 0.190 & 17.5 & @xmath2421.0 & 2.9 & e1 & pg 0953@xmath45414 & 0.239 & 18.8 & @xmath2420.2 & 1.1 & ? & pks 1004@xmath45130 & 0.240 & 16.9 & @xmath2422.0 & 1.2 & e2 & pg 1012@xmath45008 & 0.185 & 17.7 & @xmath2420.7 & 1.6 & inter . & he 1029@xmath24140 & 0.086 & 16.2 & @xmath2420.5 & 3.2 & e1 & pg 1116@xmath45215 & 0.177 & 16.9 & @xmath2421.4 & 1.4 & e2 & pg 1202@xmath45281 & 0.165 & 17.7 & @xmath2420.5 & 1.4 & e1 & 3c 273 & 0.158 & 16.0 & @xmath2422.1 & 3.7 & e4 & pks 1302@xmath24102 & 0.286 & 18.2 & @xmath2421.1 & 1.1 & e4 ? &
pg 1307@xmath45085 & 0.155 & 17.8 & @xmath2420.2 & 1.3 & e1 ? &
pg 1309@xmath45355 & 0.184 & 17.3 & @xmath2421.1 & 1.2 & sab & pg 1402@xmath45261 & 0.164 & 18.3 & @xmath2419.9 & 1.6 & sbb & pg 1444@xmath45407 & 0.267 & 18.4 & @xmath2420.5 & 1.0 & e1 ? & 3c 323.1 & 0.266 & 18.1 & @xmath2421.0 & 1.6 & e3 ? & pks 2135@xmath24147 & 0.200 & 17.4 & @xmath2421.1 & 2.6 & e1 & pks 2349@xmath24014 & 0.173 & 16.2 & @xmath2422.1 & 4.8 & inter . | observations with the wide - field camera of the _ hubble space telescope ( hst ) _ are presented for a representative sample of 20 intrinsically luminous quasars with redshifts smaller than 0.30 .
these observations show that luminous quasars occur in diverse environments that include ellipticals as bright as the brightest cluster galaxies ( 2 ) , apparently normal ellipticals ( 10 ) , apparently normal spirals with h ii regions ( 3 ) , complex systems of gravitationally interacting components ( 3 ) , and faint surrounding nebulosity ( 2 ) .
the quasar host galaxies are centered on the quasar to the accuracy of our measurements , 400 pc .
some of the host galaxies show no evidence of merging or strong gravitational interactions .
there are more radio quiet quasars in galaxies that appear to be ellipticals ( 7 ) than in spiral hosts ( 3 ) , contrary to expectations .
however , three , and possibly five , of the six radio loud quasars have detectable elliptical hosts , in agreement with expectations .
strong upper limits are placed on the possible existence of optical jets .
the luminous quasars studied in this paper occur preferentially in luminous galaxies .
the average absolute magnitude of the hosts is @xmath0 magnitudes brighter than expected for a field galaxy luminosity function .
the superb optical characteristics of the repaired @xmath1 make possible the detection of close galactic companions ; we detect eight companion galaxies within projected distances of 10 kpc from the quasars .
the presence of very close companions , the images of current gravitational interactions , and the higher density of galaxies around the quasars suggest that gravitational interactions play an important role in triggering the quasar phenomenon . and |
there are strong reasons for suspecting that the emitting plasma of @xmath0-ray bursts ( grbs ) is geometrically beamed in a cone .
the energy requirements can then be reduced below the exorbitant levels that isotropic emission would imply ( kulkarni et al .
1999 ) and in most models for the long bursts it is in any case a natural expectation borne out by simulations ( macfadyen & woosley 1999 ; macfadyen , woosley & heger 2001 ) that the relativistic outflow from a central engine should be collimated along a channel that opens up along the rotation axis of a massive star .
although the relativistic mhd that gives rise to jets is uncertain , and surely very complicated , most discussions of the radiation from gamma ray bursts ( and their afterglows ) has postulated a jet with a well - defined angle ( meszaros & rees 1997 ) , though this angle may differ for different bursts ( see , however , meszaros , rees & wijers 1998 ; salmonson 2001 ) .
we discuss here an alternative model where the jet , rather than having a uniform profile out to some definite cone angle , has a `` beam pattern '' where the power per unit solid angle ( and perhaps also the initial lorentz factor ) is maximal along the axis , but drops off gradually away from the axis .
this would be expected if there is mixing and entrainment from the borders of the funnel .
we discuss the expected time - dependence of the afterglow if it is triggered by a jet with this more general profile .
we conclude that the time of the observed break ( which , for a uniform jet viewed along its axis , depends on the cone angle ; see , e.g. , rhoads 1997 ) instead depends on the angle between the line of sight and the symmetry axis : such a jet viewed nearly head - on simulates a narrow uniform jet , whereas the afterglow from the same jet viewed more obliquely would simulate a wider uniform jet .
grb have been proposed recently to be explosions releasing a standard power that can be either injected in very different jet opening angles or distributed within the jet in some universal emission diagram , ( postnov et al .
2001 ) . while frail et al .
( 2001 , hereafter f01 ) , support with their data the first interpretation , we show that their observational results could instead be attributed to a more standard set of objects viewed at different angles to their symmetry axis .
we suppose that all long grbs have jets with a standard opening angle @xmath1 , total kinetic energy and beam profile for @xmath2 .
we consider a relativistic outflow where both the bulk lorentz factor and the energy per unit solid angle depend as power laws motivated by f01 correlation .
this is not , however , a necessary ingredient of the model as discussed in 6 ( see also fig . 3 and 4 ) ] on the angular distance from the center @xmath3 @xmath4 and @xmath5 where @xmath6 is introduced just for formal reasons to avoid a divergence at @xmath7 , but can be taken to be smaller than any other angle of interest . a lower limit to this angle is @xmath8 degrees , where @xmath9 is the maximum value to which the fireball can be accelerated ( piran 1999 ) . the power law index of @xmath10 , @xmath11 , is not important for the dynamics of the fireball and the computation of the light curve as long as @xmath12 and @xmath13 .
nevertheless , it plays a role when we want to calculate the fraction of grbs - afterglow without prompt @xmath0-ray emission , as we discuss in 6 .
consider an observer at an angle @xmath14 with respect to the axis of the jet .
he measures an isotropic equivalent energy @xmath15 from the @xmath0-ray fluence .
if also the afterglow emission is dominated by the component pointing the earth , he will infer @xmath16 as the half - opening angle of the jet by means of the break time in the light curve , @xmath17 ( sari , piran & halpern 1999 ) @xmath18 , where @xmath19 is the external medium density .
since the total energy inferred from all viewing angles is @xmath20const , the observer will derive the same conclusions obtained by f01 and panaitescu & kumar ( 2001 , hereafter pk01 ) of grbs as fireballs with the same total kinetic energy but very different jet apertures . to evaluate how the contributions of the other components add to the light curve from the zone with @xmath21 we calculate when their beamed emission include the observer direction , ( when @xmath22 ) and which energy per unit solid angle they have at that time compared to @xmath23 .
we show that under the assumptions of eq .
[ eq : top ] the afterglow light curve is indeed dominated by the fireball element along the line of sight : neither the `` core '' of the jet with @xmath24 nor the regions with @xmath25 make substantial contributions . for an effective though simplified discussion
we consider only three components of the jet corresponding to @xmath26 , @xmath27 and @xmath28 .
we call them `` cone @xmath29 '' , where @xmath30 , respectively
. this approach is justified by the fact that only the very inner parts of the jet with @xmath31 and the much wider ( @xmath32 ) outer parts could contribute and substantially modify the light curve of cone @xmath33 .
we approximate cone @xmath34 as a relativistic source moving at an angle @xmath16 with respect the observer , while the observer is approximately considered on the symmetry axis for cones @xmath33 and @xmath35 .
the cones , that are not causally connected ( @xmath36 ) , evolve independently and adiabatically in a constant density medium and spread sideways when @xmath37 drops below @xmath38 ( rhoads 1997 ) .
we consider relativistic lateral expansion of the cones so that their geometrical angles @xmath39 grow as @xmath40 since the dynamics of the fireball components changes at @xmath41 , also @xmath10 decreases with time differently before and after the break time . for @xmath42 @xmath43 where @xmath44 is the deceleration time at which @xmath45 and the jet begins to decelerate significantly . from eq .
[ eq : gamma ] to eq .
[ eq : tb ] we drop the subscript @xmath29 for an easier reading . for @xmath46 @xmath47 this on - axis calculation is valid for all @xmath48 , because @xmath49 soon after the break time and the emitting plasma enters the line of sight : from this moment we can consider the observer to be on the cone axis .
the lateral expansion starts at a time @xmath50 where the deceleration time , @xmath44 , depends on @xmath51 and consequently @xmath52 therefore from eq .
[ eq : top ] it follows that @xmath53=constant and @xmath54 cones @xmath33 and @xmath35 are hollow but they spread only outwards because the inner components which have already expanded ( for cone @xmath35 ) or are about to do so ( for cone @xmath33 ) have higher pressure .
[ eq : tb ] shows also that when @xmath55 and the part of the blast generated by the core of the jet becomes visible to the observer , cone @xmath33 has just started spreading .
the contribution of region @xmath34 to the light curve of the cone with @xmath16 is not dominant because when @xmath56 the energy per unit solid angle is comparable to @xmath57 @xmath58 the regions with @xmath59 spread at later time , so when @xmath60 the energy per solid angle of cone @xmath33 has already decreased with time but it remains always comparable to @xmath61 @xmath62 where @xmath63 is the break time for cones @xmath29 and we used eq .
[ eq : gammatb ] and eq .
[ eq : tb ] .
therefore the regions with @xmath64 , ( or those with @xmath3 substantially above @xmath16 ) , do nt determine the overall shape of the light curve , because they are hidden by the dominant emission from the component along the line of sight . in principle , the superposition of many components of energy given by eq .
[ eq:9 ] and [ eq : dtb ] may give rise to a sizable contribution .
as shown by the more detailed calculation of 3.2
( see fig 2 ) , this is not the case and the only effect is to delay by a factor of 2 the time of the break .
consequently the energy per unit solid angle that is measured modeling the afterglow is @xmath57 and the time break depends only on the viewing angle @xmath16 .
in order to compute a reasonably accurate lightcurve in our model , we adopt the following approach .
we divide the inhomogeneous jet in @xmath65 hollow cones ( all but the very central one ) , each characterized by energy per unit solid angle and lorentz factor given by eq .
[ eq : top ] .
we compute an approximate lightcurve for each sub - jet using three asymptotic behaviors .
for the cones with @xmath66 and @xmath67 we adopt the @xmath24 and @xmath25 approximations described in 2 , .
for the cones defined by @xmath68 which point towards the observer since the beginning , we again consider the observer along the symmetry axis , but the emission is calculated in a filled cone geometry , instead of the hollow one adopted for the other cases .
for @xmath69 we use the usual afterglow theory to perform the lightcurve calculations , while for @xmath66 we generalize it for an off - axis observer as explained in the following section . consider a uniform jet with lorentz factor @xmath10 and initial half - aperture @xmath70 .
the radiative process is synchrotron emission ( meszaros & rees 1997 ) and we concentrate on the power law branch of the spectrum between the peak , @xmath71 , and the cooling , @xmath72 , frequencies . the observed flux at a frequency @xmath73 , time @xmath74 and viewing angle @xmath16 is @xmath75 where @xmath76 is the emitting area and @xmath77 is the relativistic doppler factor .
@xmath78 is the comoving intensity at the comoving frequency @xmath79 and at the comoving time @xmath80 @xmath81 due to the relativistic beaming , the observed flux depends on the observer angle . in particular , if @xmath82 @xmath83 while for @xmath84 @xmath85 in this case the jet is initially a source moving at an angle @xmath16 with the observer . at the break time @xmath86
and then it starts decreasing very fast ( eq .
[ eq : gamma ] ) until @xmath87 and we can again use the approximation that the jet is viewed on - axis .
so the intensity , the change in the slope and the break time of the afterglow depend on the observer viewing angle . in fig .
[ fig : jetto ] we show the resulting off - axis lightcurve for a jet with opening angle @xmath88 as viewed with off - axis angles @xmath89 , @xmath90 , @xmath91 and @xmath92 degrees .
the flat part of the lightcurves corresponds to the time interval between the beginning of the jet spreading and the time in which the jet enters the line of sight . in this time interval
our approximations are no longer valid and we simply model the lightcurve with a flat component connecting earlier and later times .
this crude approximations is likely responsible for the slight flattening of the final lightcurves ( see below and fig . [
fig : jetti ] ) just before the break .
using eq .
[ eq : fon ] and eq .
[ eq : foff ] we can compute the total light curve as a sum over the fireball components according to the scheme described at the beginning of this section .
@xmath93\ ; + \\ \sum_{\theta_{o}-\frac{1}{\gamma } < \theta<\theta_{o}+\frac{1}{\gamma } } \big [ f_{on}\,\,\frac{1}{n}\big ] + \sum_{\theta>\theta_{o}+\frac { 1}{\gamma}}\big [ f_{on}-f_{on}(\theta_{in})\big ] , \end{array } \label{eq : lc}\ ] ] where @xmath94 is the inner edge of the cone and @xmath95 is the light curve for a jet with @xmath96 , @xmath97 and half - opening angle @xmath94 .
n is the number of jet components that are beamed around the line of sight at small times .
the results of this calculation are shown in fig .
[ fig : jetti ] for different viewing angles . on top of the solid lines we plot with a dashed line the lightcurve of a homogeneous fireball with @xmath98 and @xmath99 .
this shows that , for @xmath100 , the lightcurve of an inhomogeneous jet can be successfully modelled with the lightcurve expected for a homogeneous jet : @xmath101 for @xmath102 and @xmath103 for @xmath104 .
one of the main simplifications that we made in eq .
[ eq : lc ] is to assume that the observer is on the jet axis of the cones with @xmath105 .
the main consequence is to predict a transition between the two power - law branches sharper than what we expect from the exact integration .
this is due to the fact that , even in a uniform jet , off - axis observers sees smoother transitions ( see fig . 4 of ghisellini & lazzati 1999 ) .
recently an important observational result on the energetic content of grb < s was published by pk01 and f01 .
they found an anti - correlation between the isotropic equivalent energy , @xmath106 and the break time in the afterglow lightcurves .
f01 in particular derived @xmath107 from gamma ray fluences , @xmath108 and their data are consistent with @xmath109 .
they explain it in the framework of collimated , uniform , lateral spreading jets interacting with a constant , low density ( 0.1 @xmath110 ) , external medium .
the observer is postulated to be along the jet axis .
they assume that the afterglow emission steepens because at that time the lorentz factor has dropped to @xmath111 , so that the on - axis observer sees the edge of the jet and the lateral jet spreading becomes important . in this case
@xmath112 is predicted , which matches observation .
they convert the observed break times in jet opening angles through the formulation of sari et al .
the @xmath0-ray energy measured and corrected for the inferred geometry of the jet is clustered around @xmath113 erg .
they concluded that grbs central engines release the same amount of energy through jets with very different opening angles . in this framework , then , the wide distributions in kinetic energy per unit solid angle , which spans 3 orders of magnitude , is due only to the distribution of jet solid angles .
pk01 obtained the same results , modeling a subset of multiwavelength afterglows from which they could assign an external density , a time break and an equivalent isotropic energy for the fireball .
both f01 and pk01 found geometric angles that span an order of magnitude but strongly concentrate around @xmath114 in the framework of our model an alternative explanation of the observed relation between @xmath107 and @xmath17 can be found . as discussed in
2 and shown in fig .
[ fig : jetti ] , we can infer from observations only the properties of the cone pointing towards the observer and not those of the whole jet .
so @xmath115 and using eq .
[ eq : tb ] and eq .
[ eq : top ] we obtain @xmath116 then , the observed distribution of @xmath107 and its relation with @xmath17 is due to an inhomogeneous jet and the possibility to view it from different angles .
under our assumptions , each @xmath0-ray luminosity corresponds to a particular viewing angle . the probability to see a jet between @xmath3 and @xmath117
is given by @xmath118 @xmath119 and the highest probability is for @xmath120 .
therefore it is highly improbable to see a jet on axis .
consequently we expect more faint grbs then very luminous ones according to a luminosity function @xmath121 where @xmath122 .
since there are only a small number of grbs with observed red - shift , the comparison of our predicted luminosity function with data is far from being definitive .
recently bloom at al .
( 2001 ) published an histogram of the bolometric k - corrected prompt energies for 17 grbs .
the distribution is roughly flat from @xmath123 to @xmath124 erg but as the authors emphasize this analysis applies only to observed grbs with redshifts and several observational biases obscure the true underlying energy distribution . the main bias that overcasts faint grbs is the detection threshold of the instruments : this sample is thus flux and not volume limited .
moreover redshift determination encounters more problem for faint grbs .
based on @xmath125-hardness correlation , schmidt ( 2001 ) derived a luminosity function without using any redshift .
they had to assume how the comoving grbs rate varies with redshift and they based their calculations on three star forming rate models . in this case
a power - law luminosity function was derived , but flatter than the @xmath126 predicted in the simplest version of our model . a luminosity function
not based on assumed burst rate evolution can be derived by measuring the burst distance scale through the recently discovered variability - luminosity relation ( fenimore & ramirez - ruiz 2000 , reichart et al .
2001 ) . a cumulative analysis of a sample of 220 bursts ( fenimore & ramirez - ruiz 2000 ) yielded a power - law luminosity function @xmath127 , which compare more favorably with our prediction .
more recently , lloyd - ronning , fryer & ramirez - ruiz ( 2001 ) find that the typical luminosity of grbs evolves with redshift . as a consequence
, a flatter power - law index @xmath128 is obtained .
we stress , however , that our deduced luminosity function can be altered by distributions in total energy or geometric angles or by luminosity evolution with redshift . given the somewhat contradictory observational results discussed above , we conclude that more accurate spectral and fluence measures and a larger sample of bursts are needed for a proper comparison .
we considered inhomogeneous grbs jets with a standard total energy , opening angle and local energy distribution , @xmath129 .
we show that this jet structure can reproduce the observed correlation between isotropic energy and break - time . in this model
both measurements depend only on the viewing angle because the @xmath0-ray fluence and the afterglow emission are dominated by the components of the jet pointing towards the observer at small times .
since all cones have the same total energy @xmath130 , we recover the results of f01 and pk01 but the constrains on the geometrical beaming can be relaxed and an appealing more standard structure for all grbs can be adopted .
the jet total energy can be calculated from eq .
[ eq : top ] @xmath131 and compared to @xmath132 , the total energy inferred from observation , @xmath133 .
to give an example , for a fireball with @xmath134 and @xmath135 , we have @xmath136 , i.e. the true energy of the fireball can be one order of magnitude larger than what inferred with the models of f01 and pk01 . in addition
( 5 ) we can derive the grbs luminosity function from the probability distribution of the viewing angle and compare it to data .
the comparison is , at this stage , still uncertain because more accurate spectral and fluence measures are required to build a volume limited sample and confirm ( or rule out ) this model . in any fireball model
considering a jet - like grb structure , a certain number of afterglows without @xmath0-ray emission ( orphan afterglows ) is expected . for an homogeneous jet ,
orphan afterglows are possible only for viewing angles greater than @xmath137 . in our jet configuration
a fraction of the total area could have a lorentz factor lower than the minimum @xmath10 necessary for @xmath0-ray radiation .
consequently , considering the same opening angle , an inhomogeneous jet could produce an higher fraction of orphan afterglows than an homogeneous one .
this intrinsic fraction depends above all on the @xmath10 distribution within the jet ( @xmath11 in eq [ eq : gammadis ] ) and on the minimum lorentz factors to produce @xmath0-ray radiation and afterglow emission .
the observed number of orphan afterglows depends also on flux detection limits , grb explosion rates with redshift and cosmology .
an accurate calculation of the expected orphan afterglow rates is therefore beyond the scope of this paper .
we emphasize that we implicitly assumed the radiation efficiency of the fireball to be weakly dependent on @xmath10 or @xmath138 .
this is a plausible assumption in internal - shocks scenario .
if it was not the case and the efficiency grew with @xmath138 , a different relation between @xmath138 and @xmath3 should be postulated in order to reproduced observations . in this paper we concentrated for simplicity on a beam profile @xmath139 which is consistent observational results but it is interesting to briefly discuss other power - law relations @xmath140 ( see fig . [
fig : alfae ] ) . a decay flatter
then 2 would cause two breaks in the light curve : the first due to the cone pointing the observer when @xmath141 and the second , at later times , when the observer sees the edge of the jet and @xmath142 .
the power law index after the first break is flatter then @xmath143 because the cones with @xmath144 enter the line of sight with @xmath145 and substantially modify the light curve shape . with a steeper decay in the distribution of @xmath138 the time break and the emission after that break
would be dominated by the jet along the axis rather then by the very much weaker part directed to the observer .
the jet break would then be preceded by a prominent flattening in the lightcurve , especially for @xmath146 , difficult to reconcile with observations . in this case
we would have @xmath0-ray emission only for very small angles and the number of orphan afterglows would be much greater then expected from the @xmath147 model . from eq .
[ eq : tb ] we can derive the relation between the index @xmath148 and @xmath149 ( where @xmath150 ) . we obtain @xmath151 for @xmath152 and @xmath153 for @xmath154 , when the first break is considered . in fig .
[ fig : indi ] we plot this relation overlaid on the interval in @xmath149 allowed by observations ( we used f01 data ) .
we derive @xmath155 , at the @xmath156 level . further @xmath0-ray and afterglow observations will allow to constrain this parameter much better in the future .
besides the luminosity function discussed in 5 , there are several ways in which this model can be proved or disproved .
first , we have shown that the real total energy of the fireball can easily be an order of magnitude larger than what estimated by pk01 and f01 . in this case , after the fireball has slowed down to mild - relativistic and sub - relativistic speed , radio calorimetry ( frail , waxman & kulkarni 2000 ) should allow us to detect the excess energy .
in addition , in this model we naturally predict that the more luminous part of the fireball have higher lorentz factors .
this may help explaining the detected luminosity - variability ( fenimore & ramirez - ruiz 2000 ) and luminosity - lag ( norris , marani & bonnell 2000 ) correlations ( salmonson 2000 , kobayashy , ryde & macfadyen 2001 ; ramirez - ruiz & lloyd - ronning 2002 ) .
another constrain is given by polarization . since
fireball anisotropy is a basic ingredient of this model , inducing polarization ( ghisellini & lazzati 1999 ; see also sari 1999 and gruzinov & waxman 1999 ) .
the time evolution of the polarized fraction and of the position angle are however different from a uniform jet ( rossi et al . in preparation ) .
finally , the properties of the bursts should not depend on the location of the progenitor in the host galaxy , and therefore this model can accommodate the marginal detection of an @xmath157-offset relation ( ramirez - ruiz , lazzati & blain 2001 ) only if a distribution of @xmath70 is considered .
we thank g. ghisellini , j. granot , p. kumar , a. panaitescu and e. ramirez - ruiz for many stimulating discussions .
er and dl thank the ias , princeton , for the kind hospitality during the final part of the preparation of this work .
er thanks the isaac newton and pparc for financial support . bloom j. s. , frail d. a. , sari r. , 2001 , aj , 121 , 2879 fenimore e. e. , ramirez - ruiz e. , 2000 , apj submitted ( astro - ph/0004176 ) frail d. a. , waxman e. , kulkarni s. r. , 2000 , apj , 537 , 191 frail d. a. , et al . , 2001 , apj , 562 , l55 ghisellini g. , lazzati d. , 1999 , mnras , 309 , l7 gruzinov a. , waxman e. , 1999 , apj , 511 , 852 kobayashi s. , ryde f. , macfadyen a. , apj submitted ( astro - ph/0110080 ) kulkarni s. r. , et al . , 1999 , nature , 398 , 389 lloyd - ronning n. m. , fryer c. l. , ramirez - ruiz e. , 2001 , apj , in press ( astro - ph/0108200 ) macfadyen a. i. , woosley s. e. , 1999 , apj , 524 , 262 macfadyen a. i. , woosley s. e. , heger a. , 2001 , apj , 550 , 410 meszaros p. , rees m. j. , 1997 , apj , 476 , 232 meszaros p. , rees m. j. , wijers r. a. m. j. , 1998 , apj , 499 , 301 norris j. p. , marani g. f. , bonnell j. t. , 2000 , apj , 534 , 248 panaitescu a. , kumar p. , 2001 , 560
, l49 piran t. , 1999 , physics rep . , 314 , 575 postnov k. a. , prokhorov m. e. , lipunov v. m. , 2001 , astronomy report , 45 , 236 ramirez - ruiz e. , lazzati d. , blain a. w. , 2002 , apj , 565 , l9 reichart d. e. , lamb d. q. , fenimore e. e. , ramirez - ruiz e. , cline t. l. , hurley k. , 2001 , apj , 552 , 57 rhoads j. e. , 1997 , apj , 487 , l1 salmonson j. d. , 2000 , apj , 544 , l115 salmonson j. d. , 2001 , apj , 546 , l29 sari r. , 1999 , apj , 524 , l43 sari r. , piran t. , halpern j. p. , 1999
, apj , 519 , l17 schmidt m. , 2001 , apj , 552 , 36 | gamma ray bursts are often modelled as jet - like outflows directed towards the observer ; the cone angle of the jet is then commonly inferred from the time at which there is a steepening in the power - law decay of the afterglow .
we consider an alternative model in which the jet has a beam pattern where the luminosity per unit solid angle ( and perhaps also the initial lorentz factor ) decreases smoothly away from the axis , rather than having a well - defined cone angle within which the flow is uniform .
we show that the break in the afterglow light curve then occurs at a time that depends on the viewing angle . instead of implying a range of intrinsically different jets
some very narrow , and others with similar power spread over a wider cone the data on afterglow breaks could be consistent with a standardized jet , viewed from different angles .
we discuss the implication of this model for the luminosity function .
gamma rays : bursts x rays : general x rays : ism |
in a recent paper ( @xcite ) we demonstrated that galaxy bimodality is not just evident in color ( see @xcite ; @xcite ) but also in the joint color - structure plane ( see also @xcite ; @xcite ; @xcite ; @xcite ) . in that work we used sdss photometry and single srsic ( 1963 ;
@xcite ) profile fits to investigate the distribution of 10,095 relatively nearby luminous ( @xmath6 mag ) galaxies in the color - srsic index plane .
the red peak is comprised of highly concentrated , high - srsic index systems while the blue peak contains more diffuse , low - srsic index systems .
this is important as any movement from the blue peak to the red peak will require modifying the orbits , angular momentum and energy of the entire stellar population .
a simple inert process ( e.g. , exhaustion of the gas supply , stripping , etc . ) could not achieve this , although a violent major merger event could ( e.g. , @xcite ) .
perhaps more importantly , when the population was segregated by hubble type we found that while the early - type galaxies ( e / s0s , i.e. , bulge dominated ) lay almost exclusively in the red - concentrated peak , and the late - type spirals ( sd / irr , i.e. , disc dominated ) in the blue - diffuse peak , the mid - type spirals ( sabc , i.e. , bulge plus disc systems ) straddled both peaks with no obvious sign of bimodality .
we inferred from this that galaxy bimodality arises because of the two component nature of galaxies and that spheroidal structures ( i.e. , ellipticals and bulges ) will lie exclusively in the red - compact peak and discs in the blue - diffuse peak .
as classical bulges lie within thin rotating disc systems this argues for early spheroid formation ( via rapid merging or collapse ) followed by a more quiescent phase in which the extended disc can form . to explore this further
we have performed bulge - disc decomposition of all 10,095 galaxies in the millennium galaxy catalogue ( see @xcite ) by fitting two - component srsic - bulge plus exponential - disc models using gim2d ( @xcite ) . in this letter
we report the luminosity functions derived for various component samples ( e.g. ellipticals , bulges , discs ) and tabulate the associated luminosity and stellar mass densities for each component class . throughout this paper
we assume a @xmath7cdm cosmology with @xmath8 , @xmath9 , and adopt @xmath10(100 km s@xmath11
mpc@xmath11 ) for ease of comparison with other results .
the millennium galaxy catalogue ( mgc ) is a deep ( @xmath12 mag arcsec@xmath13 ) , wide area ( @xmath14 deg@xmath15 ) , @xmath16-band imaging and redshift survey covering a @xmath17 deg wide strip along the equatorial sky from 10h to 14h 50@xmath18 .
the mgc contains 10,095 galaxies down to @xmath19 mag , of which 9,696 have redshift information .
full details of the imaging survey can be found in liske et al .
( 2003 ) , with the spectroscopic follow - up described by driver et al .
( 2005 ; hereafter d05 ) . in allen et al .
( 2006 ) the bulge - disc decompositions were reported for all 10,095 @xmath20 mag galaxies , and a final structural catalogue produced , consisting of bulge - only ( srsic profiles ) , disc - only ( srsic or exponential profiles ) and bulge plus disc systems ( srsic plus exponential profiles ) .
luminosity functions are computed using the standard step - wise maximum likelihood ( swml ) estimator originally described by efstathiou , ellis & peterson ( 1988 ) , with samples divided into bins of absolute magnitude .
the mgc spectroscopic sample has a nominal kron magnitude limit of @xmath19 mag .
however , here we will use the gim2d total magnitudes ( derived by integrating the light profiles to infinity ) so that there is no longer a single limit that applies to the sample as a whole . to accommodate for this
, each galaxy now has a unique magnitude limit , defined by : @xmath21 .
following d05 , we restrict our sample to galaxies in the redshift range @xmath22 and within carefully defined size and surface brightness boundaries ( see d05 and liske et al .
2006 for full details ) . in d05 individual k - corrections were derived for each galaxy by comparing the total galaxy broad - band colors ( @xmath23 ) to the 27 spectral templates given in poggianti ( 1998 ) and identifying the best fitting spectrum .
having now separated the mgc galaxies into bulges and discs ( see @xcite ) the global k - correction is not necessarily valid .
however , approximately @xmath24 per cent of our sample are best fit by one component profiles ( i.e. , bulge - only or disc - only galaxies ) and for these systems we adopt the k - corrections as previously derived in d05 . for the remaining , two - component systems we consider our component colors too coarse to be used to derive robust individual k - corrections ( as our decompositions are done in a single filter only ) . for the case of blue bulges and blue discs
the global k - correction is likely to be appropriate for both ( assumming that the blue bulge has formed from the disc ) . in the case of discs surrounding classical red bulges we note that for low-@xmath25 systems the k - correction is likely to be appropriate for the discs but not the bulges . in the case of high-@xmath25 systems
we note that peletier & bacells ( 1996 ) report that such discs are typically redder .
we therefore consider it appropriate to continue to adopt the global k - correction for both our discs and blue bulges .
for the red bulges , however , we adopt the spectral template most frequently adopted by our single component red ellipticals ( a sa 15 gyr spectrum ) .
this spectrum can be represented by a fourth order polynomial valid over the redshift range @xmath26 only : @xmath27 we note that if we follow a similar procedure for the discs and blue bulges the implied characteristic turnover luminosities ( shown in table 1 ) are systematically reduced by @xmath28 mag and the implied stellar masses reduced by @xmath29 per cent . to model evolution
we assume pure luminosity evolution of the following form : @xmath30 where @xmath31 is set to @xmath32 for the global luminosity function ( see * ? ? ?
* ) , @xmath33 for blue components ( discs , blue bulges , blue ellipticals ) , and @xmath34 for red components ( red bulges and red ellipticals ) .
these values are based on the recent results reported in zucca et al .
( 2006 ) for red and blue systems ( their type 1 and 4 respectively ) .
we do not model number evolution as the redshift range is small and our merger rate estimates , based on dynamically close pairs within the mgc , are low ( see @xcite and @xcite ) .
1 shows the luminosity distributions and schechter function fits for our full galaxy sample ( upper left ) , ellipticals only ( upper right ; i.e. objects with @xmath35 after logical filtering , see fig .
13 of allen et al . 2006 ) , discs ( lower left ) and bulges ( lower right ) .
note that the schechter function in the upper left differs from that shown in d05 because the magnitudes are now based on srsic profiles integrated to infinity rather than kron magnitudes .
this difference is significant , resulting in a brighter @xmath36 value by about @xmath37 mag but a comparable faint - end slope , @xmath38 .
analysis of independent repeat observations of @xmath39 galaxies suggests that our decompositions are valid to good accuracy ( @xmath40 mag and @xmath41 mag , see allen et al .
2006 ) for componentss with luminosities with @xmath6 mag .
below this limit our decompositions become increasingly less reliable and these data are shown with open symbols . the most striking result from fig . 1 is the rapidly rising faint - end slope for the elliptical population .
this was noted previously in driver et al .
( 2006 ) and was shown to be due to contamination of the classical elliptical sample by low luminosity blue spheroids ( see also @xcite ) . in fig .
2 we show the color - structure plane defined by sdss core ( @xmath42 ) psf color versus component srsic index for the ellipticals ( upper left ) and galaxy bulges ( upper right ) . the bimodality of the ellipticals is striking , with a blue and red population being apparent .
the blue sample defines what we label blue ellipticals which were identified in ellis et al .
( 2005 ) and quantified in driver et al .
( 2006 ) . a cut at @xmath43 mag provides a clear division .
the lower panels of fig .
2 show the luminosity distributions and schechter function fits for the blue and red samples .
we see that red ellipticals and blue ellipticals follow markedly different trends and it is indeed the blue ellipticals which are responsible for the apparent upturn in the total elliptical galaxy luminosity function at very faint absolute magnitudes .
we note that when the bulges are divided in the same manner , the blue bulge luminosity function is very similar to that of the ellipticals , possibly indicating some common origin .
it is tempting to associate the blue bulges with pseudo - bulges ( see kormendy & kennicutt 2004 ) , which are believed to arise from inner disc instabilities giving rise to a swelling of the disc in the central region . as many of our blue bulges have @xmath44 mag , where our bulge - disc decompositions become unreliable , we can not unambiguously confirm this population as pseudo - bulges but for the moment adopt this nomenclature for ease of discussion .
the blue ellipticals remain somewhat intriguing and appear to define a new class of object as previously noted by ellis et al .
( 2005 ) and driver et al .
( 2006 ; see also @xcite who identify them as a rapidly fading population ) .
we are currently exploring these systems further ( @xcite ) and for the moment simply flag them as interesting . from their distinct luminosity function
it is clear they are predominantly low luminosity systems and could potentially represent the local counterparts to the luminous blue compact galaxies studied by guzman et al .
( 1997 ) and phillips et al .
( 1997 ) .
3 shows the final component luminosity functions with the red ellipticals and red bulges combined into a single red spheroid group and the blue ellipticals and blue bulges grouped together into a single blue spheroid class .
the justification for this is the similarity in the shapes and ranges of the luminosity distributions from fig . 2 .
table 1 shows the schechter function values for the luminosity function fits shown in figs . 1 , 2 & 3 , along with their associated @xmath45-band luminosity , @xmath46 , and stellar mass densities , @xmath47 .
these are derived using the following expressions : @xmath48 and @xmath49 the latter expression is first shown in driver et al .
( 2006 ) and is based on the color to mass - to - light ratios given by bell & de jong ( 2001 ) which assume a salpeter-lite imf .
note that the @xmath50 colour for each galaxy was obtained by matching the mgc to the sloan digital sky survey first data release ( abazajian et al .
from table 1 we see that @xmath1 per cent of the stellar mass is in the form of galaxy discs , @xmath3 per cent in red elliptical galaxies and @xmath4 per cent in classical red bulges .
previously it has been reported ( bell et al . 2003 ; baldry et al .
2004 ; driver et al .
2006 ) that @xmath51 per cent of the stellar mass lies in the red peak dominated by early - type galaxies .
as we have now separated the early - type galaxies into bulges and discs one expects that the spheroid ( elliptical+bulge ) stellar mass density should be lower than the early - type stellar mass density , which of course includes the associated disc components of the lenticular galaxies . examining the color - structure plane ( fig .
2 upper panels ) for ellipticals and bulges we see that the red populations ( of each type ) lie in the same location .
this is indicative of a shared origin for red ellipticals and classical red bulges .
when combined the red spheroids account for @xmath2 per cent of the stellar mass density .
hence the bulk of the stellar mass ( @xmath52 per cent ) exists in the two classical structures originally defined by de vaucouleurs ( 1959 ) .
the remaining @xmath53 per cent lie in the form of blue ellipticals ( @xmath54 per cent ) and blue bulges ( @xmath5 per cent ) .
these latter two populations can therefore be considered minor from a cosmological perspective .
furthermore , as the blue bulges are likely to represent either difficulties in the decomposition ( e.g. , bars ) or pseudo - bulges ( disc swelling ) , their stellar mass could arguably be added , in either case , to that of the galaxy discs ( i.e. , @xmath55 per cent ) .
the nature of the blue ellipticals ( previously dubbed blue spheroids ) remains uncertain but they appear to constitute a very small fraction of the stellar mass budget , although we must note the near divergent faint - end slopes .
the primary conclusion then is that the stellar mass is mainly divided between two distinct structures : blue 2d discs and red 3d spheroids .
finally we note that the results presented in this letter are based on @xmath16-band data and therefore susceptible to dust attenuation ( see @xcite ) .
bell & de jong ( 2001 ) argue that the effect on an individual galaxy s stellar mass is less than one might expect ( see their fig .
1 ) because the observed decrease in total luminosity is offset by the increased stellar mass - to - light ratio inferred from the redenned colors , therefore yielding comparable final stellar masses . we explore this in detail in driver et al .
( 2007 ) and note that while the masses are modified somewhat the final stellar mass breakdown is not dramatically altered .
the millennium galaxy catalogue consists of imaging data from the isaac newton telescope and spectroscopic data from the anglo australian telescope , the anu 2.3 m , the eso new technology telescope , the telescopio nazionale galileo , and the gemini telescope .
the survey has been supported through grants from the particle physics and astronomy research council ( uk ) and the australian research council ( aus ) .
the data and data products are publicly available from http://www.eso.org/@xmath56jliske/mgc/ or on request from j. liske or s.p . driver .
abazajian k. , et al .
, 2003 , aj , 126 , 2081 allen p. , driver s.p . , graham a.w . , cameron e. , liske j. , cross n.j.g . , de propris r. , 2006 , mnras , 371 , 2 baldry i. , glazebrook k. , brinkmann j. , ivezic z. , lupton r.h . ,
nichol r.c . ,
szalay a.s .
, 2004 , apj , 600 , 681 ball n. , loveday j. , brunner r.j . , 2006 , mnras , submitted , ( astro - ph/0610171 ) barnes j.e . , hernquist l. , 1992 , nature , 360 , 715 bell e. , mcintosh d.h .
, katz n. , weinberg m.d .
, 2003 , apjs , 149 , 289 bell e. , de jong r.s . , 2001 ,
apj , 550 , 212 choi y. , park c. , vogeley m. , 2006 , apj , in press , astro - ph/0611607 conselice c.j . , 2006 , mnras , submitted , astro - ph/0610016 de propris r. , liske j. , driver s.p . , allen p.d . ,
cross n.j.g .
, 2005 , aj , 130 , 1516 de propris r. , liske j. , conselice c. , driver s.p .
, graham a.w .
, allen p.d .
, 2007 , mnras , submitted de vaucouleurs g. , 1959 , hdp , 53 , 275 driver s.p . , liske j. , cross n.j.g . , de propris r. , allen p.d . , 2005 , mnras , 360 , 81 driver s.p . , et al
, 2006 , mnras , 368 , 414 driver s.p . , popescu c. , tuffs r.j . , liske j. , graham a.w . ,
allen p.d . , de propris r. , 2007 , mnras , submitted efstathiou g. , ellis r.s .
, peterson b.a . , 1988 ,
mnras , 232 , 431 ellis s.c . , driver s.p . , allen p.d . , liske j. , bland - hawthorn j. , de propris r. , 2005 , mnras , 363 , 1257 ellis s.c . , driver s.p . , liske j. , graham a.w . , 2007 , mnras , in preparation graham a.w . , driver s.p . , 2005 , pasa , 22 , 118 guzman r. , jangren a. , koo d.c .
, bershady m.c .
, simard l. , 1998 , apj , 495 , 13 ilbert o. , et al . , 2006 , a&a , 453 , 809 kormendy j. , kennicutt r.c.jr . , 2004 , ara&a , 42 , 603 liske j. , lemon d. , driver s.p . , cross n.j.g . , couch w.j . , 2003 , mnras , 344 , 307 liske j. , driver s.p . , allen p.d .
, cross n.j.g . , de propris r. , 2006 , mnras , 369 , 1547 park c. , choi y. , vogeley m. , gott ii j.r . , blanton m.r . , 2006 ,
apj , in press , astro - ph/0611610 peletier r.f .
, bacells m. , aj , 111 , 2238 phillips a.c . , et al .
, 1997 , apj , 489 , 543 poggianti b. , 1997 , a&as , 122 , 399 sersic j .-
, 1963 , baaa , 6 , 41 shao z. , xiao q. , shen s. , mo h.j . , xia x. , deng z. , 2006 , apj , submitted , astro - ph/0611714 simard l. , et al .
, 2002 , apjs , 142 , 1 strateva i. , et al . , 2001 ,
aj , 122 , 1861 zucca e. , et al . , 2006 , a&a , 455 , 879 lcccccc all & @xmath57 & @xmath58 & @xmath59 & @xmath60 & @xmath61 & @xmath62 + all discs & @xmath63 & @xmath64 & @xmath65 & @xmath66 & @xmath67 & @xmath68 + bulges & @xmath69 & @xmath70 & @xmath71 & @xmath72 & @xmath73 & @xmath74 + ellipticals & @xmath75 & @xmath76 & @xmath77 & @xmath78 & @xmath79 & @xmath80 + blue ellipticals & @xmath81 & @xmath82 & @xmath83 & @xmath84 & @xmath85 & @xmath86 + red ellipticals & @xmath87 & @xmath88 & @xmath89 & @xmath90 & @xmath91 & @xmath92 + blue bulges & @xmath93 & @xmath94 & @xmath83 & @xmath95 & @xmath85 & @xmath96 + red bulges & @xmath97 & @xmath98 & @xmath99 & @xmath100 & @xmath101 & @xmath102 + ellipticals + bulges & @xmath103 & @xmath104 & @xmath105 & @xmath106 & @xmath107 & @xmath108 + red ( ellipticals + bulges ) & @xmath109 & @xmath110 & @xmath111 & @xmath112 & @xmath113 & @xmath114 + blue ( ellipticals + bulges ) & @xmath115 & @xmath116 & @xmath117 & @xmath118 & @xmath119 & @xmath120 + | we derive the luminosity functions of elliptical galaxies , galaxy bulges , galaxy pseudo - bulges and galaxy discs from our structural catalogue of 10,095 galaxies .
in addition we compute their associated luminosity and stellar mass densities .
we show that spheroidal systems ( elliptical galaxies and the bulges of disc galaxies ) exhibit a strong color bimodality indicating two distinct types of spheroid which are separated by a core color of @xmath0 mag .
we argue that the similarity of the red elliptical and the red bulge luminosity functions supports our previous arguments that they share a common origin and surprisingly find that the same follows for the blue ellipticals and blue bulges , the latter of which we refer to as pseudo - bulges . in terms of the stellar mass budget
we find that @xmath1 per cent is currently in the form of discs , @xmath2 per cent in the form of red spheroids ( @xmath3 per cent ellipticals , @xmath4 per cent bulges ) and the remainder is in the form of blue spheroidal systems ( @xmath5 per cent blue ellipticals and @xmath5 per cent pseudo - bulges ) . in terms of galaxy formation
we argue that our data on galaxy components strongly supports the notion of a two - stage formation process ( spheroid first , disc later ) but with the additional complexity of secular evolution occurring in quiescent discs giving rise to two distinct bulge types : genuine classical bulges and pseudo - bulges .
we therefore advocate that there are three significant structures underpinning galaxy evolution : classical spheroids ( old ) ; pseudo - bulges ( young ) and discs ( intermediate ) .
the luminous nearby galaxy population is a mixture of these three structural types .
the nature of the blue elliptical galaxies remains unclear but one possibility is that these constitute recently collapsed structures supporting the notion of mass - dependent spheroid formation with redshift . |
this paper deals with radial solutions of the system of equations @xmath0 where @xmath1 and @xmath2 is an ignition type nonlinearity , that is , there exists @xmath3 such that @xmath4 system arises as a model problem for some reaction diffusion systems in combustion theory
. we will describe the connection with these models at the end of the section .
we call @xmath5 the trivial solution of . note that for any solution @xmath6 of one has @xmath7 as can be deduced from the maximum principle .
we are mainly interested in existence and multiplicity of nontrivial radial solutions of .
the first observation in this direction is that for @xmath8 large has no nontrivial solutions .
indeed , by the maximum principle @xmath9 and then @xmath10 if @xmath8 is large then @xmath11 for all @xmath12 $ ] and we conclude that @xmath13 .
better estimates for the quantity @xmath14 are given in section [ estimate on eps * ] .
suppose that @xmath15 is a nontrivial radial solution .
then for some @xmath16 we must have @xmath17 .
otherwise @xmath18 and then @xmath19 , and by the liouville theorem @xmath20 .
since @xmath21 this would imply @xmath22 and then @xmath15 is trivial .
so for a nontrivial solution @xmath15 , since @xmath23 , there is @xmath24 such that @xmath25 the parameter @xmath26 , which we will write simply as @xmath27 , then serves to distinguish different solutions .
[ thm existence1 ] assume @xmath28 \to [ 0,+\infty)$ ] is continuous , satisfies and for some @xmath29 @xmath30 .\end{aligned}\ ] ] then @xmath31 and there exist @xmath32 such that for @xmath33 there are at least 2 solutions of .
one of them has bounded @xmath27 as @xmath34 and the other has @xmath27 in the range @xmath35 as @xmath34 , where @xmath36 is fixed .
a very natural and interesting question is the stability of the solutions constructed in theorem [ thm existence1 ] .
based on the works @xcite we conjecture that the solution with bounded @xmath27 is unstable and that in part of branch of solutions with large @xmath27 the solution is stable , at least with respect to radial perturbations . in some cases one may want to consider a discontinuous nonlinearity , such as the heavisde function @xmath37}$ ] . with this example in mind
we introduce the following hypothesis @xmath38 $ and $ \lim_{u\to \theta+ } f(u)$ exists } .\end{aligned}\ ] ] [ thm existence2 ] assume @xmath28 \to [ 0,+\infty)$ ] satisfies , , and . then @xmath31 and there exist @xmath32 such that for @xmath33 there are at least 2 solutions of .
one of them has bounded @xmath27 as @xmath34 and the other has @xmath27 in the range @xmath35 as @xmath34 , where @xmath36 is fixed .
the solution constructed in theorem [ thm existence2 ] is such that the set @xmath39 is finite , and hence the equation holds a.e . in @xmath40 . the motivation to consider a discontinuous nonlinearity is only mathematical .
however the example @xmath41}$ ] is interesting since it provides a situation where explicit calculations are possible .
theorem [ thm existence2 ] shows that in part the conclusions obtained for @xmath41}$ ] remain valid for more general non - linearities . for @xmath41}$ ] explicit calculations lead to an equation for @xmath27 and @xmath42 in order for a radial solution to exist . in figure [ fig1 ]
we show the numerical solution for this relation when @xmath43 , with @xmath27 in the vertical axis and @xmath44 in the horizontal axis .
it shows that for @xmath45 there are 2 solutions .
solutions in the lower branch satisfy @xmath46 in @xmath47 that is , the reaction takes place in the ball of radius @xmath27 .
the same happens for points in the upper branch which are to the right of the special point marked in the graph . to the left of that point the solution satisfies @xmath46 in an annulus of the form @xmath48 .
thanks to the explicit form of the relation bewteen @xmath27 and @xmath42 we can compute the asymptotic behavior of the curve as @xmath49 , and we find that @xmath50 where @xmath51 is the unique solution of the system of equations @xmath52 [ fig1 ] because of the information on the heaviside nonlinearity one can conjecture that for general @xmath2 there should be a similar relation for @xmath27 and @xmath42 as @xmath53 .
we present in section [ apriori estimates ] nonexistence results for general ignition nonlinearities satisfying and , that capture this relation , and roughly speaking say that no solution can exist if @xmath54 is either too large or too small , provided @xmath27 is taken large enough .
using these nonexistence results and degree theory we can give a proof of theorem [ thm existence1 ] .
this is done in section [ existence of solutions ] . in section [
sect exsit discont ] we give the proof of theorem [ thm existence2 ] by approximating the discontinuous nonlinearity by continuous ones .
section [ the heaviside ignition ] is devoted to the explicit computations for the heaviside function .
finally section [ estimate on eps * ] contains a finer estimate of @xmath55 . as mentioned before , system arises in connection with some models in combustion theory , more precisely , in the flame ball problem for a weakly premixed gas sensitive to radiative heat losses . in such a mixture , it is known that apparently stationary spherical structures appear , which are called flame balls @xcite . in @xcite
the following reaction diffusion system has been used to model a combustion process where flame balls can appear : @xmath56 where @xmath57 is the temperature , @xmath58 the reactant concentration , @xmath59 , @xmath60 are the reactant concentration and tempereature at infinity , and @xmath61 , @xmath62 , @xmath63 and @xmath64 are respectively the specific heat capacity at constant pressure , the perfect gas constant , the chemical heat release and the molecular mass of the reactant .
the term @xmath65 represents radiative losses .
the reaction is characterized by the one - step arrhenius kinetics @xmath66 where @xmath67 is a constant .
furthermore , the hydrodynamics effects are neglected , i.e the density @xmath68 , the thermal conductivity @xmath69 and the diffusion coefficient @xmath70 are constant .
see also numerical simulations in @xcite . after the seminal work @xcite , the traveling front problem for systems like has been investigated by several authors , for example , @xcite . in absence of radiation ,
i.e. when @xmath71 , there are many works dealing with , see for instance @xcite and the references therein .
also we remark that the stationary version of without radiation , leads to system with @xmath72 which reduces to a scalar equation , since @xmath73 .
there is a huge amount of literature concerning existence of radial ground states for semilinear equations , so we mention here only some classical references @xcite on the problem in entire space .
when the problem is treated in a bounded domain see the book @xcite and @xcite for multiplicity results in the case of arrehnius non - linearity .
the paper @xcite contains interesting numerical computations of the bifurcation diagram in the case of the full coupled system in a interior of a sphere
. a common simplification of under the assumption of large activation energy , that is @xmath74 , is to assume that the source term for the reaction is concentrated on a very thin layer , typically a sphere .
this approach is taken for example in @xcite and leads to the free boundary problem @xmath75 where @xmath76 is the radius of the front where the reaction takes place , @xmath77 is the dirac measure and @xmath78 is the front temperature . in @xcite
the authors analyze the stability of stationary solutions of . in a similar framework , existence and stability of flame balls and travelling flame balls
have also been studied in @xcite .
we arrive at by introducing the following simplifications : * assume the radiative loss to be linear , i.e. , @xmath79 where @xmath80 , * approximate @xmath81 by @xmath82 where @xmath83 is an activation temperature and @xmath84 is a cut - off function satisfying @xmath85 in @xmath86 and @xmath87 in @xmath88 . as in @xcite one can model radiative heat losses using stefan s law @xmath89 for some constant @xmath8 .
when @xmath57 is close to @xmath90 we can write @xmath91 . as a step towars understanding more general situations , we assume that this linear relation holds for all @xmath57 , that is , we assume a ) .
other linear or piecewise linear approximations have been used before , for instance in @xcite .
assumption b ) corresponds the a standard approximation in combustion theory to avoid the _ cold boundary difficulty _ , see @xcite . after introducing dimensionless variables @xmath92 , @xmath93 , corresponding to temperature and reactant concentration ,
the stationary version of becomes @xmath94 where @xmath95 is the lewis number , @xmath96 , @xmath97 and @xmath98 is an ignition type function , that is , there is @xmath99 such that @xmath100 if @xmath101 .
we stress that our results are valid for any value of @xmath95 . indeed ,
since we are considering stationary solutions , the following change of variables will allow us to assume that @xmath102 .
letting @xmath103 , @xmath104 transforms system into , where @xmath105 and @xmath106 .
we observe that @xmath2 is still is an ignition type nonlinearity .
the purpose in this section is to establish nonexistence results in some ranges of the parameters . given a nontrivial solution @xmath15 of let @xmath107 be such that @xmath108 setting @xmath109 , @xmath110 these new functions satisfy @xmath111 in the sequel we will study in the following set of functions @xmath112 let @xmath113 be such that @xmath114 we consider now functions @xmath115 such that @xmath116 [ nonexistence large epsilon beta2 ] there is no solution in @xmath117 to if @xmath118 since @xmath119 we have an explicit formula @xmath120 > from this we find @xmath121 similarly , since @xmath122 for @xmath123 and @xmath124 we have @xmath125 where @xmath126 .
this yields @xmath127 integrating the equation for @xmath93 in ( 0,1 ) implies @xmath128 let @xmath129 be such that @xmath130 . then integrating the equation for @xmath92 in @xmath131 yields @xmath132 this formula together with and gives @xmath133 and it follows that @xmath134 [ nonexistence small epsilon beta2 ] there is @xmath135 , @xmath36 depending only on @xmath136 , @xmath137 , @xmath138 such that for all @xmath139 , @xmath140 and all @xmath2 satisfying , , there is no solution in @xmath117 to the system .
we treat the case @xmath8 since the situation @xmath72 is similar .
as before @xmath120 > from this we find @xmath141 similarly @xmath142 where @xmath126 .
this yields @xmath127 integrating the equation for @xmath92 we see that @xmath143 > from this and it follows that @xmath144 integrating the equation for @xmath93 in ( 0,1 ) implies @xmath145 and combined with yields @xmath146 integrating the equation for @xmath93 in @xmath147 we obtain @xmath148 integrating this on @xmath149 with @xmath150 yields @xmath151 in particular , with @xmath152 @xmath153 this and give @xmath154 now going back to we obtain @xmath155 now consider the function @xmath156 which satisfies @xmath157 in @xmath40 . integrating the relation @xmath158 in @xmath147 @xmath159 so that @xmath160 it follows that @xmath161 observe that integrating on @xmath162 we find @xmath163 and since @xmath164 @xmath165 in particular , using , @xmath166 * step 1 . * for any @xmath167 there exists @xmath168 depending on @xmath169 , @xmath136 and @xmath137 only such that @xmath170 } u < \frac{1+\theta}{2 } \quad \forall \beta \ge \beta_1 .\end{aligned}\ ] ] to prove this , suppose that @xmath171 } u \ge \frac{1+\theta}{2 } $ ] and let @xmath172 $ ] be such that @xmath173 .
let @xmath174 so that from @xmath175 then @xmath176 \quad \hbox{such that } \quad |s - r| \le \frac{1-\theta}{4 m}.\end{aligned}\ ] ] using and since @xmath177 @xmath178 if @xmath179 let @xmath180 $ ] and if @xmath181 let @xmath182 $ ] . using and : @xmath183 we take @xmath36 small only depending on @xmath137 such that @xmath184 then @xmath185 and the claim follows . * step 2 .
* for any @xmath167 there exists @xmath29 such that @xmath186 } u -\theta ) h_0\left(\frac{\max_{[r_0,1 ] } u-\theta}{2 } + \theta \right ) \le \frac{c}{\beta^2}.\ ] ] the constant @xmath187 depends only on @xmath169 and @xmath137 .
the conclusion from this is that @xmath188 } ( u - \theta ) \le 0,\end{aligned}\ ] ] and this is uniform with respect to @xmath92 and @xmath2 . as before , setting @xmath174 we have @xmath189 for @xmath190 by .
let @xmath191 $ ] be a point such that @xmath192 } u$ ] and let us write @xmath193 } u$ ] .
then @xmath194 \quad \hbox{such that } \quad |r - r_m| \le \frac{u_m - \theta}{2m}.\ ] ] since @xmath195 , using , , , we obtain @xmath196 thus there exists @xmath187 depending only on @xmath169 , @xmath137 such that for @xmath197 @xmath198 this proves the claim .
* step 3 . * for @xmath199 @xmath200 } ( u - \theta).\end{aligned}\ ] ] indeed multiplying the equation @xmath201 by @xmath202 and integrating over @xmath203 we get @xmath204 by @xmath205 since @xmath206 we have @xmath207 finally using @xmath208 } ( u - \theta ) \beta^2
\int_a^1 r^2 v f(u ) \ , d r \le \max_{[a,1 ] }
( u - \theta ) .\ ] ] * step 4 . * we finish the proof of the lemma using a modification of pohozaev s identity .
let @xmath209 be a fixed number .
multiplying by @xmath210 and integrating over @xmath203 @xmath211 a computation shows that the left hand side is given by @xmath212 thanks to we find @xmath213 } ( u - \theta).\ ] ] since @xmath214 we obtain from , and the previous estimate @xmath215 } ( u - \theta).\end{aligned}\ ] ] let us compute @xmath216 recalling that @xmath217 this shows that @xmath218 define @xmath219 then @xmath220 these terms can be estimated using : @xmath221 and hence by @xmath222 } ( u-\theta)\ ] ] similarly @xmath223 } ( u-\theta ) .\ ] ] to estimate the remaining terms we use @xmath224 where the last inequality is consequence of , and provided @xmath225 ( @xmath226 is fixed ) .
hence @xmath227 } ( u-\theta),\ ] ] and this implies also @xmath228 } ( u-\theta).\ ] ] therefore @xmath229 } ( u-\theta ) \end{aligned}\ ] ] where @xmath230 .
> from , , , we arrive at @xmath231 } ( u-\theta ) \quad \forall \beta \ge \beta_1(a ) .\end{aligned}\ ] ] we fix @xmath36 even smaller if necessary so that @xmath232 then fix @xmath233 such that @xmath234 then for yields @xmath235 } ( u-\theta ) \quad \forall \beta \ge \beta_1(a)\ ] ] which is not possible for @xmath27 large enough by .
in this section we prove theorem [ thm existence1 ] , and throughout it we assume that @xmath2 is continuous and satisfies and .
we work in the banach space @xmath236 ) , \ ; u(1 ) = 0 , \ ; \beta\in \r \}\ ] ] endowed with its natural norm @xmath237 ) } + |\beta| , \ ] ] where @xmath238 is fixed . define for @xmath239 $ ] @xmath240,\ ] ] where @xmath241 .
let us consider with nonlinearity @xmath242 that is @xmath243 to apply the non - existence results of the previous section we need to exhibit a function @xmath136 satisfying and . for this purpose define @xmath244\ : \
f \ge t \hbox { in } [ r,1 ] \ } \quad \hbox{for all } t \ge 0,\ ] ] and @xmath245.\ ] ] the following properties then follow from these definitions : 1 .
@xmath98 is strictly increasing , continuous from the left and satisfies @xmath246 .
2 . @xmath247 for @xmath248 $ ] , @xmath249 for @xmath250 $ ] , @xmath251 is nondecreasing and continuous in @xmath252 $ ] .
3 . @xmath253 in @xmath252 $ ] .
the function @xmath254 satisfies and and the nonlinearity @xmath242 satisfies , , for all @xmath255 $ ] .
since any solution to is bounded above by 1 we may redefine @xmath256 as a constant for @xmath257 .
thus we may assume that for some constant @xmath258 we have @xmath259.\ ] ] we define a nonlinear map @xmath260 as follows .
let @xmath261 . then solve @xmath262 this problem has a unique solution which can be found for example by minimizing @xmath263 with @xmath264 . since @xmath265 by standard elliptic regularity @xmath266)$ ] . then find @xmath107 and @xmath92 such that @xmath267 admits a solution .
this problem has a unique solution @xmath268 which furthermore is @xmath269)$ ] ( this assertion is verified in the proof of lemma [ t continuous compact ] below ) .
we define @xmath270 observe that @xmath268 is a fixed point of @xmath271 if and only if @xmath272 is a solution of .
[ t continuous compact ] write @xmath273 .
then @xmath274 \to x\ ] ] is compact .
moreover @xmath275.\end{aligned}\ ] ] first we remark that the solution @xmath93 to satisfies @xmath276 because @xmath277 is a subsolution and @xmath278 is a supersolution .
given @xmath279 let @xmath280 be the solution to @xmath281 note that @xmath280 is explicit : @xmath282 and hence @xmath283 let @xmath284 .
then choosing @xmath285 , multiplying by @xmath280 and integrating in @xmath286 we find @xmath287 > from the boundary condition in it follows that @xmath27 is explicitly given by @xmath288 where @xmath285 . by @xmath289
> from formulas , we see that @xmath290 using this inequality , standard elliptic estimates and the facts that @xmath291 and @xmath242 is uniformly bounded we obtain . from here
we deduce that @xmath57 is continuous and compact
. indeed , for the latter assertion , note that if @xmath292 be a bounded set of @xmath293 $ ] then @xmath294 is a bounded set in @xmath295 ) \times \r$ ] for any @xmath296 and taking @xmath297 it follows that this set is precompact in @xmath269 ) \times \r$ ] . [ lemma unique eps=0 t=0 ] the operator @xmath298 has a unique fixed point @xmath299 in @xmath300 .
in this situation @xmath301 and hence the system reduces to @xmath302 with the additional requirement that @xmath303 let @xmath304 .
then the equation for @xmath305 becomes @xmath306 this equation is of logistic type and many properties are well known ( see @xcite ) : 1 .
any solution @xmath305 to satisfies @xmath307 , and either @xmath308 or @xmath309 in @xmath286 .
2 . has a nontrivial solution if and only if @xmath310 where @xmath311 is such that @xmath312 and @xmath313 denotes the first dirichlet eigenvalue of @xmath314 in the unit ball .
3 . for @xmath310
there is a unique non - trivial solution , which we write as @xmath315 .
4 . @xmath315 is monotone increasing with respect to @xmath27 and @xmath316 uniformly on compact sets of @xmath286 and @xmath317 > from the above properties it follows that @xmath318 moreover @xmath319 it follows that there is a unique @xmath310 such that @xmath320 we call this value @xmath321 and the let @xmath322 .
then @xmath323 is the fixed point of @xmath298 .
[ sol is nondegenerate ] let @xmath299 denote the unique fixed point of @xmath298 found in lemma [ lemma unique eps=0 t=0 ] .
then @xmath299 is nondegenerate .
let us write @xmath324 the solution of @xmath325 of lemma [ lemma unique eps=0 t=0 ] .
we have to verify that the linearization of @xmath326 around this solution is an invertible operator .
note that the operator @xmath298 involves only the nonlinearity @xmath327 and that the fixed point @xmath299 from lemma [ lemma unique eps=0 t=0 ] satisfies @xmath328 in @xmath329 and @xmath330 .
therefore for @xmath331 in a neighborhood of @xmath332 in the topology of @xmath269)$ ] we have @xmath333 .
hence @xmath298 is differentiable at @xmath299 .
we next compute the derivative of @xmath298 at @xmath299 in the direction of @xmath334 , which we write as @xmath335 then @xmath336 where @xmath337 and @xmath338 are computed as follows . to compute @xmath339 we have linearize with respect to @xmath340 and then set @xmath341 , @xmath342 . since for @xmath343
, we have @xmath344 and @xmath345 , the function @xmath339 satisfies @xmath346 where @xmath347 is adjusted so that @xmath339 satisfies both boundary conditions , @xmath348 is found solving @xmath349 and @xmath350 is the solution of with @xmath351 , @xmath352 and @xmath343 . as explained in the proof of lemma [ lemma unique eps=0 t=0 ] then @xmath353 to compute @xmath354 we linearize with respect to @xmath355 and then set @xmath341 , @xmath342
. therefore @xmath354 satisfies @xmath356 where @xmath357 is chosen so that both boundary conditions are satisfied and @xmath358 is computed from the equation @xmath359 we need to find the kernel of @xmath360 , that is solutions
@xmath361 of : @xmath362 using the notation above we may write @xmath363 , @xmath364 .
we claim that @xmath365 to prove this , let @xmath366 .
then a calculation shows that @xmath367 and @xmath368 .
this implies @xmath369 and follows .
> from the formulas we get @xmath370 but also @xmath331 satisfies @xmath371 the linear elliptic operator @xmath372 with dirichlet boundary condition arises as the linearization of around the solution @xmath332 , and it is well known that it has a positive first eigenvalue . since @xmath373 , for positive @xmath374 we would have @xmath375 and @xmath376 and for negative @xmath374 we would have @xmath377 and @xmath378 , which are both impossible since @xmath379 .
hence @xmath380 and also @xmath381 .
this shows that the linear equation @xmath382 admits only the trivial solution .
since @xmath383 is compact we see that @xmath384 is an isomorphism . by lemma
[ nonexistence small epsilon beta2 ] there exists @xmath36 and @xmath385 such that for all @xmath197 , @xmath140 and all @xmath239 $ ] there is no solution in @xmath117 to the system .
this means that @xmath271 has no fixed point @xmath268 if @xmath197 and @xmath140 .
let @xmath386 and define @xmath387 ) } < r_1 , \ ; 0 < \beta < \beta_1 \ } .\end{aligned}\ ] ] let @xmath388
. then @xmath389 is a bounded open set of @xmath300 and for @xmath390 and @xmath391 the operator @xmath271 has no fixed point in @xmath392 . indeed , suppose @xmath393 is a fixed point of @xmath271 .
it is not possible that @xmath394 by lemma [ nonexistence small epsilon beta2 ] .
the case @xmath395 is also impossible .
this means that @xmath396 and @xmath397 ) } = r_1 $ ] .
but by inequality we would have @xmath398 ) } \le c ( 1+\sinh(\sqrt{\varepsilon}\beta ) + \beta^2 ) < c ( 1+\sinh(\delta ) + \beta_1 ^ 2 ) = r_1\ ] ] which is impossible .
thus the leray - schauder degree @xmath399 is well defined for @xmath400 , @xmath239 $ ] . by lemmas [ lemma unique eps=0 t=0 ] and [ sol is nondegenerate ] we have @xmath401 and by homotopy invariance @xmath402.\ ] ] this shows @xmath271 has at least one fixed point in @xmath389 for @xmath403 and @xmath255 $ ] .
in particular the system has a solution with bounded @xmath27 for any @xmath404 .
we know there exists @xmath31 such that for @xmath405 the system has no solution , and hence @xmath406 has no fixed points in @xmath300 for such @xmath42 .
let @xmath407 and define @xmath408 ) } < r_2 , \ ; 0 < \beta < \beta_2 \}\end{aligned}\ ] ] where @xmath409 then @xmath271 has no fixed points on @xmath410 for @xmath411 and @xmath239 $ ] . then using a homotopy along @xmath412 $ ] we find @xmath413 this implies that for @xmath414 the system has at least another solution and that this solution has @xmath27 in the range @xmath415 .
since @xmath416 is arbitrary we obtain the same conclusion for @xmath417 .
in this section we prove theorem [ thm existence2 ] .
we assume that @xmath2 satisfies , , and that @xmath418 for @xmath419 define @xmath420 then @xmath421 is continuous and satisfies and with a fixed constant .
moreover there is @xmath136 satisfying and and such that @xmath422 for all @xmath423 . such @xmath136 can be taken for instance as @xmath424 for some @xmath425 small enough . by theorem [ thm existence1 ]
there is @xmath426 such that the system admits 2 solutions @xmath427 , @xmath428 for all @xmath429 .
moreover there is a fixed number @xmath430 such that @xmath431 and @xmath432 for all @xmath419 and @xmath429 . from now on we fix @xmath429 and study the limit as @xmath433 of any of the 2 solutions which we call just @xmath434 with parameter @xmath435 .
since @xmath435 is bounded and by standard elliptic estimates we may assume that @xmath436 , @xmath437 and @xmath438 in @xmath269)$ ] .
since @xmath439 is uniformly bounded we may extract a further subsequence such that @xmath440 weakly- * in @xmath441 .
the function @xmath331 then satisfies @xmath442 } f$ ] in @xmath252 $ ] .
since @xmath443 satisfies a linear equation for @xmath444 and vanishes at infinity it has an explicit formula @xmath445 setting @xmath446 for @xmath444 we see that @xmath15 satisfy @xmath447 the set @xmath448 is open ( relative to @xmath449 and for any @xmath450 we have @xmath451 as @xmath452 .
it follows that @xmath453 in @xmath454 .
outside @xmath454 we have @xmath455 a.e .
indeed , let @xmath456 , @xmath457 .
then @xmath458 } f_n(u_n ) \psi + \int_{d^c \cap [ 0,1 ] } f_n(u_n ) \psi.\ ] ] since @xmath459 in @xmath454 by dominated convergence we have @xmath460 } f_n(u_n ) \psi \to \int_{d \cap [ 0,1 ] } f(u ) \psi .\ ] ] on the other hand , if @xmath461 @xmath462 so that @xmath463 on @xmath464 . then @xmath465 } ( f_n(u_n ) -\eta ) \psi = \int_{d^c \cap[0,1 ] } ( f_n(u_n ) -\eta)^+ \psi - \int_{d^c \cap[0,1 ] } ( f_n(u_n ) -\eta)^- \psi
\le o(1)\ ] ] where @xmath466 denotes a sequence converging to 0 as @xmath467 .
it follows that @xmath468 } f_n(u_n ) \psi \le \int_{d^c \cap [ 0,1 ] } \eta \psi + o(1)\ ] ] and hence @xmath469 } f(u ) \psi + \int_{d^c \cap [ 0,1 ] } \eta \psi.\ ] ] this shows that @xmath455 a.e . in @xmath464 .
our main objective is to show that the complement of @xmath454 is finite .
if @xmath470 whenever @xmath471 then @xmath464 is discrete and since it is contained in @xmath252 $ ] it is finite . let us analyze the case where for some @xmath472 $ ] we have @xmath471 and @xmath473 .
let @xmath474 then @xmath475 and @xmath476 , @xmath477 .
we assert that there is a small interval @xmath478 , @xmath425 such that @xmath46 in @xmath478 . to prove this we start ruling out the possibility that @xmath479 for some infinite sequence @xmath480 .
we actually may assume that if @xmath423 is even then @xmath46 on @xmath481 and if @xmath423 is odd @xmath482 on @xmath481 .
let us see that the following holds @xmath483 where @xmath484 and @xmath485 denotes a sequence bounded by @xmath486 with @xmath187 independent of @xmath423 as @xmath433 .
suppose first that @xmath482 on @xmath481 and define @xmath487 by @xmath488.\ ] ] then @xmath489.\ ] ] @xmath490 this equation implies that @xmath487 and @xmath491 are uniformly bounded on @xmath252 $ ] .
therefore @xmath492\ ] ] hence integrating @xmath493\ ] ] and @xmath494 .\end{aligned}\ ] ] this proves in this case .
now suppose that @xmath46 on @xmath481 . defining @xmath487 as before we now obtain the following equation @xmath495.\ ] ] @xmath496 where @xmath497.\ ] ] this equation implies that @xmath487 , @xmath491 are uniformly bounded on @xmath252 $ ] .
thus @xmath498.\ ] ] multiplying this equation by @xmath487 and integrating in @xmath252 $ ] yields @xmath499 which shows that @xmath500 and proves in this case . using inductively we find @xmath501 letting @xmath502 , and using that @xmath130 yields @xmath503 choose a subsequence @xmath504 as @xmath505 such that @xmath506 for all @xmath507 .
then @xmath508 but implies that @xmath509 .
it follows that @xmath510 which is impossible as @xmath505 .
this establishes the assertion that @xmath92 can not oscillate infinitely many times around @xmath137 to the right of @xmath169 .
it follows that for some @xmath425 either @xmath46 in @xmath478 or @xmath482 in @xmath478 .
the latter can in fact not occur by the hopf lemma .
since @xmath92 satisfies the ode @xmath511 in @xmath478 and in this interval @xmath46 we see that @xmath92 is @xmath512 in @xmath513 .
since @xmath169 is a minimum of @xmath92 restricted to @xmath513 it follows that @xmath514 . which yields the following inequality @xmath515 now , if @xmath93 is constant in @xmath516 $ ] this means that @xmath517 a.e . in @xmath516
$ ] and @xmath101 in @xmath516 $ ] .
actually @xmath482 in @xmath518 by the strong maximum principle and this finishes the proof in this case . if @xmath93 is not constant , then @xmath519 for all @xmath520 .
let us verify that @xmath464 is discrete .
suppose that for some @xmath521 we have @xmath522 .
if @xmath523 then @xmath524 is isolated . if @xmath525 then integrating the equation for @xmath92 we have @xmath526 for @xmath527 near @xmath524 we have @xmath528 in the second line above we have used that @xmath519 , @xmath529 and . in the third line above we may say that @xmath530 if @xmath527 is sufficiently close to @xmath524 by continuity , @xmath519 and ( we may regard @xmath2 as continuous here since we are working with values above @xmath137 ) .
this shows that @xmath92 is strictly convex in a neighborhood of @xmath524 and hence there are no other points in @xmath464 close to @xmath524 .
this shows that @xmath464 is discrete , hence finite , and finishes the proof of the theorem .
in this section we perform explicit calculations for the ignition nonlinearity @xmath37}$ ] , where @xmath3 .
we first reduce the differential equations to a finite number of equations in some parameters . in a second part
we obtain rigorously the bifurcation diagram of figure [ fig1 ] for small @xmath42 .
we rewrite in the form @xmath531 and @xmath532 where @xmath533 is a parameter to be adjusted .
observe that problem can explicitly be solved .
indeed , the following functions solve @xmath534 hence the last condition in becomes @xmath535 assuming that @xmath46 , in @xmath536 the subsystem is reduced to the following linear problem @xmath537 for @xmath538 , observe that the two following functions @xmath539 solve equations - .
> from the formula for @xmath93 we can see that @xmath540 now to obtain a solution of subsystem , it then remains to adjust @xmath27 and @xmath42 to get .
this equation is @xmath541 which after simplification yields @xmath542 let us define @xmath543 it can be shown that there is @xmath31 such that for @xmath544 , @xmath545 has 2 zeros , @xmath546 .
the function @xmath547 has a unique zero @xmath548 .
a calculation ( see lemma [ u > theta small eps ] ) shows that for @xmath8 small and @xmath549 the solution @xmath92 obtained by the formula satisfies @xmath46 in @xmath329 . along the lower branch @xmath550 , @xmath551 we observe numerically that @xmath46 in @xmath329 . after reaching @xmath55
we continue the curve along the upper branch @xmath552 now decreasing @xmath42 from @xmath55 to 0 until we reach a critical value @xmath553 where the corresponding solution given by satisfies @xmath46 in @xmath554 but @xmath555 .
[ analysis reaction annulus ] under this assumption the problem decouples again into two subsystems @xmath559 and @xmath560 where @xmath561 are parameters to be found . using the conditions at @xmath562 and @xmath563 we will be able to reduce the parameters to only @xmath27 and @xmath84 which will be implicitly defined as functions of @xmath42 by 2 equations .
observe that since @xmath93 is harmonic in @xmath558 with the dirichlet condition @xmath564 , we deduce that @xmath565 and therefore @xmath566 .
let us now solve explicitly .
first , observe that the function @xmath567 satisfies @xmath568 therefore , choosing @xmath569 and @xmath67 such that @xmath570 + b[\beta sh(\beta ) -ch(\beta ) ] , \end{aligned}\ ] ] it follows that @xmath93 satisfies the right boundary conditions on @xmath571 .
solving for @xmath569 and @xmath67 we find @xmath572 to obtain a full solution it remains to adjust @xmath533 in such way that @xmath573 . but @xmath574
+ b[\frac{\beta sh(\beta r)}{r } -\frac{ch(\beta r)}{r^2}]\ ] ] and setting @xmath575 in the above equation yields @xmath576 + b\left[\frac{\beta sh(\beta \eta)}{\eta } -\frac{ch(\beta \eta)}{\eta^2}\right ] = 0 \\ & a\big(\eta\beta
ch(\beta \eta)-sh(\beta \eta)\big ) + b \big ( \eta\beta sh(\beta \eta)-ch(\beta \eta)\big)=0\end{aligned}\ ] ] substituting now the values of @xmath569 and @xmath67 in the above equation , it follows @xmath577 where @xmath578 therefore , we have @xmath579 which after simplification reads @xmath580}{\beta\left[\big(\eta\beta- th(\eta \beta)\big)th(\beta ) + 1 - \eta\beta th(\eta\beta ) \right ] } .\ ] ] thus we may compute now @xmath581 : @xmath582 > from we know that in @xmath558 , we have the following @xmath583 therefore , @xmath584=\frac{\theta ch(\eta\beta\sqrt{{\varepsilon}})}{\eta sh(\eta\beta\sqrt{{\varepsilon}})}[\eta\beta\sqrt{{\varepsilon } } -th(\eta\beta\sqrt{{\varepsilon}})].\end{aligned}\ ] ] let us now find the solution @xmath92 in @xmath556 .
as for the construction of @xmath93 , let us observe that for @xmath538 , @xmath585 solves the following equation @xmath586 by taking @xmath587 we easily verify that @xmath588 and @xmath589 .
it remains now to impose the boundary condition on @xmath590 . from the formula for @xmath92 we have @xmath591 since @xmath573 .
thus the equation @xmath592 becomes @xmath593 and combined with becomes @xmath594 . &
\end{aligned}\ ] ] we solve numerically equations and with @xmath27 and @xmath84 as unknowns that depend on @xmath42 , for @xmath595 , where @xmath416 is the critical value of @xmath42 described in subsection [ sub reaction ball ] . the result from this numerical computation
is shown in figure [ fig1 ] .
[ u > theta small eps ] for @xmath8 suitably small @xmath550 is a smooth function of @xmath42 and the solution given by corresponding to @xmath549 satisfies @xmath46 in @xmath329 .
let @xmath430 the unique solution of the equation @xmath596 then @xmath597\end{aligned}\ ] ] for @xmath598 , using that @xmath599 , we end up with @xmath600 computing now the positive roots of @xmath601 , yields @xmath602 for @xmath603 . therefore , we achieve @xmath604 , for @xmath42 in @xmath605 .
let us denote @xmath606 hence , for any @xmath607 in @xmath608 fixed there exists two possible solution to @xmath609 .
moreover , @xmath610 . to obtain a solution , to problem
, we still need to show that the constructed solutions effectively satisfies the conditions @xmath46 in @xmath536 .
let us observe that @xmath611 satisfies : @xmath612 if @xmath613 then using the maximum principle , it follows that @xmath614 . since @xmath615 , it follows that @xmath616 .
hence , we end up with the condition @xmath617 we can conclude with a final estimate on @xmath42 , namely since @xmath618 , and @xmath619 is a decreasing function , we have the following uniform estimates on @xmath42 @xmath620 we therefore have construct a solution to the problem ( 1 ) , when @xmath607 belongs to @xmath621 where @xmath622 for @xmath8 small the system , has a unique solution @xmath623 with @xmath85 , @xmath552 is a smooth function of @xmath42 and @xmath92 defined by satisfies @xmath46 in @xmath624 . the 2 equations and can be written in the form @xmath625 where @xmath626 instead of @xmath42 , @xmath27 , @xmath84 consider the variables @xmath627 , @xmath628 , @xmath629 defined by the following relations @xmath630 consider also the function @xmath631 where @xmath632 \\ g_2(t , a , x ) & = e \left ( a \eta \cosh(a \eta ) - \sinh(a \eta ) \right ) + f \left ( a \eta \sinh(a \eta ) - \cosh(a \eta ) \right ) \\ & \quad - \eta \theta\left ( \frac{a \eta}{\tanh(a\eta ) } - 1 \right ) \end{aligned}\ ] ] where @xmath633 are as before .
we see that we have a solution to the system if and only if @xmath634 . using the fact that @xmath635 it is not difficult to verify that @xmath636 maybe extended in a @xmath637 manner for @xmath343 , @xmath638 with the values @xmath639 note that the equation @xmath640 is equivalent to @xmath641 which is seen to have a unique solution @xmath642 since the left hand side defines a strictly increasing function of @xmath628 which goes to zero as @xmath643 and goes to @xmath644 as @xmath645
now the equation @xmath646 can be written as @xmath647 the above left hand side is strictly increasing with limits @xmath648 as @xmath649 and 1 as @xmath650 . using the inequality @xmath651
we see that @xmath652 which yields @xmath653 .
hence @xmath654 thus has a unique solution @xmath655 .
as mentioned before the largest value of @xmath42 such that has a nontrivial solution satisfies @xmath656 where @xmath657 \ } , \end{aligned}\ ] ] and @xmath658 in the particular case of @xmath37}$ ] , @xmath3 we obtain the value @xmath659 .
this estimate can be sharpened . for this , given @xmath660
let us introduce @xmath661 for @xmath662 define also @xmath663 as the smallest zero of @xmath664 in the interval @xmath554 .
we @xmath665 be the largest @xmath666 in @xmath252 $ ] such that @xmath664 is positive on @xmath667 . before proving the above proposition , let us recall a result on the scalar problem on a bounded domain @xmath670 : @xmath671 there is a vast literature concerning the existence of positive solutions of the above equation .
we will just mention a necessary and sufficient condition for the existence of positive solutions .
[ cn ] ( clment and sweers @xcite ) let @xmath670 be a bounded smooth domain and @xmath672 .
assume there exists @xmath673 with @xmath674 such that @xmath675 and @xmath676 on @xmath677 .
then there exists positive solution @xmath678 to with @xmath679 if and only if @xmath680 for all @xmath681 .
fix @xmath682 . from our assumption , we have @xmath683 .
let @xmath684 a smooth function such that @xmath685 and @xmath686 .
this is always possible since @xmath683 .
observe now that from the necessary and sufficient condition given in theorem [ cn ] , the following scalar problem @xmath687 does not have any positive solution for any bounded domain @xmath688 .
now , we argue by contradiction , assume there exists a non trivial couple @xmath6 solution of the system .
an easy computation shows that @xmath92 is a sub - solution of the problem with @xmath689 .
take @xmath690 small , and define @xmath691 so for @xmath77 small enough,(i.e @xmath692),we have : @xmath693 therefore @xmath694 does not satisfy the necessary existence condition of theorem [ cn ] .
now , take @xmath695 .
since @xmath92 goes uniformly to 0 when @xmath696 goes to infinity , we have @xmath697 as @xmath698 .
thus , with @xmath699 , @xmath700 is a sub - solution of the following scalar problem : @xmath701 observe that the constant @xmath702 is a super - solution of and @xmath703 , then we can apply the monotone iterative scheme to obtain at least one positive solution which contradicts theorem [ cn ] . in the case @xmath37}$ ]
where @xmath3 with a similar argument we can show that for @xmath704 the system has no nontrivial solution .
indeed , given any @xmath425 choose a smooth function @xmath705 such that @xmath706 for @xmath707 and @xmath708 for @xmath709 . with the same argument as before , has no nontrivial solution for @xmath710 where @xmath711 is given by with @xmath2 replaced by @xmath712 .
a computation then shows that as @xmath713 , @xmath714
. * acknowledgment . *
j. coville warmly thanks professor sivashinsky for proposing the formulation of this problem and enlightening discussions .
he is also indebted to professor dolbeault for useful conversations on this problem .
j. coville was partially supported tel aviv university , universit paris dauphine and max planck institute for mathematics in the science .
j. dvila was supported by fondecyt 1057025 , 1090167 .
both authors acknowledge the support of ecos - conicyt project c05e04 .
h. berestycki , b. larrouturou , _ quelques aspects mathmatiques de la propagation des flammes prmlanges .
_ nonlinear partial differential equations and their applications . collge de france seminar , vol .
x ( paris , 19871988 ) , 65129 , pitman res . notes math . ser .
, 220 , longman sci . tech . ,
harlow , 1991 .
h. berestycki , b. larrouturou , j .-
roquejoffre , _ mathematical investigation of the cold boundary difficulty in flame propagation theory .
_ dynamical issues in combustion theory ( minneapolis , mn , 1989 ) , 3761 , i m a vol . math .
, 35 , springer , new york , 1991 .
j. brindley , n. a. jivraj ; j. h. merkin ; s. k. scott , _ stationary - state solutions for coupled reaction - diffusion and temperature - conduction equations .
ii . spherical geometry with dirichlet boundary conditions .
london ser . a 430 ( 1990 ) , no .
1880 , 479488 .
j. buckmaster , g. joulin , p. ronney , _ the structure and stability of nonadiabatic flame balls .
_ combustion and flame * 79 * ( 1990 ) , no . 3 - 4 , 381392 .
v. hutson , j. lpez - gmez , k. mischaikow , and g. vickers , _ limit behaviour for a competing species problem with diffusion _ , dynamical systems and applications , world sci .
, vol . 4 , world sci .
publ . , river edge , nj , 1995 , pp .
343358 . c. lederman , j .-
roquejoffre , n. wolanski , _ mathematical justification of a nonlinear integrodifferential equation for the propagation of spherical flames .
_ ann . mat .
pura appl .
( 4 ) 183 ( 2004 ) , no .
2 , 173239 . | in this paper , we construct radially symmetric solutions of a nonlinear noncooperative elliptic system derived from a model for flame balls with radiation losses .
this model is based on a one step kinetic reaction and our system is obtained by approximating the standard arrehnius law by an ignition nonlinearity , and by simplifying the term that models radiation .
we prove the existence of 2 solutions using degree theory . |
the study of the relation between classical and quantum dynamical chaos has greatly improved our understanding of the behavior of quantum systems .
an issue which may be of interest in several situations is the stability of motion . in this respect , even though liouville equation , which governs the evolution of classical distribution functions , is linear , the exponential sensitivity of classical trajectories with respect to perturbing the initial conditions leads to a strong dynamical instability , and characterizes classical chaos . like the classical liouville equation , the schrdinger equation is also linear .
however , the quantum evolution of states is stable and this qualitative difference is clearly apparent in the loschmidt echo numerical experiments of ref .
the problem of the stability of quantum motion under perturbations in the hamiltonian has recently gained a renewed interest @xcite , also in connection with quantum computation @xcite .
the quantity of central interest in these investigations is the fidelity @xmath0 ( also called loschmidt echo ) , which measures the accuracy to which a quantum state can be recovered by inverting , at time @xmath1 , the dynamics with a perturbed hamiltonian .
the main interest has been focused on classically chaotic systems for which , besides numerical experiments , some theoretical tools are available , like random matrix theory and semiclassical methods . however , in the general case , the phase space structure is mixed with chaotic components and islands of stability . if the motion starts inside an integrable island , then it very much resembles the motion in integrable systems .
contrary to chaotic systems , which are dynamically unstable but structurally stable , integrable systems are dynamically stable but very sensitive to external perturbations .
therefore the analysis of the fidelity requires particular care and one may expect it to be dependent on initial conditions and on the type of perturbation .
indeed , the decay of fidelity in integrable systems has been discussed in recent papers @xcite , and very different behaviors have been found .
et al . _ have shown the existence of a regime in which the quantum fidelity for classically integrable systems decays as a power law , with an anomalous exponent of purely quantum origin @xcite .
prosen and nidari have instead discussed a regime in which quantum fidelity exhibits a much faster gaussian decay @xcite .
both regimes have been also discussed by eckhardt in his analysis of the decay of classical fidelity @xcite , in which the problem of the evolution of classical phase space densities has been addressed for linearized flows . in the present paper ,
we discuss the behavior of fidelity for integrable classical systems . besides being of interest on its own , our classical study will allow us to understand the main mechanisms for the fidelity decay and therefore will constitute a valuable reference point for the quantum analysis . here
we show the existence of a critical border depending on the _ shape _ of the perturbation , which separates two different types of fidelity decay : a power law decay @xmath2 , where @xmath3 is the dimension of the system , and a much faster decay of ballistic type .
we stress that the type of decay depends on initial conditions and on the shape of the perturbation but , for small enough perturbations , not on its strength .
we derive an analytical expression for the critical border and our theoretical results are confirmed by a numerical analysis on three different models : the twist map , the rectangular billiard and the kicked rotor .
the outline of the paper is as follows : in section ii we develop a general theory for the decay of classical fidelity in integrable systems .
section iii demonstrates numerically the validity of our theory in two different typical examples of integrable systems , the twist map and the rectangular billiard , and in an almost integrable system , the kicked rotor .
our conclusions are drawn in section iv .
finally , in the appendix a , we discuss the long - time relaxation to equilibrium .
the quantum fidelity is defined as the overlap at time @xmath1 of the states @xmath4 and @xmath5 , obtained by the evolution of the same initial state @xmath6 with the unperturbed hamiltonian @xmath7 and the perturbed hamiltonian @xmath8 , respectively .
the fidelity is then given by @xmath9 this expression can be equivalently rewritten in terms of the wigner functions as @xmath10 where @xmath3 is the number of degrees of freedom . since the wigner functions can be considered as the quantum analogues of the classical phase space densities , we define the classical fidelity as @xmath11 where @xmath12 are the square normalized classical phase space densities ( @xmath13 ) .
we note that @xmath14 is the classical limit of @xmath0 . as the density evolution is unitary in both classical and quantum mechanics , instead of evolving two densities forward in time and calculating their overlap
, we may first evolve the initial density @xmath15 forward in time with the unperturbed hamiltonian @xmath7 , and then evolve this density backward in time with the perturbed hamiltonian @xmath8 .
we denote the density obtained in such a way as @xmath16 .
the fidelity is then given by the overlap of the density @xmath16 with the initial density @xmath15 : @xmath17 such an approach is more convenient for our discussion of the classical fidelity of integrable systems . for the following discussion
we assume that the perturbation of the integrable system is of the kam type , namely that , for small enough perturbations @xmath18 , most of the tori of the system are only slightly deformed but not destroyed .
therefore for most of the tori the transformation from old action - angle variables @xmath19 to new ones @xmath20 is possible , in such a way that the new actions are constants of motion of the perturbed system . to the first order in the perturbation strength @xmath21 the transformation can be written as @xmath22 after the forward unperturbed evolution up to time @xmath1 we have @xmath23 then we perform a backward evolution of the perturbed system from time @xmath1 to time @xmath24 , getting @xmath25 the frequency vectors @xmath26 and @xmath27 characterize the linear ( in time ) evolution of the angle variables in integrable systems .
expressed in the original action - angle variables , the overall evolution can be written as @xmath28 where the error term @xmath29 is due to the change from one set of variables to the other at time @xmath1 and the reverse change at time @xmath24 . since the perturbation is small , we may write the change in frequency as @xmath30 where @xmath31 denotes the change of frequencies on the unperturbed torus @xmath32 and @xmath33 gives the change of the action variables caused by the perturbation ( written in eq .
( [ eq : transi ] ) to the first order in @xmath21 ) .
if we consider the angle variables to be uniformly distributed at the time @xmath1 at which the motion is inverted , we may introduce the distribution @xmath34 , \label{eq : probaction}\ ] ] which gives the probability density for the transition from the torus characterized by the action variables @xmath32 in the unperturbed coordinates to the torus with action variables @xmath35 in the perturbed coordinates .
the @xmath21 scaling has been chosen in order for the function @xmath36 itself not to depend on @xmath21 in the linear approximation . in the generic case ,
the motion on a torus is ergodic .
it is , however , not random and therefore , in order for the equation ( [ eq : probaction ] ) to well describe transitions between the tori , one needs to consider an ensemble of tori in the vicinity of the chosen actions @xmath32 , as only for an ensemble of tori with different frequencies we can expect the angle variables to be uniformly distributed after a sufficiently long time @xmath1 .
indeed , the spread of frequencies given by @xmath37 translates into the spread of angle variables @xmath38 .
the time for this spread in the angle variable to become comparable to @xmath39 is @xmath40 where @xmath41 is the characteristic width of the initial phase space density distribution @xmath15 along the action direction ( to simplify writing , we have given eq .
( [ ttheta ] ) for the one - dimensional case ) .
thus , our theory based on eq .
( [ eq : probaction ] ) is valid for times @xmath42 .
since the final angle variables , after the forward and backward evolutions , depend on the actions of the perturbed system , we compute the distribution of the angle variables from the distribution of the perturbed actions as @xmath43 using the expressions ( [ eq : theta2 t ] ) and ( [ eq : omegaexpand ] ) , we obtain @xmath44 @xmath45^{-1 } \left(\frac{{\bf \theta}_{0}- { \bf \theta}_{2t}+{\cal o } ( \epsilon ) } { \epsilon t}-\frac{\delta { \bf \omega}}{\epsilon}\right)\right ) .
\label{eq : main}\ ] ] this expression is the kernel for the combined forward and backward evolution of the phase space densities .
we assume that the width @xmath41 along the action direction of the initial density @xmath15 is much larger than the change of the action variable induced by the perturbation , that is @xmath46 ( to simplify writing we have given the condition ( [ nui ] ) for the one - dimensional case ) .
therefore the effects of the forward and backward evolutions are felt mainly in the change of the angles variable .
this means that the evolution from the initial phase space density @xmath15 to @xmath16 is given , up to corrections of order @xmath21 , by @xmath47 the fidelity @xmath14 can then be computed by inserting @xmath16 into eq.([fidc2 ] ) .
the kernel @xmath48 is stretched linearly in time , while at the same time it moves ballistically ( linearly with time ) with velocity @xmath49 . under the assumption that the perturbation of the shape of the tori is not divergent ( as it is the case for most of the tori in a kam regime )
, the distribution @xmath50 has a bounded support which is determined by the change of the shape of the tori due to the perturbation .
we can see from eq .
( [ eq : main ] ) that at long times the argument of the function @xmath36 is given by @xmath51^{-1 } ( \delta { \bf \omega}/\epsilon)$ ]
. therefore the long time behavior of @xmath52 depends on whether the value of @xmath51^{-1 } ( \delta { \bf \omega}/\epsilon)$ ] falls within the support of @xmath36 or not . in the first case , @xmath53^{-1 } ( \delta { \bf \omega}/\epsilon))$ ] is different from zero and therefore the kernel @xmath52 drops @xmath2 . in the latter case , @xmath54 , and therefore @xmath52 drops
the transition between these two regimes is determined by the equality @xmath55^{-1 } \frac{\delta { \bf \omega}}{\epsilon } , \label{deltais}\ ] ] where @xmath56 are the coordinates of the border of the support of @xmath36 .
we can therefore draw the following conclusions : if the perturbation of a classical integrable system is such that the primary effect is the change of the shape of the tori , then the expected decay of fidelity is @xmath2 . on the contrary , if the change of the frequencies of the tori is the dominant effect , then we expect a ballistic decay of fidelity , that is the center of mass motion of the phase space densities after the forward and backward evolutions is responsible for a drastic drop of fidelity .
such a decay takes place as soon as the centers of mass of the densities @xmath16 and @xmath15 are separated in the angle variables @xmath57 by more than their characteristic width @xmath58 . as it can be seen from eq .
( [ fidc2 ] ) , the exact form of the fidelity drop in the ballistic regime depends on the tails of the initial distribution @xmath15 : for instance , a gaussian tail gives a gaussian decay of fidelity , whereas a sharp border induces a sharp drop to zero of fidelity .
finally , it is important to stress that the type of decay , power law or ballistic , depends on initial conditions and on the shape of the perturbation .
however , it does not depend on the strength of the perturbation , provided that it is sufficiently small . indeed , eq . ( [ deltais ] )
shows that @xmath56 is @xmath21-independent ( to the first order in @xmath21 ) , since @xmath59 .
as a first example we consider the perturbed twist map , defined by @xmath60 where the angle @xmath61 determines the mixture between purely perturbing the shape of the tori ( @xmath62 ) or purely changing their frequencies ( @xmath63 ) .
this parametrization allows us to change the type of the perturbation without changing its overall magnitude .
the change of frequency associated with the perturbation is given by @xmath64 the conserved action variable of the @xmath21-perturbed system is , to the first order ( in @xmath21 ) approximation , given by @xmath65 indeed , inserting this expression into the mapping ( [ eq : skew2 ] ) , one can easily verify that @xmath66 .
the transition probability function @xmath67 for this system can thus be obtained by means of eq .
( [ eq : probaction ] ) : @xmath68 @xmath69,\ ] ] which gives
@xmath70 the range of the support of the distribution @xmath67 is between @xmath71 .
the critical value @xmath72 for which @xmath73 is therefore determined by @xmath74 in fig .
[ fig : skew ] we show the numerically computed behavior of fidelity for this system as a function of time for various values of the parameter @xmath61 .
we take as initial phase space density a rectangle centered around the point @xmath75 with sides of length @xmath76 .
the perturbation strength is @xmath77 . to compute fidelity
, we follow the evolution of @xmath78 trajectories , which at @xmath79 are uniformly distributed inside the above rectangle .
the fidelity is then given by the percentage of trajectories that return back to this region after the forward and backward evolutions . in all cases
we observe an initial plateau during which the fidelity does not decay appreciably .
this plateau persists until the time @xmath80 at which the width of the kernel ( [ eq : main ] ) or the shift of its center become comparable to the width @xmath81 of the phase space density along the angle variable . in either case
this time is @xmath82 according to eq .
( [ deltais ] ) , we expect the behavior to change from algebraic decay to ballistic one at the value of the parameter @xmath83 .
for the chosen initial conditions , eq .
( [ eq : critskew ] ) gives @xmath84 .
indeed , the change from an algebraic fidelity decay @xmath85 when @xmath86 to a sharp drop of fidelity when @xmath87 is clearly seen in fig .
[ fig : skew ] .
an interesting feature is that , approaching the critical value @xmath72 , we observe that the fidelity decay , power law or ballistic , sets in after longer and longer times .
this fact has a clear explanation : the value @xmath88^{-1 } ( \delta { \bf \omega}/\epsilon))$ ] , which determines the long time behavior of the evolution kernel ( [ eq : main ] ) , diverges close to the critical value @xmath83 ( see eq .
( [ eq : distskew ] ) ) .
when the fidelity decay is power law , we have @xmath89 , with the constant @xmath90 . since @xmath91 becomes larger and larger close to the critical point , the fidelity decay must be postponed to longer and longer times . on the other hand ,
when the long time fidelity decay is ballistic , we can see from eq .
( [ eq : main ] ) that the argument of the function @xmath36 goes outside the support of @xmath36 after a time that becomes longer close to the critical point . only after this time
the fidelity drops off .
of course the decay can not be postponed indefinitely since the exact condition ( [ eq : critskew ] ) can be satisfied only for a single torus , while we always deal with a family of tori upon which the initial phase space density @xmath15 rests .
0 ( full line ) , 0.8 ( dashed ) , 0.892 ( dot - dashed ) , 1.0 ( dot - dot - dashed ) and @xmath92 ( dotted ) . the @xmath93 decay is shown as a thin dotted line , width=302 ] we also checked numerically that , provided that the perturbation is much smaller than the characteristic widths @xmath41 of the initial density ( that is , the requirement ( [ nui ] ) is fulfilled ) , the type of behavior does not alter with changing the actual size of the perturbation @xmath21 , as it is expected from our theory .
we simply rescale the time @xmath94 after which the fidelity decay starts , in agreement with eq .
( [ tplateau ] ) . to illustrate the fidelity decay in integrable systems with more than one degree of freedom , we consider the following system : @xmath95 where the unperturbed hamiltonian @xmath96 describes the motion of a particle bouncing elastically inside a rectangular billiard and the perturbation is given by @xmath97 again , depending on the value of the parameter @xmath98
, the perturbation mainly affects either the shape of the tori or their frequencies .
we use the first order perturbation theory of hamiltonian systems ( see , e.g. , ref . @xcite ) to determine the effects of the perturbation .
what we need to find is a set of action - angle coordinates such that , to the first order in the perturbation strength @xmath21 , the hamiltonian ( [ hrectangle ] ) in these new coordinates can be written as a function of only the new actions , namely @xmath99 introducing the generating function @xmath100 @xmath101,\ ] ] we get @xmath102 @xmath103,\ ] ] @xmath104 @xmath105.\ ] ] substituting the above expressions into the hamiltonian ( [ hrectangle ] ) , we get @xmath106 the new frequencies are then given by @xmath107 @xmath108 thus the frequencies changes read as follows : @xmath109 @xmath110 as in the previous example , the above expressions allow us to find the transition probability function @xmath111 it can be seen that the support for the distribution @xmath112 is the rectangle @xmath113 and @xmath114 . 0 ( full line ) , 0.232 ( dashed ) , 0.3 ( dot - dashed ) and @xmath92 ( dotted ) . the @xmath115 decay is shown as a thin dotted line.,width=302 ] in fig .
[ fig : rect ] we show the decay of fidelity for this system for various values of @xmath98 .
the parameters of the system have been chosen as follows : @xmath116 , @xmath117 . in all cases
the initial phase space density is a hyper - rectangle centered around @xmath118 , @xmath119 , @xmath120 and @xmath121 with all sides of length @xmath122 .
the perturbation parameter is @xmath123 and the number of trajectories @xmath124 . for the above initial conditions and parameters @xmath125 , @xmath126 , the critical value @xmath127 which separates the power law and the ballistic fidelity decay
is determined by the equality ( [ deltais ] ) in the direction of the @xmath128 variable .
indeed , when @xmath98 increases , @xmath129^{-1}(\delta { \bf \omega}/\epsilon)$ ] goes outside the support of @xmath36 at first along this direction .
this gives @xmath130_u , \label{betau}\ ] ] where the right - hand side is the @xmath128-component of the vector @xmath131 therefore we get @xmath132 substituting the chosen values of @xmath133 and @xmath126 , we find that the critical value is equal to @xmath134 .
this theoretical expectation is confirmed by the numerical data of fig .
[ fig : rect ] , which show a crossover from a power law fidelity decay ( for @xmath135 ) to a ballistic decay ( for @xmath136 ) .
the results are very similar to the case of the twist map , including the fact that , close to the critical value @xmath137 , the decay is postponed to longer times .
is should be stressed that the algebraic fidelity decay , differently from the twist map case , is now inversely proportional to the square of the time , in agreement with our theoretical expectation for a two - dimensional system ( see eq .
( [ eq : main ] ) ) . , @xmath138 , and @xmath78 trajectories , using two different initial phase space densities , one centered at the point @xmath139 ( full line ) and the other centered at @xmath140 ( dashed line ) . in the first case the fidelity decays ballistically , in the latter case inversely proportional to time .
both initial densities are @xmath141 wide in the @xmath142 and @xmath143 directions .
the @xmath93 curve is shown as a thin dotted line.,width=302 ] as a last numerical example , we consider the kicked rotor map that is given by @xmath144 @xmath145 as it is known , for @xmath146 the system is almost integrable , namely its phase space is dominated by invariant tori .
there is a stable fixed point at @xmath147 and a separatrix which divides the phase space into two regions : a section of librational motion around the stable fixed point inside the separatrix , and a section of rotational motion outside the separatrix .
we perturb the system by varying @xmath148 .
the important point is that the type of the perturbation chosen strongly affects the frequencies of the tori in the librational section , while it mainly perturbs the shape of the rotational tori .
therefore the same system and perturbation should lead to two completely different types of fidelity decays , power law or ballistic , depending on the choice of the initial conditions .
[ fig : rotor ] confirms this expectation : if the initial density @xmath15 is inside the separatrix the fidelity decay is ballistic , otherwise it is power law .
in this work we have studied the decay of the fidelity of classical motion for integrable systems .
our main result is the following : for small enough perturbations , the type of the decay of fidelity for integrable systems depends not on the strength of the perturbation but on its shape and on initial conditions . more precisely , the fidelity exhibits two completely different behaviors , namely an algebraic decay if the perturbation mainly affects the shape of the tori , and a faster , ballistic decay , if the main effect of the perturbation is to change the frequencies of the tori .
we have also given clear numerical demonstrations of the transition between the two types of behaviors , induced by changing the shape of the perturbation or the initial conditions .
this result poses interesting questions with respect to the quantum mechanical picture . due to the correspondence principle
, there should exist regimes where both types of decay may be observed .
it is however expected that , for small perturbations , quantum mechanics would favor the ballistic type decay , as demonstrated in @xcite .
indeed the algebraic decay is due to the transitions between tori which , for small perturbations , are suppressed in quantum mechanics , due to tori quantization and subsequent gaps between them .
the classical - quantum correspondence will be the topic of further studies .
this work was supported in part by the ec rtn contract hprn - ct-2000 - 0156 , the nsa and arda under aro contract no .
daad19 - 02 - 1 - 0086 , the pa infm `` weak chaos : theory and applications '' , and the prin 2002 `` fault tolerance , control and stability in quantum information precessing '' .
the results of the previous sections do not tackle the asymptotic decay of fidelity for integrable systems @xcite .
indeed , we neglected the contributions to the evolution kernel ( [ eq : main ] ) that stem from the fact that the angle variables are cyclic .
this means that , after a time which is @xmath149 , we need to take into account the contributions to ( [ eq : main ] ) not only at @xmath150 but also at all @xmath151 , where @xmath152 is a vector of integer numbers .
we limit ourselves to the case of a single torus . of course the fidelity
@xmath14 is strictly zero for a single torus , and therefore it should be understood that we take the limits @xmath153 let us consider the initial density @xmath154 to be defined on the whole @xmath57 space ( without @xmath155 periodicity ) , while the kernel @xmath156 is defined as the periodic function obtained from the original kernel : @xmath157 assuming that the initial density is square normalized , the fidelity can be written as @xmath158 where @xmath159 denotes the fourier transform : @xmath160 since @xmath161 in the fourier picture this becomes @xmath162 we may write the original kernel ( [ eq : main ] ) in the simplified form @xmath163 where @xmath164^{-1 } \frac{{\bf y}}{\epsilon } \right ) \left|\frac{\partial { \bf \omega}}{\partial { \bf i}}\right|^{-1 } $ ] and @xmath165 .
the fourier transform of the kernel ( [ eq : perker ] ) is therefore given by @xmath166 then the formula @xmath167 leads to @xmath168 this result , coupled with equations ( [ eq : fidfour ] ) and ( [ eq : evolfour ] ) , finally leads to @xmath169 as we can see , the behavior of fidelity in the limit @xmath170 is given by the tails of the fourier transform of the kernel @xmath171 .
the origin of the kernel is the projection of the perturbed tori onto unperturbed ones , and we expect singularities in such a projection . these singularities induce a power law decay in the tails of the fourier transform of the kernel and thus are responsible for the asymptotic power law decay of fidelity . in the single degree of freedom situation , the typical singularity of projection to be encountered leads to @xmath172 , as it can also be seen in the twist map example ( [ eq : distskew ] ) .
this type of singularity leads to the fourier transform @xmath173 such an expression leads to following asymptotic fidelity decay : @xmath174 @xmath175 where @xmath176 and @xmath177 is some periodic function with period @xmath39 .
we note that eq .
( [ asymptotic ] ) gives an overall @xmath178 fidelity decay together with a superimposed oscillatory behavior .
this is the typical asymptotic relaxation of fidelity for a single torus in integrable systems with a single degree of freedom .
if one considers a finite interval of actions @xmath41 , the decay ( [ asymptotic ] ) must be averaged over @xmath41 , and therefore , due to the oscillatory nature of eq .
( [ asymptotic ] ) , it can be faster that @xmath179 .
the extension to the many - dimensional case requires a complex analysis of the singularities encountered in the projection of the perturbed tori onto the unperturbed ones and is beyond the scope of the present paper .
99 g. casati , b.v .
chirikov , i. guarneri , and d.l .
shepelyansky , phys .
lett . * 56 * , 2437 ( 1986 ) .
a. peres , phys . rev .
a * 30 * , 1610 ( 1984 ) .
jalabert and h.m .
pastawski , phys .
lett , * 86 * , 2490 ( 2001 ) .
cucchietti , h.m .
pastawski , and d.a .
wisniacki , phys .
e * 65 * , 045206(r ) ( 2002 ) .
jacquod , p.g .
silvestrov , and c.w.j .
beenakker , phys .
e * 64 * , 055203(r ) ( 2001 ) , n.r .
cerruti and s. tomsovic , phys .
lett . * 88 * , 054103 ( 2002 ) .
v.v . flambaum and f.m .
izrailev , phys .
e * 64 * , 036220 ( 2001 ) .
karkuszewski , c. jarzynski , and w.h .
zurek , phys .
lett . * 89 * , 170405 ( 2002 ) .
g. benenti and g. casati , phys .
e * 65 * , 066205 ( 2002 ) .
d.a . wisniacki and d. cohen , phys .
e * 66 * , 046209 ( 2002 ) .
t. prosen and m. nidari , j. phys .
a * 35 * , 1455 ( 2002 ) .
weinstein , s. lloyd , and c. tsallis , phys .
lett . * 89 * , 214101 ( 2002 ) .
w. wang and b. li , phys .
e * 66 * , 056208 ( 2002 ) .
y. adamov , i.v .
gornyi , and a.d .
mirlin , preprint cond - mat/0212065 . i. garcia - mata , m. saraceno , and m.e .
spina , preprint nlin.cd/0301025 .
j. vanicek and e.j .
heller , preprint quant - ph/0302192 .
t. prosen and m. nidari , j. phys .
a * 34 * , l681 ( 2001 ) .
g. benenti , g. casati , s. montangero , and d.l .
shepelyansky , phys .
lett . * 87 * , 227901 ( 2001 ) .
j. emerson , y.s .
weinstein , s. lloyd , and d.g .
cory , phys .
lett . * 89 * , 284102 ( 2002 ) .
b. eckhardt , j. phys .
a * 36 * , 371 ( 2003 ) ph .
jacquod , i. adagideli , and c.w.j .
beenaker , europhys . lett .
* 61 * , 729 ( 2003 ) .
corben and p. stehle , _ classical mechanics _
( 2nd edition ) , ( robert e. kreiger publishing , new york , 1974 ) p. 244
for chaotic systems this problem has been addressed by g. benenti , g. casati , and g. veble , preprint nlin.cd/0208003 , to be published in phys . | we study both analytically and numerically the decay of fidelity of classical motion for integrable systems .
we find that the decay can exhibit two qualitatively different behaviors , namely an _ algebraic decay _
, that is due to the perturbation of the shape of the tori , or a _ ballistic decay _ , that is associated with perturbing the frequencies of the tori .
the type of decay depends on initial conditions and on the shape of the perturbation but , for small enough perturbations , not on its size .
we demonstrate numerically this general behavior for the cases of the twist map , the rectangular billiard , and the kicked rotor in the almost integrable regime . 2 |
the exclusive electroproduction of mesons on the nucleon is an important tool to better understand the nucleon structure and , more generally , the transition between the low energy hadronic and high energy partonic domains of the quantum chromodynamics ( qcd ) theory . the cebaf ( `` continous electron beam accelerator facility '' ) of the jefferson laboratory ( jlab ) at newport news ( usa ) with its high intensity electron beam and its large acceptance spectrometer clas @xcite ( cebaf large acceptance spectrometer ) offers a great opportunity to study the exclusive electroproduction of mesons , in particular those that have multi - particle decays .
this proceeding presents some first results for the exclusive electroproduction of the vector mesons @xmath0 on the proton at clas .
these results come from the e1-dvcs experiment : the data were collected between march and may 2005 with a beam energy of 5.75 gev and an integrated luminosity of @xmath2 fb@xmath3 .
it is the first ever measurement of cross section for the @xmath0 electroproduction channel .
the presented results are still preliminary and more details can be found in ref . @xcite .
the final state that is analyzed is @xmath4 . the final state particles @xmath5 , @xmath6 and @xmath7 were detected in clas and the exclusive process was identified by selecting a neutron with the missing mass technique .
figure [ fitq2xb ] shows the acceptance - corrected invariant mass @xmath8 distributions for each @xmath9 bin .
one sees clearly the @xmath10 peaks around 775 mev .
they are , however , sitting on top of a non - negligeable non - resonant , two - pion background .
we have used two contributions to fit these spectra : a skewed breit - wigner distribution to describe the resonant structure of the @xmath0 and a parametrization of the non - resonant two - pion continuum determined from simulations ( mostly @xmath4 phase space ) . in order to extract the total cross section of the reaction @xmath11 , we have calculated for each @xmath9 bin the area of the breit - wigner ( green curve in fig .
[ fitq2xb ] ) , the distributions being already normalized and corrected for the acceptance .
a separation of the longitudinal cross sections from the transverse ones was performed by analysing the pion decay angles of the @xmath0 and relying on the schc ( @xmath12-channel helicity conservation ) , which was checked experimentally .
acceptance - corrected distributions , showing fits for the background subtraction . in black : experimental data ; in green : breit - wigner of @xmath0 ; in purple : @xmath8 projection of the the non - resonant continuum @xmath13 reaction ; in red : total fit result . in this proceeding , units are arbitrary for the vertical scale.,width=529 ]
we will compare in the following the extracted cross sections of the exclusive electroproduction of the @xmath0 with two theoretical approaches : on the one hand , the hadronic approach based on regge theory and meson trajectory exchanges in the @xmath1-channel and on the other hand , the partonic approach based on the handbag diagram and gpds .
these two approaches are illustrated in fig .
[ reggegpds ]
. this comparison will be useful in order to better understand the domain of validity of these two approaches and constrain their inputs .
-channel exchange ( left ) and of the handbag diagram ( right ) for exclusive vector meson electroproduction.,title="fig:",height=113]-channel exchange ( left ) and of the handbag diagram ( right ) for exclusive vector meson electroproduction.,title="fig:",height=113 ] the regge approach is most appropriate at low @xmath14 above the resonance region and at forward angles ( where the cross section is largest ) .
it consists of exchanges of meson `` trajectories '' in the @xmath1-channel . in the following
, we will use the `` jml '' acronym to refer to the specific model developed by j .-
laget @xcite . for the @xmath0 channel , the exchange trajectories are those of the @xmath6 and of the @xmath0 . the free parameters in this model are the coupling constants at the hadronic vertices ( most of them being well constrained ) and the mass scales of the electromagnetic form factors .
figure [ xsectionrhoplaget ] shows the total cross section @xmath15 as a function of @xmath16 for fixed @xmath14 compared with the jml calculation shown with the red curve .
the jml model can successfully reproduce the cross sections for almost all of our @xmath17 range .
the decreasing cross section with @xmath16 can be explained in the regge theory by an intercept ( @xmath18 ) of the @xmath0 trajectory which is dominant : @xmath19 . at fixed @xmath14 , for the reaction @xmath20 .
the red curves are the result of the jml model .
units are arbitrary on the y axis as the results are still preliminary.,width=529 ] the formalism of gpds is valid in the so - called bjorken regime , i.e. @xmath21 with @xmath22 finite .
it was proven @xcite that the dominant process for exclusive meson electroproduction , in the bjorken limit , is given by the so - called handbag diagram represented in fig .
[ reggegpds ] .
the handbag diagram is based on the notion of factorization in leading - order / leading - twist pqcd between a hard scattering process , exactly calculable in pqcd , and a nonperturbative nucleon structure part that is parametrized by the gpds . for mesons ,
the factorization of the handbag diagram is only valid for the longitudinal part of the cross section . in the following
we discuss the two particular gk @xcite and vgg @xcite gpd - based calculations that provide quantitative results for the longitudinal part of the exclusive meson cross sections .
figure [ theoryrhoplus ] shows longitudinal cross section of @xmath23 as a function of @xmath16 and for fixed @xmath14 .
the results of the calculations of the gk and vgg model are shown , respectively , with the red and the blue curves .
we see that both models fail to reproduce the data .
this discrepancy can reach an order of magnitude at the lowest @xmath16 values .
the trend of these particular gpd calculations is to decrease as @xmath16 decreases , whereas the data increase .
the same behavior was observed , in the low @xmath16 region , for the exclusive electroproduction of the @xmath24 @xcite . at fixed @xmath14 , for the reaction @xmath20 .
the red and the blue curves are the results of the gk and vgg models .
units are arbitrary on the y axis as the results are still preliminary .
, width=529 ] one can conclude that either the gpd formalism is not valid for our @xmath14 coverage ( due to important higher twist contributions for instance ) or there is a fundamental contribution , which is missing in the way the gpds are parametrized in the gk and vgg models .
the clas collaboration has measured the largest ever set of data for the exclusive electroproduction on the proton of the vector mesons @xmath24 @xcite , @xmath25 @xcite and @xmath26 @xcite .
figure [ tslopes ] shows the slope of the differential cross section @xmath27 for the @xmath0 , @xmath24 , @xmath25 and @xmath26 channels as a function of @xmath16 ( in the top part ) and as a function of @xmath14 ( in the bottom part ) .
one can see the same trends of this slope , @xmath28 , for all mesons channels , which can be interpreted in simple and intuitive terms in the following way : * @xmath28 increases with @xmath16 : the size of the nucleon increases as one probes the high @xmath16 values ( i.e. the sea quarks ) , which could mean that the sea quarks tend to extend to the periphery of the nucleon .
* @xmath28 decreases with @xmath14 : as we go to large @xmath14 , the resolution of the probe increases and we tend to see smaller and smaller objects . as a function of @xmath16 ( on the top ) and as a function of @xmath14 ( on the bottom ) for the @xmath24 , @xmath25 , @xmath26 and @xmath0 channels.,width=604 ]
we have presented ( preliminary ) first ever measurement of cross sections for the exclusive electroproduction of @xmath0 on the proton using the clas detector at jlab .
these cross sections can be interpreted with two approaches : * hadronic approach : the jml model describes well most of the features of the @xmath0 cross sections up to @xmath29 gev@xmath30 .
* partonic approach : gpd models fail to describe longitudinal @xmath0 cross sections especially for low @xmath16
. 9 b.a . mecking et al .
instr . meth .
a * 503 * , ( 2003 ) 513 .
a. fradi , thesis univ .
paris - sud at orsay , 2009 .
laget , phys .
b * 489 * ( 2000 ) 313 . ; + f. cano and j .- m .
laget , phys .
d * 65 * ( 2002 ) 074022 . ; + j. m. laget , phys .
d * 70 * ( 2004 ) 054023 . ; + f. cano and j. m. laget , phys .
b * 551 * ( 2003 ) 317 .
collins , l. frankfurt and m. strikman , phys.rev.d * 56 * ( 1997 ) 2982 .
goloskokov and p. kroll , eur .
* 42 * ( 2005 ) 281 ; eur.phys .
j.c * 50 * ( 2007 ) 829 .
m. vanderhaeghen , p.a.m .
guichon , m. guidal , phys . rev .
d * 60 * , 094017 ( 1999 ) .
s. morrow et al . , eur .
phys . j. a * 39 * , ( 2009 ) 5 .
l. morand et al . , eur .
phys . j. a * 24 * ( 2005 ) 445 .
j. p. santoro et al . , phys .
c * 78 * ( 2008 ) 025210 . | we present preliminary results of the exclusive electroproduction of @xmath0 on the proton at clas .
we discuss the interpretation of the cross sections in terms of @xmath1-channel reggeon exchanges and in terms of generalized parton distributions ( gpds ) formalism . |
understanding the gravitational collapse of a massive neutron star is of central importance for its connection to some of the most energetic astrophysical phenomena .
such an event may take place within a core - collapse supernova @xcite , or in the late stage of a binary neutron star merger @xcite , and is widely believed to power some types of gamma - ray bursts . in recent years
it has been realized that newly formed stellar mass black holes may be prodigious sources of electromagnetic energy , in addition to driving strong kinetic outflows @xcite .
we are faced with the exciting possibility of probing the most extreme forms of gravitational collapse using coordinated measurements of electromagnetic transients and gravitational waves .
refining models that make observational predictions for joint gravitational and electromagnetic radiation is critical in order to establish efficient observation campaigns for both traditional astronomers and gravitational - wave astronomers .
it is also a key step toward predicting the delay between peak electromagnetic and gravitational wave emission , the electromagnetic emission pattern ( i.e. the beaming angle ) and the electromagnetic and gravitational wave spectrum .
our focus here is on the evolution of the neutron star s magnetic field during its collapse to a black hole .
we employ fully self - consistent relativistic calculations that follow the dense stellar material as well as the strong electromagnetic and gravitational fields .
a particularly intense electromagnetic transient is generated if the initial magnetic field is very strong ( @xmath0 g ) , that is , if the star is a magnetar .
rapid rotation will enhance the energy of the transient , to a degree that can only be derived by a full time - evolution of the electromagnetic and gravitational fields .
rapid rotation also provides a context for generating strong magnetic fields : when the neutron star is accreting , the shear layer at its surface is a promising site for dynamo action . a distinct magnetosphere will emerge and hold off the accretion flow if even @xmath1 of the energy that is dissipated in the shear layer is converted to a poloidal magnetic field . understanding the evolution of the magnetic field around
an _ isolated _ star is therefore potentially of key relevance for the collapsar and binary merger scenarios of gamma - ray bursts .
recent success in studying the behavior of plasmas around magnetized , spinning , stable neutron stars within flat spacetime has been presented in refs .
@xcite . to study the collapse problem
, one needs general relativity to account for the role of spacetime curvature ; and general relativistic magnetohydrodynamics ( grmhd ) to determine the internal evolution of the star .
the electromagnetic ( em ) phenomena outside the star can be approached in a variety of ways : through a full grmhd calculation , which generally is very expensive given the low matter density ; or more approximately using vacuum em and force - free equations . in this paper
, we compare all three approaches ( restricting to ideal mhd in the case of grmhd ) . within general relativity , studies paying attention to neutron star collapse have been presented in the context of isolated stellar collapse ( e.g. @xcite ) and binary neutron star mergers ( e.g. @xcite ) with different degrees of realism .
the role of electromagnetic fields are typically examined within general relativistic _
ideal _ mhd and studies have been presented for single @xcite and binary star systems @xcite .
such simulations have illustrated important details of the extreme dynamical behavior induced by the system which could trigger tremendously energetic phenomena .
newtonian and general relativistic simulations have shed light on how a binary neutron star merger can amplify pulsar - strength magnetic fields by several orders of magnitude .
the resulting hypermassive star is generically unstable to black hole formation , which opens up the tantalizing possibility that the increasing rotation rate and magnetic field strength would drive an intense electromagnetic outflow . to study such a scenario ,
our goal here is to combine the grmhd approach with a suitable description of the magnetically - dominated region outside a collapsing star , by applying the force - free approximation .
our approach is related to that of ref .
@xcite , which matched a simplified , analytic , relativistic solution for the interior of a collapsing star ( dust - ball ) with a numerical solution of the coupled einstein - maxwell equations in its exterior .
boundary conditions at the star s surface are provided by the analytic , ideal mhd solution .
we go beyond this simplified scenario in three significant ways .
first , our stars are evolved through collapse consistently by evolving the ideal grmhd equations , and we can therefore consider in principle any kind of compact star . in particular , we have studied both rigidly rotating and non - rotating stars , whereas only the non - rotating case was considered in @xcite .
second , the magnetosphere is described within the force - free approximation , which , as argued in ref .
@xcite , is a much more realistic model that , for instance , has been instrumental in understanding pulsar spin down @xcite .
third and last , our matching of the exterior solution with the interior one is dynamical .
thus the force - free solution can adapt to time - dependent fields sourced by the star .
our approach thus allows us to examine many interesting scenarios and we apply it here to study the behavior of collapsing , magnetized , compact stars ( either rotating or not ) , the behavior of surrounding plasma ( as described within the force - free approximation ) and possible electromagnetic radiation induced by the system .
we compare our results with recent estimates in ref .
@xcite for the non - rotating case which predict that the collapse process is smooth and that the magnetic field remains anchored to the star as a black hole forms , leaving a final black hole with a split - monopole field configuration . our work , which follows the dynamics of the system , indicates that in both cases the stars radiate significant electromagnetic energy in which reconnection plays a crucial role , and that this radiation ceases shortly after the formation of a black hole . with force - free
, such a black hole loses its electromagnetic hair within a dynamical time scale .
we describe our hybrid approximation and the evolution equations in sec .
[ sec : approach ] , followed by a summary of the numerical techniques in sec .
[ sec : numerical_techniques ] .
our choice of initial data is described in sec .
[ sec : id ] , and several tests of the hybrid approach are presented in sec .
[ sec : testing ] . the new results for collapsing ,
magnetized stars are explained in detail in sec .
[ sec : physics ] , while the astrophysical consequences are described in secs .
[ sec : astrophysicsi ] and [ sec : magnetar ] .
we conclude in sec .
[ sec : concluding_remarks ] with some final comments .
in the presence of matter and electromagnetic fields , the einstein equations must be suitably coupled to both the maxwell and hydrodynamics equations .
this coupling is achieved by considering the stress energy tensor @xmath2 with contributions from matter and electromagnetic energy given respectively by @xmath3 u_a u_b + p g_{ab } \ , , \\
t_{ab}^{\rm em } & = & { f_a}^c f_{bc } - \frac{1}{4 } g_{ab } f^{cd } f_{cd } \ , .\end{aligned}\ ] ] a perfect fluid with pressure @xmath4 , energy density @xmath5 , internal energy @xmath6 , and four - velocity @xmath7 describes the matter state , and the faraday tensor @xmath8 describes the electromagnetic field . the fluid and electromagnetic components are directly coupled through ohm s law , which closes the system of equations by defining the electric current 4-vector @xmath9 as a function of the other fields .
a general relativistic expression can be obtained by considering a multifluid system of charged species @xcite , leading to a fully non - linear propagation equation for the spatial component of the current .
however , it usually suffices to consider a simplified version accounting for an algebraic relation between the current and the fields @xcite @xmath10 where @xmath11 is the conductivity of the fluid .
the well - known scalar ohm s law is recovered for @xmath12 .
the equation of motion for the fluid and electromagnetic field are obtained from _ the conservation laws _
@xmath13 which , together with the einstein equations @xmath14 , complete the system of equations governing the dynamics .
once the appropriate form of ohm s law and the conductivity have been specified , using for instance the algebraic relation of eq .
( [ ohm_law ] ) , the resulting equations typically involve vastly different scales , rendering the implementation of these equations quite costly from a computational point of view .
fortunately , for specific regimes certain useful approximations capture the _ most relevant _ physics while bypassing the most strenuous difficulties . for our current purposes , involving magnetized neutron stars ,
the following relevant approximations can be defined and employed in different regimes : * the _ ideal - mhd _ equations are obtained by requiring that the current remains finite in the limit of infinite conductivity , @xmath15 .
this condition also implies the vanishing of the electric field measured by an observer co - moving with the fluid ( @xmath16 ) .
this approximation is appropriate for the highly conducting matter expected in neutron stars .
however , the numerical evolution of the ideal mhd equations typically fails in low density regions where the inertia of the electromagnetic field is a few orders of magnitude larger than that of the fluid unless sufficient resolution is available .
such resolution requirements increase as the ratio of the electromagnetic to the fluid s inertia ( or magnetic to fluid s pressure ) grows .
consequently , the approach becomes costlier in regions with decreasing physical relevance with respect to bulk matter motion .
such a situation arises in particular in `` vacuum '' regions due to the standard practice of maintaining a density floor ( a so called _ atmosphere _ ) in regions of low density to exploit advanced numerical techniques for relativistic hydrodynamics .
the atmosphere s density is much smaller than that inside the star , so this approach does not affect the relevant matter physics .
however outside the star , the fluid inertia ( or pressure ) is typically much smaller than that of the electromagnetic field in magnetized cases and one generally encounters a large number of numerical difficulties .
different numerical strategies are often introduced to avoid them but such measures can limit one s ability to extract appropriate physics in these regions . *
the _ force - free _ maxwell equations are obtained by assuming the fluid s inertia is much smaller than that of the electromagnetic fields . as a result @xmath17 which in turn implies @xmath18 ( @xmath19 ) @xcite .
this assumption therefore allows one to ignore the explicit time - evolution of the fluid as its dynamics are implicitly prescribed by the charge and current distribution .
this approximation is well suited to the dynamics of the low density , magnetically dominated plasma surrounding a compact object , but it can not account for the physics in dense regions . * the _ vacuum _ maxwell equations are trivially recovered assuming no coupling with matter ( @xmath20 ) .
this is a natural approximation in vacuum regions , far away from the compact objects of interest .
as mentioned , computational costs are presently major obstacles to employ the general equations .
it is thus highly desirable to define a scheme able to model all the relevant regimes in highly dynamical systems simultaneously since none of these three approaches captures all the expected behavior .
in particular we have in mind interacting binary neutron stars , black hole - neutron star binaries as well as the collapse of a magnetized star considered here .
all three systems are believed to play an important role in understanding gamma ray bursts and other energetic events driven by compact stellar - mass objects . in what follows ,
we describe and apply a _ hybrid _ approach which , while approximate , can account for the dynamical interaction of both gravitating and electromagnetically driven fluids
. such an approach allows one to study the magnetosphere s behavior , in particular energetics and field topology providing important clues for understanding relevant systems . to do so , we take into account that the electromagnetic inertia of a region with high conductivity ( i.e. inside the star ) would be orders of magnitude larger than that of the plasma region .
thus , we can ignore the back - reaction of the electromagnetic field in the plasma region onto that inside the star ( i.e. , a `` passive '' magnetosphere ) .
we exploit this observation to define our approach , which employs both the ideal and force - free approximations suitably matched around the stellar surface .
the matching procedure is such that the star s electromagnetic field provides the boundary conditions for the surrounding region treated with the force - free approximation , but the behavior of the magnetospheric plasma on the high density stellar interior is ignored .
notice that this approach , in a sense , can be regarded as a natural extension of that adopted for studying pulsar - spin down by magnetosphere interactions @xcite to general relativistic dynamical contexts . in these works ,
only the behavior of the magnetospheric plasma is studied and the star s influence is accounted for by boundary conditions .
we here account for the possible stellar dynamics and consequent influence on the boundary conditions defined for the force - free approximation . in practical terms , we regard our system as described by two sets of electromagnetic fields : ( i ) the ideal mhd fields @xmath21 and ( ii ) the force - free fields @xmath22 , governed by their respective equations
. both sets of fields are defined with respect to observers orthogonal to the spacelike hypersurfaces employed to foliate the spacetime . thus identifying the appropriate fields for matching
is direct .
the ideal mhd equations are evolved over the entire computational domain , as is customary .
the matching region is determined by fluid density being some value above the vacuum region ( though alternatives based on other physical quantities are obviously possible ) .
in particular , the force - free fields are evolved only in regions where @xmath23 .
dynamic boundary conditions are applied to the force - free fields on the surface at which @xmath24 using the ideal mhd fields @xmath25 these conditions are applied in the spirit of penalty techniques @xcite in which the equations are modified with driving terms at boundary points in order to enforce the desired boundary condition . in particular , we extend the penalty technique across a number of grid points to effect both the boundary conditions and to `` turn - off '' the evolution of the force - free fields within the higher density regions .
notice that this extension is not mathematically rigorous , but its usage will be justified by examining the solution in different test applications and comparing with the expected behavior . in practice we define a smooth kernel @xmath26 defined as @xmath27 where typically we adopt @xmath28 and @xmath29 , being @xmath30 the value for the density of the atmosphere . in the collapsing cases studied ,
this value is rescaled in time by the ratio of maximum density to the initial maximum density ( see appendix ) .
the values found in eqs .
( [ efromb]-[eq : bbc ] ) along with the kernel eq .
( [ eq : kernel ] ) make their appearance in the equations determining @xmath31 ( together with `` constraint cleaning '' fields @xmath32 , introduced to ensure constraints are well behaved through the evolution @xcite ) which are @xmath33 the final term on the right - hand side of these equations is a penalty factor , which is introduced to impose interface conditions and ensure that a discrete energy norm is bounded .
details of this `` penalty technique '' are presented in @xcite and examples of applications in general relativity can be found in @xcite .
the above equations , together with the einstein and grmhd equations are implemented as described in @xcite to which we refer the reader for further details .
we adopt finite difference techniques on a regular cartesian grid to solve the system . to ensure sufficient resolution in an efficient manner we employ adaptive mesh refinement ( amr ) via the had computational infrastructure that provides distributed , berger - oliger style amr @xcite with full sub - cycling in time , together with an improved treatment of artificial boundaries @xcite .
the refinement regions are determined using truncation error estimation provided by a shadow hierarchy @xcite which adapts dynamically to ensure the estimated error is bounded within a pre - specified tolerance .
the spatial discretization of the geometry and force - free fields is performed using a fourth order accurate scheme satisfying the summation by parts rule , and high resolution shock capturing methods based on the hlle flux formulae with ppm reconstruction are used to discretize the fluid variables @xcite .
the time - evolution is performed through the method of lines using a third order accurate runge - kutta integration scheme , which helps to ensure stability of the numerical implementation @xcite .
we adopt a courant parameter of @xmath34 so that @xmath35 on each refinement level @xmath36 . on each level
, one therefore ensures that the courant - friedrichs - levy ( cfl ) condition dictated by the principal part of the equations is satisfied . to extract physical information
, we monitor several quantities : ( i ) the matter variables and spacetime behavior , ( ii ) the electromagnetic field configuration and fluxes , and ( iii ) the electromagnetic newman - penrose ( complex ) radiative scalar ( @xmath37 ) .
this scalar is computed by contracting the maxwell tensor with a suitably defined null tetrad @xmath38 and it accounts for the energy carried off by outgoing waves to infinity .
the luminosity of the electromagnetic waves is @xmath39 additionally we monitor the ratio of particular components of the maxwell tensor @xmath40 which , in the stationary , axisymmetric case , can be interpreted as the rotation frequency of the electromagnetic field @xcite .
initial data for the hybrid equations involve the intrinsic metric ( @xmath41 ) and extrinsic curvature ( @xmath42 ) on a given hypersurface , as well as the magnetized fluid configuration in terms of its primitive variables @xmath43 .
the initial data for the geometry and the fluid of rigidly rotating neutron stars are provided by the lorene package _ magstar _
@xcite , which adopts a polytropic equation of state @xmath44 with @xmath45 rescaled to @xmath46 .
because the fluid inertia of a neutron star is many orders of magnitude larger than its electromagnetic one , the magnetic field will have a negligible effect on both the geometry and the fluid structure , and so it can be specified freely .
unless noted otherwise , in our simulations we have chosen a dipolar structure for the initial magnetic field .
the electric fields are set by assuming the ideal mhd condition eq .
( [ efromb ] ) , with zero fluid velocity in the exterior region .
additionally , we require initial data for the force - free fields @xmath31 . inside the star
, they are defined to be exactly the same as their ideal mhd counterparts . outside the star , the magnetic field
is well defined by the dipolar solution , while the electric field is computed by assuming that the magnetosphere rotates rigidly with the star up to @xmath47 ( @xmath48 is the stellar radius ) , and imposing again the ideal mhd condition for the electric field .
this configuration provides consistent data for the problem ; however such data will not necessarily conform to the physical situation considered and so an unphysical early transient will be generated .
we first establish that the adopted procedure is indeed able to capture correctly the dynamics of relevant systems .
since the ideal mhd approximation is self - consistently evolved throughout the computational domain , our tests must address the behavior of the force - free fields .
to do so , we first examine the convergence of our implementation and then illustrate that the approach provides the expected behavior by comparing with certain recently studied cases . adopting a stable and non - rotating stellar solution with mass @xmath49 and equatorial radius @xmath50 , from which we remove all initial pressure , we examine convergence of the force - free fields as the star collapses .
this scheme contains `` standard '' sources of error , such as : ( i ) truncation error of our finite difference approximation of maxwell s equations , ( ii ) errors associated with our application of the force - free conditions , ( iii ) the various numerical errors ( truncation and constraint violations ) associated with our grmhd implementation as well as ( iv ) errors associated with our dynamic boundary condition that matches the force - free fields to the corresponding mhd fields inside the star .
a detailed analysis of these errors is delicate and involved ; however we illustrate that the fields obtained by this approach converge to a unique solution with increasing grid resolution .
we therefore evolve the collapse of a non - rotating star at three different resolutions with fixed mesh refinement ( fmr ) .
we subtract a high resolution run ( with coarse level grid of @xmath51 points ) and a medium resolution run ( of dimension @xmath52 ) and do similarly for the medium and low resolutions ( of dimension @xmath53 ) .
these two differences are plotted at @xmath54 ( i.e. , when the radius of the star starts to shrink ) in fig .
[ fig : nonrotatingconvergence ] which shows that the difference decreases with increasing resolution , consistent with convergence .
next we consider a stationary force - free solution representing a rotating neutron star with a monopole magnetic field @xcite , and compare to our results . to minimize effects due to oscillations of the star ( induced by perturbations to the star induced by the discretization ) , the geometry and mhd fields are kept to their initial values and are not evolved .
the initial data correspond to a rigidly rotating neutron star near the mass shedding limit , with a mass @xmath55 and an equatorial radius @xmath56 , rotating with a period of @xmath57 .
the magnetic field is given by @xmath58 , regularized conveniently near the origin .
as described earlier , the initial electric field satisfies the ideal mhd condition in a rigidly co - rotating magnetosphere which extends initially up to @xmath59 .
the evolution is performed in a cubic domain of length @xmath60 with only two fmr grids and the star is placed at the origin .
the maximum resolution is @xmath61 , so that roughly 30 grid points cover the star .
this resolution is sufficient for this test as both the geometry and matter variables are kept fixed and only the force - free equations are evolved until reaching a quasi - stationary configuration .
the behaviour of different fields across the surface of the star after the solution has settled are displayed in fig .
[ fig : monopole_bfield ] , which plots radial profiles of : the kernel function @xmath62 , the density @xmath5 , and the non - trivial components of the magnetic field .
as illustrated in the figure , the radial magnetic field only changes outside the star , where there also appears a toroidal component indicating the rotation of the magnetic fields in the magnetosphere .
overall the electromagnetic fields are seen to relax rather quickly to a state approaching the expected one .
another measure of the obtained solution is given by the ( normalized ) rotational frequency of the magnetic field @xmath63 in fig .
[ fig : monopole_omega ] .
as time progresses , this value approaches @xmath64 as expected in a radially smooth way .
a particularly challenging test of our approach is the aligned rotator solution ( see @xcite ) .
the aligned rotator is a numerical solution of the force - free equations outside a rotating surface representing a star that demonstrates closed field lines within the light cylinder ( lc ) .
this problem has been studied by a number of authors @xcite working in flat space with a computational domain consisting only of the stellar exterior . in these
works the star s influence is accounted for through suitable boundary conditions derived from the expected electromagnetic field at its surface .
these efforts were specifically motivated to obtain the solution in the magnetosphere of isolated pulsars and , in particular , to understand a possible spin - down mechanism that works even when the star s dipolar field is aligned with its angular momentum .
as already mentioned , our approach differs in that we solve for the stellar interior , we work within curved space and our computational grid does not conform to the star s geometry .
because of these differences and also because our initial data for the force - free fields is only nonzero in the immediate neighborhood of the star , achieving the expected late - stage stationary solution dynamically constitutes a demanding test .
we note that the time and length scales required for such a test with respect to a realistic star are too computationally demanding if we are to capture the main physical aspects of the solution ( i.e. field topology , location of the light - cylinder , development of a current sheet , etc ) . instead of a realistic star , we adopt a rapidly rotating star so that the light cylinder is brought closer to the star where there is enough resolution to resolve it and render the dynamical time scales shorter and easier to follow numerically .
to achieve such rapid rotation , we resort to an unstable star but prevent the instability from disrupting the star by artificially freezing both fluid and geometry .
consequently we only evolve the force - free equations and compare the obtained configuration with the expected solution .
we adopt a rotating star with a mass @xmath65 , equatorial / polar radius of @xmath66 and rotational period @xmath67 .
the light cylinder for this star is located at @xmath68 .
the electromagnetic field is initially set similar to that in the previous monopole test , but with a poloidal magnetic field given by a dipole outside the star .
the evolution is performed in a cubic domain of @xmath69 with four fmr levels with resolutions @xmath70 .
the fmr hierarchy consists of centered cubes with side lengths @xmath71 , so that there are roughly 70 points across the star in the equatorial plane .
we evolve this star until the solution settles to a quasi - stationary solution .
[ fig : pulsar_bfield ] illustrates the magnetic field topology at three representative times . in the last frame
, we show the configuration to which the field settles , illustrating features predicted by known solutions for the aligned rotator . in particular , one can observe a region of closed field lines bounded at the expected radius , denoting the light cylinder .
having tested our hybrid approximation to em field evolution , we now focus on the collapse of a magnetized star .
the star may or may not be rotating , the rotating case being of greater astrophysical relevance .
in particular , we are interested in studying how the magnetosphere responds , and the distribution of escaping electromagnetic radiation . here
it is essential to capture both the dynamics of the collapsing star and the surrounding magnetoplasma , since the two are tightly coupled .
it is worth recalling that the introduction of rotation introduces a fundamental difference in the radiative output of stationary ( non - collapsing ) stars as calculated in the electrovacuum and force - free limits : the spindown torque of an aligned rotator vanishes in electrovacuum but does not in the force - free case .
furthermore , this torque is strong and its magnitude varies only modestly with inclination angle @xcite .
previous studies of the magnetosphere of a collapsing star are fairly limited .
some pioneering work in @xcite focused on the electrovacuum behavior of a non - rotating star . while this study included the effects of strong gravity , by neglecting plasma or rotation it was not able to capture the winding of the magnetic field during the collapse .
the analytic study by @xcite estimated the em output during the collapse by applying the formula for the spindown luminosity of an equilibrium rotator , @xmath72 , where @xmath48 is the stellar radius , @xmath73 the surface magnetic field , and @xmath74 the spin frequency .
however , when the collapse time is shorter than the rotation period ( as must be the case if the star is to avoid a rotational hang - up ) , the outer magnetosphere near the light cylinder is not able to follow the change in the surface magnetic field , and this equilibrium formula does not apply . instead
, twisting of the closed magnetic field lines is expected at a radius @xmath75 with the time for an alfvn wave to propagate from the star out to @xmath76 along a dipole field line . ] , where @xmath77 is the collapse time .
the corresponding toroidal magnetic field is @xmath78 , where @xmath79 in a dipole geometry .
the power injected into the magnetosphere is @xmath80 which is larger than the equilibrium spindown power by a factor @xmath81 . to determine the dependence of @xmath82 on @xmath48 , equation ( [ eq : lwind ] )
must be combined with the conservation of magnetic flux , @xmath83 constant ; and the appropriate scaling between @xmath74 and @xmath48 ( @xmath84 for self - similar collapse ) .
the collapse does not , in general , follow a simple power - law relation between @xmath85 and @xmath48 , but in the special case of pressureless collapse from a large radius ( @xmath86 ) one obtains @xmath87 . it should be re - emphasized that equation ( [ eq : lwind ] ) represents energy stored in the magnetosphere , and so the maximum amount of energy that can only escape to infinity after the collapse is completed . during the last , relativistic stages of the collapse , additional physical effects arise .
spacetime curvature has a mixed effect on magnetic field winding in the magnetosphere .
on the one hand , the rotation frequency is reduced with respect to the newtonian value ; on the other hand , torsional alfvn waves in the force - free magnetosphere slow down significantly as the horizon approaches the surface of the star .
the net result , as we show , is that the magnetic field becomes strongly wound up if the star is rotating close to breakup before the collapse , and develops @xmath88 after the collapse . to clarify the influence of plasma and rotation , and to make useful comparisons ,
we have made a parallel set of runs in the electrovacuum approximation .
these are obtained straightforwardly by setting @xmath89 and not enforcing the conditions @xmath90 in eqs .
( [ maxwellext_3 + 1_eq1a]-[maxwellext_3 + 1_eq1d ] ) . the collapse of a magnetized , but non - rotating , neutron star was studied in @xcite , where a magnetic field frozen into the stellar surface was matched to an exterior , vacuum solution of maxwell s equations .
this calculation followed the transition to a black hole , and the ringdown of the em field threading the horizon . in reality , the medium outside the star is an excellent electrical conductor .
as is argued in @xcite , a combination of large voltages and strong gravitational fields will trigger runaway pair creation near compact objects @xcite . in a strongly dynamic situation
, the number of particles generated can be enhanced even further , e.g. by a kolmogorov - like transfer of energy from large - scale waves to internal plasma heat @xcite .
the remnant of a binary merger or a rapidly rotating stellar core collapse is also a strong neutrino source and , therefore , its magnetosphere will be filled with a much denser baryonic plasma compared with pulsar magnetospheres @xcite . but in most cases , the plasma around compact objects is magnetically dominated and its dynamics is nearly force free .
+ + therefore , we expect significant differences in the time evolution between the electrovacuum exterior assumed by @xcite and a force - free magnetosphere . to anchor the external force - free ( or electrovacuum ) solution , we adopt a marginally unstable , non - rotating star with a mass @xmath91 and a radius @xmath92 .
the numerical domain extends up to @xmath93 and contains three centered fmr grids of sequentially half sizes ( and hence twice better resolved ) .
the highest resolution grid has @xmath94 .
the star collapses to a black hole in @xmath95 @xcite .
we set @xmath96 at the onset of an apparent horizon and display the magnetic field and the density at various times in fig .
[ fig : collapse_norot1 ] .
this force - free solution has several salient features .
there is an early transient in which the dipole magnetic field relaxes to a solution consistent with the physical configuration .
subsequently , as the star collapses , the magnetic field lines are gradually stretched along the equatorial plane .
after @xmath97 an apparent horizon appears in the interior of the star , which grows as it swallows all the remaining fluid in @xmath98 . as the outer layers of the star are accreted by the black hole , the stretched magnetic field lines near the equatorial plane reconnect and form closed loops that carry away electromagnetic energy and magnetic flux ( note the field loops in the last frame of fig .
[ fig : collapse_norot1 ] ) .
the em field evolution shows qualitative differences in the force - free and electrovacuum runs , especially after the black hole forms ( fig .
[ fig : collapse_norot2 ] ) . starting from a common dipole structure at the beginning of the collapse , the magnetic field in the force - free case becomes radially stretched near the magnetic equator , and maintains a consistent sign . in the electrovacuum evolution ,
the stellar magnetic field disconnects more readily from the exterior .
changes in the connectivity of the magnetic field follow the appearance of zones where @xmath99 , as is demonstrated in figs .
[ bsq_esq ] and [ fig : collapse_norot3 ] . for a more quantitative discussion
, we consider in fig .
[ fig : collapse_emfluxes ] the electric and magnetic fluxes , computed over a surface located at @xmath100 , as a function of time . as expected , both the total ( signed ) fluxes remain small throughout the simulation , indicating that essentially no spurious magnetic / electric charges are created during the collapse .
the unsigned magnetic flux decays exponentially in both the electrovacuum and force - free cases .
interestingly , the decay rate in the former case matches the @xmath101 quasi - normal mode for electromagnetic perturbations of a schwarzschild black hole with mass @xmath49 @xcite . the decay rate of the magnetic flux in the force - free run is roughly twice the electrovacuum result
. one could translate the measured e - folding time @xmath102 into an `` effective reconnection speed , '' @xmath103 by observing that reconnection takes place within @xmath104 ( with @xmath105 the horizon radius ) .
our results indicate @xmath106 .
this result may appear paradoxical at first sight , and deserves some comment .
electrovacuum magnetic fields effectively reconnect ( by converting to electric - dominated fields ) at the speed of light , as we discuss in more detail in sec .
[ s : quasi ] .
why then is the decay faster in the force - free case ?
the answer appears to reside in the self - inductance of the black hole .
as magnetic field lines reconnect through the equator , they generate a strong toroidal emf . in a vacuum , this emf sources a magnetic flux of the opposing sign .
when conducting matter is present , there is no oscillation in the sign of the flux threading the hole .
rather , reconnection is a monotonic process and , after the formation of an x - point , the magnetic field lines interior to the x - point fall through the horizon .
+ + + in order to estimate the efficiency with which em energy is radiated to infinity , we have scaled the time - integrated luminosity to the peak electromagnetic energy contained in the magnetosphere .
this peak energy is reached approximately at the formation of the apparent horizon . in the absence of rotation
, it is @xmath107 times the initial dipole energy , as determined by the conservation of magnetic flux .
the measured radiated energy is , in turn , a fraction @xmath108 of @xmath109 .
numerically , @xmath110 , and one finds @xmath111 for a ( initial ) polar field @xmath112 and radius @xmath113 .
hence @xmath114 the radiative efficiency is very small in the force - free case , @xmath115 ; the rest of the peak energy is swallowed by the black hole .
this is illustrated in fig .
[ fig : collapse_energyrad_nonrotating ] , which shows how @xmath108 grows with time in both the force - free and electrovacuum runs .
the radiated energy is @xmath116 , from eq .
( [ energy_peak_magnetosphere ] ) , expressed in terms of magnetar - strength magnetic fields .
most of this is radiated in a short interval @xmath117 ms surrounding the collapse , with an average luminosity @xmath118 erg s@xmath119 .
the radiative efficiency measured at @xmath100 is an order of magnitude higher in the electrovacuum case .
this indicates that a larger proportion of the electromagnetic energy falls into the black hole in the force - free run , instead of escaping to infinity .
an important check of this result is to measure the dissipation at current sheets in both the force - free and electrovacuum simulations . in both cases , we find that the integral of @xmath120 is a small fraction of the total em energy in the magnetosphere .
( the constraint @xmath121 is maintained in the force - free case by applying a small enhanced resistivity ; hence the reduction in electric field energy appears as @xmath120 dissipation . )
an important feature of this radiation in both cases is its predominantly dipolar structure in energy flux ( i.e. , @xmath122 ) . in the force - free case , energy is radiated in a rather continuous manner and it propagates outwards with a velocity @xmath123 , equivalent to a mild lorentz factor of @xmath124 . for the electrovacuum case
the energy is radiated mainly in two long bursts instead of the several , and with shorter periods , resulting in the force - free case . at later times , when the fluid has completely fallen into the black hole , the field lines that were dragged toward the horizon reconnect near the equatorial plane in a few sequential bursts , expelling most of the remaining magnetic flux .
this behavior appears to be associated with the formation of x - type singularities in the magnetic field . since the field is stretched radially and then reconnects near the horizon , the resulting electromagnetic pulse bunches up in the radial direction , which explains the structure seen in the last panel of fig .
[ fig : collapse_norot1 ] .
x - point reconnection appears to happen easily in force - free plasmas ; in ohmic plasmas it is associated with inhomogeneities in the electrical resistivity ( e.g see @xcite ) . in the absence of a detailed microphysical model for the resistivity , fine details such as these
should be treated with caution , and the calculation should be viewed as illustrative .
some important features of the vacuum evolution of the electromagnetic field around a collapsing star can be understood by neglecting the effects of spacetime curvature , and by considering a simplified trajectory for the surface of the star .
if the collapse starts at a finite time , then an initially potential magnetic field evolves into a hybrid structure that consists of an inner potential magnetic field that matches the surface boundary condition as determined by the conservation of magnetic flux ; and a transient electromagnetic wave that propagates into the original field structure . of especial interest
is the appearance of zones within this wave structure that are dominated by the electric field .
if the collapse continues for a long time ( the final stellar radius @xmath48 is small compared with the initial radius ) , then this zone where @xmath99 extends over a wide range of radius . since a realistic magnetosphere may contain enough free charges to limit the growth of @xmath125 , this provides a nice example of how the nearly force - free evolution of an electromagnetic field can lead to strong dissipation .
in the absence of rotation , a spherically symmetric collapse implies a radial fluid velocity inside the star .
as long as the stellar surface contracts with a speed @xmath126 , the magnetic field near it is approximately potential . for a pure multipole of order @xmath127 , @xmath128\ ] ]
( with @xmath73 the magnetic field at the star s surface ) . a toroidal electric field @xmath129 is present at the surface of the star , assuming its interior to be perfectly conducting .
the junction condition at the surface ensures the continuity of @xmath130 .
the surface magnetic field therefore increases in accordance with simple flux conservation , @xmath131 at a fixed radius @xmath132 , the magnetic field grows weaker : in the case of a simple dipole , the stellar magnetic moment scales as @xmath133 .
if the star were to reach infinite density at a finite time @xmath85 , then a strong toroidal electric field would develop at @xmath134 .
the inner potential zone would shrink along with the star as @xmath135 approaches @xmath85 .
the external electromagnetic field can then be obtained by rescaling the radius , @xmath136 , and transforming derivatives according to @xmath137x(\xi , t)$ ] .
it is simplest to solve for the vector potential @xmath138 , from which the poloidal magnetic field and toroidal electric field are derived .
the boundary condition at the surface of the star is @xmath139 = { b_s(t ) r_s^2(t)\over \ell } { dp_\ell\over d\theta } = { r_{s0 } a_{\phi 0}(\theta)\over\ell } { dp_\ell\over d\theta}.\ ] ] from eq .
( [ eq : fluxcon ] ) , @xmath140 is constant . substituting @xmath141 into the wave equation @xmath142 and adopting a collapse law @xmath143 ( here @xmath144 for pressureless collapse from a large radius )
, we find @xmath145\partial_\xi^2 g \cr & & \quad\quad\quad -\left(1+{1\over\alpha}\right)\xi\partial_\xi g - { c^2\over \dot r_s^2 } { \ell(\ell+1)\over \xi^2}g.\end{aligned}\ ] ] here @xmath146 is a dimensionless time coordinate ; hence @xmath147 .
the electromagnetic field is constructed from the solution to eq .
( [ eq : diffeq ] ) using @xmath148 one recovers the usual potential solution @xmath149\,(r / r_s)^{-(\ell+1)}$ ] where @xmath150 .
a self - similar solution is also available in the case of collapse at a uniform speed , @xmath151 , if the collapse starts at a very large initial radius
. then one can take @xmath152 and the electromagnetic field is a function only of @xmath153 . in this case
, the magnetic field can retain a dipolar form out to large distances @xmath154 from the star .
restricting to @xmath155 gives @xmath156 although the magnetic field is identical to that sourced by a stationary dipole of magnitude @xmath157 , the electric field energy dominates outside a distance @xmath158 from the star .
the similarity solution is accurate out to the larger distance @xmath159 , where @xmath160 is the stellar radius at the beginning of the collapse . to understand how this inner solution for the electromagnetic field matches onto the initial potential magnetic field that was present prior to the collapse , or to consider cases other than constant @xmath161
, one must calculate the full time - dependent solution to ( [ eq : diffeq ] ) .
we have done this by evolving @xmath162 , @xmath163 and @xmath164 using a centered discretization of eq .
( [ eq : diffeq ] ) and employing a small kreiss - oliger dissipation value @xmath165 ) to damp high - frequency noise in each of these variables .
the resulting electromagnetic field profile is plotted in fig .
[ 1d_em_profile ] , with distance normalized to the initial radius of the star .
one observes in fig .
[ bsq_esq ] the emergence of an extended zone with @xmath99 , as is expected from the similarity solution of eq .
( [ eq : similar ] ) .
the magnetic field is relatively stronger in the polar regions , since @xmath166 at small @xmath167 .
eventually the amplitude of the outgoing wave disturbance becomes large enough that @xmath168 changes sign .
after this happens , the magnetic field remains dipolar inside the radius @xmath169 , but disconnects from the external zone of undisturbed potential field .
this transition is illustrated in fig.[1d_em_field_lines ] using two snapshots corresponding to the two most extended field profiles in fig .
[ 1d_em_profile ] .
the first part of the collapse leads to a re - arrangement of the magnetic field outside the star , while electromagnetic energy flows inward : the poynting flux , @xmath170 is negative inside @xmath171 .
after the horizon forms and reaches the surface of the star , the compression of the magnetic field stops , and some of the trapped magnetic field can be radiated to infinity .
the angular distribution of this radiated energy reflects the symmetry of the initial field . in the case of a dipole
, the energy is carried away by closed magnetic loops and is mainly channeled through the magnetic equator .
we note that the appearance of zones with @xmath99 in the electrovacuum solution implies a fundamental difference with the alternative force - free solution . in this case , @xmath172 throughout the collapse , so consistent force - free evolution is obtained only by removing energy from the electric field . from an mhd perspective
, such a transition signals the appearance of nearly luminal plasma motions , where the inertia of even a small residue of entrained matter can become important .
this simple example shows that there can be profound differences in the macroscopic structure of the electromagnetic fields when even a small amount of conducting matter is present . in the complete absence of free charges , macroscopic zones where @xmath99 are the natural consequence of the time - evolution of an em field that , initially , is purely magnetic .
the quasi - normal mode ( qnm ) behavior observed shows two remarkable features : an oscillation in the sign of the magnetic flux threading each hemisphere , and the appearance of an equatorial zone with @xmath99 .
these two features can be related to each other using a simple planar analogy .
consider an initial magnetic field configuration @xmath173 ( @xmath174 ) for @xmath175 ( @xmath176 ) .
this is familiar from studies of conducting fluids , where in the case of uniform resistivity one finds a slow flow of matter to a thin current sheet at @xmath177 .
the sheet thickness is limited by the flow of matter along the sheet to large @xmath178 ( sweet - parker reconnection ) .
but now we are interested in the case where the conducting matter is absent , and the electromagnetic field evolves according to the vacuum wave equation .
one finds , instead , a growing zone of pure _ electric _ field @xmath179 for @xmath180 , which maintains a uniform sign across the initial magnetic null surface .
a pure magnetic field is converted to a pure electric field . in the case of em fields localized around a black hole ,
this conversion of magnetic to electric fields occurs at the magnetic equator , where the poloidal field lines merge together . the toroidal electric field that is created sources a poloidal magnetic flux through the horizon of the opposing sign to the pre - existing flux .
in this way , a continuing interconversion of magnetic and electric fields can be maintained .
+ + + + the collapse of a rotating , magnetized star produces an interesting generalization of the relativistic wind problem for a stationary star ( e.g @xcite ) . as in the case of spherical collapse ,
qualitatively new effects are introduced after the formation of a horizon .
we have chosen an unstable , rotating model star with a mass @xmath55 and equatorial radius @xmath181 .
the star rotates with a period @xmath182 , so that the light cylinder is initially located at @xmath183 .
the numerical domain and resolution are identical to those employed in the non - rotating case . to remove unphysical transients
we evolve the force - free equations for a couple of periods with both geometry and matter fixed , and afterwards all equations are evolved .
this approach ensures the force - free fields relax to a configuration consistent with the physical scenario considered .
the expected field configuration emerges during this startup phase , with a closed , corotating magnetosphere extending out to the light cylinder ( lc ) .
the goldreich - julian ( g - j ) current structure is present , with an outflow along the polar field lines balanced by a return flow through an equatorial current sheet .
the different stages of the collapse are represented in fig .
[ fig : collapse_rot1 ] , while that the angular velocity of the star is displayed in fig .
[ fig : collapse_omegastar ] . as the star contracts , and its rotational frequency increases , the instantaneous lc approaches the star .
differential rotation develops in the magnetosphere , due to the lack of causal contact between the star and the lc , and the magnetic field is wound in the toroidal direction .
furthermore , the deepening of the gravitational potential forces significant changes in the magnetic field profile , by pulling the field lines more tightly toward the star .
the poloidal magnetic field strengthens due to flux freezing in the star , just as in the non - rotating case , but now most of the em energy is in the toroidal component . near the poles , the field lines twist around , generating a cone - like structure . in general ,
the magnetic field preserves a stretched dipolar topology for a longer time than in the non - rotating case , up to the point that all the fluid is swallowed by the black hole . as the black hole forms inside the star , and the fluid falls inward ,
many of the features in the em field evolution are preserved from the non - rotating runs .
snapshots of the field profile , charge density and poynting flux , taken at various times close to horizon formation , are displayed in figs .
[ fig : collapse_q_rotating ] and [ fig : collapse_bphi2_rotating ] . just as in the non - rotating case
, the topology of the magnetic field lines changes dramatically : the y - point structure in the magnetic field disappears from the equatorial regions , and a current sheet extends inward to the horizon .
this current sheet is subject to spasmodic episodes of reconnection , the details of which may depend on the prescription for the electric resistivity and its variation with radius .
in less than a millisecond , the magnetic flux threading the horizon has almost completely vanished .
this is qualitatively similar to the rapid evolution of a black hole interacting with a dipolar force - free configuration , as seen by @xcite .
in contrast with what is argued in @xcite , magnetic reconnection prevents the emergence of a relatively long - lived split - monopole configuration . in sec .
[ s : resist ] we discuss whether force - free or ideal mhd calculations more accurately describe the global structure of the current sheet , and the implications of the slower reconnection that is seen in the mhd calculation of @xcite . + + + + the evolution of the magnetic flux during the collapse is displayed in fig .
[ fig : collapse_emfluxes_rotating ] , again computed on a surface located at @xmath100 .
the total ( signed ) flux is small through most of the simulation , and only at late times does it becomes comparable to the absolute value of the magnetic flux . in the electrovacuum case ,
the final decay is governed by the main quasi - normal modes of the rotating black hole .
one observes faster decay in the force - free run , just as as in the non - rotating case , possibly due to the greater self - inductance of the vacuum black hole .
the radiated electromagnetic energy is displayed in fig .
[ fig : collapse_energyrad_rotating ] , rescaled again with respect to the energy peak in the magnetosphere .
the simulations indicate that around @xmath184 of the energy stored in the magnetosphere is radiated during the collapse in the force - free case , similarly to the electrovacuum case .
this is in clear contrast to the non - rotating case , where a much larger energy is radiated in the electrovacuum solution than in the force - free one .
we note that although the energy of the magnetosphere is similar in the non - rotating and rotating cases ( we find @xmath185 ) , the inclusion of rotation leads to a @xmath186fold enhancement in the em energy radiated by a collapsing force - free magnetosphere : @xmath187 .
hence eq .
( [ energy_peak_magnetosphere ] ) gives @xmath188 , resulting in a strong average luminosity of @xmath189 erg s@xmath119 during the collapse .
the distribution of the radiated energy is essentially quadrupolar , since the newman - penrose scalar @xmath37 has angular dependence mainly determined by an @xmath190 mode .
this radiation propagates outwards with a velocity @xmath191 , which is very similar to what was obtained in the non - rotating case . as displayed in fig .
[ fig : collapse_fluxangles_rotating ] , most of the energy is radiated more efficiently near the angle @xmath192 with a peak intensity shortly after the formation of the black hole ( for comparison purposes , fig .
[ fig : collapse_fluxangles_rotating_ev ] illustrates the behavior in the electrovacuum case ) .
subsequently this emitted radiation fades away rapidly as the magnetic flux is radiated away .
we now consider the implications of our simulation results for high - energy transient phenomena .
the very luminous gamma - ray bursts ( grbs ) are generally believed to result from the formation of a stellar - mass black hole by sudden gravitational collapse , either in the core of a massive star @xcite , or after the merger of a neutron star with another compact object @xcite .
a rapidly rotating magnetar is an interesting alternative @xcite .
our focus here is on ultraluminous em outflows : the same general mechanism is strongly favored in other contexts such as pulsar synchrotron nebulae @xcite and agn jets @xcite .
our calculations focus on the _ transition _ between a magnetar and a black hole , as a key part of the engine that drives a grb .
the implication is that a build - up of magnetic flux in the surface layers of a rapidly rotating neutron star ( formed , e.g. , in a binary merger ) can driven an electromagnetic outflow during the collapse of the star , independently of any blandford - znajek process operating afterward .
as we explain here , this has some advantages over an em wind operating before the collapse , in that this brief em transient is likely to be significantly cleaner .
we also compare the net em output with that expected from a post - collapse jet , in the situation where the surface layers of the neutron star , and not the surrounding disk , are the dominant zone for magnetic flux generated by a dynamo process .
our calculations have revealed strongly dissipative processes at work in the em field surrounding the collapsing star .
these include the reconnection of field lines near the magnetic equator , large - amplitude oscillations in the field , and ( in vacuum calculations ) the formation of extended zones where @xmath193 , which in the presence of conducting matter imply relativistic motions of the entrained particles transverse to the magnetic field .
the ejection of loops of magnetic field is observed in our non - rotating simulations .
the dynamical evolution of these loops has interesting implications for the gamma - ray flares from gravitationally stable magnetars ( see sec . [
sec : magnetar ] ) .
although our calculations take into account the presence of plasma implicitly by enforcing @xmath172 , adding the associated force - free current and enforcing @xmath194 , we expect that strong heating will occur in practice .
where the magnetic field reaches @xmath0 g , the implied temperature is above @xmath195 mev .
such a high density of electrons and positrons is created that macroscopic zones of non - vanishing @xmath125 as are required by most pulsar discharge models ( e.g. @xcite ) can not be maintained .
effective heating can occur by other channels : for example , long - wavelength gradients in the magnetic field are converted efficiently to internal energy if the magnetic field becomes turbulent , so that a wide spectrum of wave motions is formed that extends down to very small scales @xcite
. the relativistic outflow that is emitted by a rapidly collapsing star should therefore be quite hot .
large - scale magnetic fields and a relativistic photon - electron - positron plasma will both contribute substantially to the energy flux .
magnetar outbursts provide a fairly direct example of this phenomenon .
the star that collapses to form a black hole passes through three distinct phases : a pre - collapse phase during which it emits a magnetized wind if it rotates ; the dynamic collapse phase ; and , if an orbiting disk is present as it is following a neutron star merger
an accretion phase that is accompanied by a relativistic jet . since
our current simulations focus on the intermediate step , it is worth examining its relative contribution to the total em output of the star .
a large enough em output @xmath196 is measured in our rotating force - free simulation to power some short grbs , especially if some account is made for beaming . here
the requirement is that the surface magnetic field is strong enough to hold off any accretion flow . at an accretion rate @xmath197
, this implies a polar magnetic field stronger than @xmath198^{1/2 } \cr & \sim & 7\times 10^{15}\,\left({\dot m\over m_\odot~{\rm s}^{-1}}\right)^{1/2}\quad{\rm g}.\end{aligned}\ ] ] here @xmath199 is the circular speed , approximated as keplerian . in this section
we take a stellar mass @xmath200 and a radius @xmath201 , as appropriate to a hot and rapidly rotating neutron star . the dipole field energy before the collapse is @xmath202 we estimate @xmath203 , given that the energy released is about 0.2 times the peak magnetic energy , which in turn is @xmath204 times @xmath205 . from eq .
( [ eq : edipole0 ] ) we obtain @xmath206 the precise numerical value depends non - linearly on the initial specific angular momentum @xmath207 through the factor @xmath208 : faster initial spins imply stronger winding of the magnetic field during the collapse .
the magnetic field of an isolated neutron star acts as a couple between its reservoir of rotational energy and a dissipative outflow . even in the case of a ( gravitationally stable ) magnetar ,
the magnetic energy begins to dominate the rotational energy only at an advanced age , as the star spins down .
therefore the output of the pre - collapse phase could be very large compared with the release of em energy during the collapse .
comparing the spindown energy radiated over a time @xmath209 with the external magnetic energy ( [ eq : dele ] ) gives @xmath210 here we have substituted the spindown power of an aligned , force - free rotator @xcite , @xmath211 where @xmath212 .
an isolated , rotating star would radiate energy equal to @xmath205 in a few milliseconds .. in the context of binary neutron star mergers , one requires a magnetosphere to emerge from the very strong shear layer near the surface of the merger remnant ( see section [ s : dynamo ] for further discussion of how this could happen ) .
if the neutron star is formed hot , a lengthy pre - collapse spindown phase would cause significant difficulties with the application to short grbs , because the wind generated during the pre - collapse phase is heavily loaded with nucleons and @xmath213 particles that are driven outward by charged - current absorption of electron - type neutrinos near the neutrinosphere @xcite .
the connection between short grbs and neutron star mergers would also be disfavored if this pre - collapse outflow lasted longer than @xmath214 ms , given a characteristic short grb lifetime of 0.03 - 0.3 seconds @xcite .
on the other hand , the survival of the merger remnant for @xmath215 ms would have the advantage of allowing stronger amplification of the magnetic field before the remnant collapses .
the possibility of longer - lived merger remnants for some configurations has been raised by recent simulations that employ a realistic , finite - temperature eos @xcite .
although two merging neutron stars are expected initially to have magnetospheres , the torus formed by the tidal disruption of the lighter star has a pressure vastly exceeding that of a typical pulsar dipole .
the torus would , therefore , suppress a magnetosphere around the newly formed merger remnant .
the neutrino - driven wind that emerges from the polar regions of the remnant will comb out the magnetic field , but this field need not initially be coherent across the star or dynamically important . by the same token ,
if a torus were entirely absent , then the magnetic field threading the star would dissipate rapidly after the black hole forms , and the post - collapse phase would contribute negligibly to the output of the star .
it is useful to express the em power in terms of the `` open '' magnetic flux that connects the surface of the star to the outflowing wind . in the case of an isolated star ,
this is the flux extending beyond the light cylinder , @xmath216 where @xmath217 is the dipolar magnetic flux threading the star .
one can then re - write eq .
( [ eq : lsd_dipole ] ) as @xmath218 after the star forms a distinct magnetosphere , the accretion torus can continue to influence the wind power by modifying the open magnetic flux .
let us suppose that the magnetic pressure dominates the torus ram pressure out to an equatorial distance @xmath219 .
approximating the magnetosphere by a dipole , a fraction @xmath220 of the stellar flux is trapped by the torus . in this situation , it is still possible for @xmath221 to exceed eq .
( [ eq : phiop ] ) , because the torus can extend inside the stellar light cylinder , @xmath222 during our simulations of the collapse of an isolated star , we observe that the magnetic field lines are strongly twisted , so that most of the closed magnetic flux opens out . then @xmath223 where @xmath224 is the angular velocity of the horizon .
a torus also plays an important role after the black hole forms by trapping a certain fraction of the magnetic flux that threads the star .
if the magnetic field is strong enough to hold off the torus from the star before the collapse , then it will have a similar effect after the collapse .
we therefore calculate the blandford - znajek power emerging from the horizon by assuming a uniform flux density out to some radius @xmath225 . from eq .
( 8.65 ) of @xcite , one gets , @xmath226 \left({\omega_h r_h\over c}\right)^2\,r_h^2\langle b\rangle^2 c\cr & \alt & { 1\over 30\pi^2 c } \left({\phi_h\omega_h}\right)^2 , \end{aligned}\ ] ] where @xmath227 is the flux density threading the region interior to the torus , and @xmath228 . the difference in the normalizations of eqs .
( [ eq : lsd2 ] ) and ( [ eq : lbz ] ) largely reflects the fact that the torque on the black hole is maximized when the magnetic field has an angular velocity @xmath229 .
we can now relate eq .
( [ eq : lbz ] ) to the em power generated before the collapse .
the spin angular momentum is approximately conserved during the collapse , @xmath230 , and we also set @xmath231 .
then @xmath232 where @xmath233 .
the angular frequency of the black hole is @xmath234 , and is related to @xmath235 pre - collapse by @xmath236 where @xmath237 is the schwarzschild radius .
one expects @xmath238 after the collapse if most of the magnetic flux threading the star is open before the collapse , and the pre - collapse magnetosphere is limited in size .
the proportion of the trapped flux threading the hole is @xmath239 we can now show that the em power is suppressed immediately following the collapse . substituting equations ( [ eq : omratio ] ) and ( [ eq : phiratio ] ) into ( [ eq : lbz ] ) gives @xmath240 with a coefficient @xmath241 .
additional power will flow along the magnetic field lines that thread the ergosphere , but this portion of the black hole magnetosphere will mix with the accretion flow and may be less strongly magnetized .
the interesting conclusion here is that the rapid twisting up of the magnetic field during the collapse can generate a larger em output than a fairly extended jet emission after the collapse .
it is worth summarizing the three main sources of this result : i ) before the horizon forms , the em power is proportional to @xmath242 rather than @xmath243 ; ii ) the magnetic flux remains pinned in the star for a few rotation periods during the collapse ( before the onset of the black hole ) , and then springs out to fill a larger volume ; and iii ) the relation between rotation frequency and angular momentum is enhanced by a factor @xmath244 prior to the collapse ; in other words , relativistic gravity has the effect of softening the growth of the rotation frequency as the star collapses
. a reduction in the trapped flux ( due to outward diffusion of the magnetic field into the torus ) would , in this situation , initially _ increase _ the blandford - znajek power flowing from the horizon .
the pressure of the trapped field approximately balances the ram pressure of the accretion flow at some point outside the horizon .
therefore a reduction in the trapped flux allows the flow to reach closer to the black hole , and attain higher pressures , before being interrupted .
although we are considering the flux originating in a dynamo process before the collapse ( sec .
[ s : dynamo ] ) , continued flux generation in the torus by the magnetorotational instability @xcite could play a role in modulating the jet power .
let us consider the em energy that would be radiated following a binary ns merger , if the remnant survives long enough to form a magnetosphere ( sec .
[ s : dynamo ] ) .
taking @xmath245 , and normalizing to an accretion rate @xmath246 s@xmath119 through eq .
( [ eq : edipole0 ] ) , gives @xmath247 erg .
we expect that the gain factor @xmath248 depends non - linearly on the pre - collapse rotation rate , since the winding of the magnetospheric field results from a competition between differential rotation and torsional wave motion .
in addition , @xmath196 depends indirectly on the torus mass and pressure through the strength of the magnetic field that is required to hold off the torus material from the neutron star surface .
the larger accretion rate in a collapsar environment implies a larger transient energy .
it is possible to make a direct comparison with the blandford - znajek jet that follows the collapse , if the magnetic flux that threads the black hole is left behind by the collapsing magnetar . combining equations ( [ eq : lsd ] ) and ( [ eq : lbz_2 ] ) , and taking pre - collapse rotation period and radius @xmath249 msec and @xmath250 km
, one finds that an energy @xmath251 would be radiated by a bz jet over @xmath252 ms .
equivalently , a blandford - znajek jet from a @xmath253 black hole with @xmath254 would generate a power ( [ eq : lbz ] ) @xmath255 erg / s . a recent binary merger simulation with grmhd and dynamical gravity @xcite presented a polar magnetic field @xmath256 g developing from a much weaker seed field in the torus .
the equation of state used gave a fast collapse to a bh ( within @xmath257 ms ) , and therefore precluded the surface shear dynamo that we have conjectured .
( note that this polar field could be affected by numerical resistivity , e.g. @xcite ) . comparing the output @xmath196 with eq .
( [ eq : dele ] ) , one sees that the pre - collapse star would release a comparable energy within @xmath258 ms after forming a magnetosphere . however , it is well known ( e.g. @xcite ) that such an outflow would bear a much higher density of nucleons than a blandford - znajek jet from a black hole , due to the absorption of electron - type neutrinos and anti - neutrinos .
we expect that this nucleon loading would be strongly suppressed in the dynamical collapse phase , due to i ) the relatively short duration of the emission ; ii ) the strongly wound field geometry ( @xmath259 ) ; and iii ) redshifting effects .
combining these effects suggests a significant suppression @xmath260 in the nucleon loading , so that the mass ejected would essentially be that present in the magnetosphere before the collapse . the immediate aftermath of a binary neutron star merger is distinguished from disk accretion onto a black hole , in that the velocity shear is strongest where the orbiting material makes a transition from centrifugal to hydrostatic support .
another feature which distinguishes the merger remnant from ordinary accreting neutron stars ( x - ray pulsars ) is that it does not initially have an ordered magnetosphere .
long after the first stage of the merger is complete , the velocity shear provides a tremendous source of free energy for amplifying a magnetic field .
the power dissipated in the material settling onto the neutron star surface is @xmath261\cr & \sim & 3\times 10^{52}\,\left({\dot m\over 0.1m_\odot~{\rm s}^{-1}}\right ) f_{\rm shear}\quad{\rm erg~s^{-1}},{\mbox { } \nonumber \\ \mbox { } } \end{aligned}\ ] ] where @xmath262 ^ 2 $ ] .
note that the surface shear becomes more radially concentrated with time : as differential rotation is erased in the interior of the merger remnant , the surface shear is maintained by continuing accretion .
the inner part of the shear layer develops positive @xmath263 and the magnetorotational instability is extinguished .
a strong _ feedback mechanism _ is present which allows rapid magnetic field growth , but causes this growth to saturate once the star is able to form a magnetosphere that holds off the accretion flow .
when the magnetosphere is present , the accreting material follows the magnetic field and reaches the star at the same angular velocity .
shearing of the magnetic field in the outer layers of the star is therefore turned off .
this effect is clearly demonstrated in the 3d accretion simulations of @xcite .
only a tiny fraction of the accretion energy @xmath264 must be converted to a poloidal magnetic field to hold off the accretion flow : substituting ( [ eq : bmag ] ) into the dipole energy ( [ eq : edipole0 ] ) gives @xmath265 over a duration @xmath209 .
low - mass x - ray binaries provide a nice example of systems where this feedback appears to operate .
the millisecond radio pulsars that are descended from them have magnetic fields that are just strong enough ( @xmath266 g ) to hold off an eddington - level accretion flow onto a neutron star but not much stronger .
the absence of persistent pulsations in the majority of lmxbs then requires that the magnetic field be aligned with the angular momentum of the accretion flow .
now let us consider how the magnetic field evolves in the surface shear layer .
the field present initially in the merging stars is rapidly amplified by a kelvin - helmholtz instability @xcite , or the magnetorotational instability @xcite .
rapid growth of the magnetic field on large scales is sensitive to the speed of magnetic reconnection in the fluid , and requires three - dimensional motions .
the two - dimensional wrapping of a magnetic field by kelvin - helmholtz vortices does not generate net flux ; and the initial growth length of the mri is very small compared with the scale height of the torus .
mri growth is fastest on a scale @xmath267 in a torus of mass @xmath268 @xcite .
the hydrostatic structure of the inner shear layer makes it easier for the magnetic field to be pinned and retained than it would be in the surrounding torus .
the magnetic field threading the shear layer is wound up , and since @xmath269 , the mean toroidal flux density grows at least in a linear manner .
if a hot , rotating , and massive neutron star ( @xmath270 ) can survive collapse for more than @xmath258 ms , as is suggested by recent simulations of @xcite , then this toroidal field becomes quite strong .
even if the seed poloidal field is as weak as @xmath271 g within the range of pulsar fields then the toroidal field reaches @xmath272\delta t\ ,
b_{p,0}\cr & = & 1\times 10^{16}~\left({\delta t\over 100~{\rm ms}}\right)\left({\omega_c(r_{\rm ns})-\omega_{\rm ns}\over 10 ^ 4~{\rm s^{-1}}}\right){\mbox { } \nonumber \\ \mbox { } } & & \quad\quad\quad\quad\times \left({b_{p\,0}\over 10^{13}~{\rm g}}\right)\quad{\rm g}.\end{aligned}\ ] ] here @xmath273 is the angular frequency at the surface of the torus .
exponential magnetic field growth becomes possible when the wound - up field can rise buoyantly through the shear layer .
the large dissipated power ( [ eq : shearpower ] ) will generate a strongly positive entropy gradient in the inner part of the shear layer .
where the material is convectively stable , only magnetic fields stronger than @xmath274 g can directly overcome the pressure of the overlying material .
but even in this case , the intense flux of electron - type neutrinos allows a magnetic field stronger than @xmath275 g to rise buoyantly on the alfvn timescale , by erasing gradients in entropy and electron fraction that impede buoyancy @xcite .
the buoyancy time is @xmath276 across a pressure scale height @xmath277 .
thus an exponential feedback loop appears quite likely in this situation .
one observes that the large - scale dipole magnetic field that emerges may be sensitive to the seed magnetic field , in the sense that a _ minimal _ seed field is required for the linearly wound field to reach the buoyancy threshold .
nonetheless , it is also clear that magnetar - strength magnetic fields do not require magnetar - strength seed fields in the presence of persistent surface shear . the strength of the magnetosphere that eventually emerges results from a competition between the finite rate of amplification and the diminishing shear stress at the surface of the star .
at an age of @xmath278 , a poloidal magnetic field stronger than @xmath279 g ( @xmath280 g ) will begin to apply a strong , negative feedback on the surface shear .
once the torus mass drops to the point that accretion is mediated mainly by magnetic stresses , the accretion rate is @xmath281 s@xmath119 . here
@xmath282 and @xmath283 are the initial torus mass and diffusion time in this viscous phase , and @xmath213 is the viscosity coefficient .
the accretion rate drops as the torus material spreads outward , as @xmath284 , as long as the torus remains geometrically thick and conserves its angular momentum ( e.g. @xcite ) .
we now consider the observational imprint of a magnetar if it forms by the accretion of a thin layer of strongly sheared material , but is ( initially ) gravitationally stable .
we focus on the rate of energy loss and the spin history of the star .
there is a substantial reduction in the energy release by spindown , compared with a star that initially rotated as a solid body , due to the internal rearrangement of angular momentum .
approximate solid - body rotation is attained on the poloidal alfvn timescale @xmath285 , which is generally much shorter than the magnetic - dipole spindown time @xmath286 most of the shear energy is then dissipated _
internally_. a delayed collapse is possible if the magnetar only slightly exceeds the maximum mass for a non - rotating , zero - temperature neutron star .
the loss of rotational support by a magnetic wind would trigger a collapse on the rotational braking time .
quite generally , the accretion of a thin layer of strongly sheared material provides an attractive mechanism for creating magnetars , and so the results of this section should have a broader application to the galactic magnetar population , even if these stars are well below the maximum mass .
several possible channels are available : the rotation of the accreting material could be generated by an instability of the supernova shock @xcite ; or the neutron star could be exposed to material with very strong vorticity in a merger event , e.g. when it merges with a companion co white dwarf , or with the core of an evolving be star .
the spindown power that is released by the magnetar is then reduced significantly compared with a star which rotates uniformly with the same angular velocity as the surface material . the final spin period resulting from the accretion of a layer of mass @xmath287 and rotation period @xmath288 onto a neutron star of total mass @xmath289 is @xmath290^{-1 } p_{\rm shear}\cr & = & 10.5\,j_{\rm shear}^{-1}\left({m_{\rm ns}\over 2\,m_\odot}\right)\quad{\rm ms},\end{aligned}\ ] ] where @xmath291 . as the star spins down , it deposits a rotational energy @xmath292 in the surrounding shock wave . for a magnetar of polar field @xmath112 ,
the corresponding spindown luminosity is fairly modest , @xmath293 and the star enters the extended spindown phase after a time @xmath294 in this situation , only a very strong internal magnetic field ( @xmath295 g ) would induce sufficient triaxiality in the star that the gravity wave torque competed with the external electromagnetic torque @xcite .
escaping loops of magnetic field are observed in our non - rotating simulations .
these loops are formed by magnetic reconnection , which means their size and rate of formation are sensitive to the treatment of resistivity , and to the disregard of stresses associated with bulk plasma flow by the adoption of the force - free equations .
an outgoing plasmoid ( containing plasma and a closed magnetic field ) has a well - defined center of mass : we find that its bulk motion is measurably smaller than the speed of light , @xmath296 and @xmath297 , at a distance @xmath298 times the pre - collapse neutron star radius @xmath299 .
this is significantly less relativistic than the bulk motion expected for a gas of freely expanding particles released by the collapsed star ( @xmath300 ) . after addressing each of these issues ,
we make contact with the giant gamma - ray flares of the galactic magnetars .
these appear to involve the ejection of an energetic plasmoid ( @xmath301 erg ) , but due to the shearing of the magnetic field lines rather than the collapse of the star @xcite . when magnetic field lines of an opposing sense are stretched out and forced into contact , the rate at which they reconnect is sensitive to the microscopic model of resistivity . in an ohmic plasma with a uniform resistivity ,
a long current sheet forms and reconnection is very slow ; fast reconnection with an x - point geometry depends on a local maximum in the resistivity @xcite . in some contexts , such as the solar corona ,
the microscopic explanation for this behavior may be provided by the hall terms in the conductivity @xcite .
these are relatively less important if the plasma is loaded with @xmath302 pairs , as would be expected in the magnetosphere of a collapsing magnetar . in a fluid
, small - scale hydromagnetic turbulence appears to greatly accelerate the reconnection rate @xcite ; but whether such a process can operate in a low-@xmath303 plasma of astrophysical dimensions is not yet determined . in the present work we find evidence for relatively fast reconnection of magnetic field lines , @xmath304 , as calculated in the force - free approximation ( sec .
[ subsec : nonrotating ] ) .
this results from a change in topology of the field lines ( the formation of an x - point ) .
it is not due to strongly enhanced dissipation in an extended equatorial current sheet : @xmath120 dissipation is measured to be small in both the rotating and non - rotating cases . such a concern arises in force - free calculations of pulsar magnetospheres , where the magnetic field is less dynamic and is anchored in the star . there
the absence of fluid pressure support in the current sheet leads to a rapid collapse of the magnetic field toward the sheet , unless explicitly compensated @xcite . in the present case , after the black hole forms the inflow of magnetic flux toward the equator can continue in a more dynamic manner into the star , or out to the computational boundary .
a recent treatment of reconnection in the magnetosphere of a stationary black hole @xcite illustrates a slower field decay when employing an ideal mhd treatment , as compared to a force - free approach .
( resistivity in the mhd case arises through the numerical approximation . )
such a slower decay is also observed at late stages after the formation of the black hole in our simulations when comparing force - free and ideal mhd fields and our implementation does not cope with as large magnetizations as that in @xcite . ] ; though by this time the field strengths are orders of magnitude below their peak values .
a comparison of force - free and mhd reconnection calculations which explores more general resistivity models , and their influence on the reconnection geometry , remains to be developed .
reconnection is usually studied in a context where the plasma flow speed along the magnetic field is a small fraction of the alfvn speed : for example , in a steady mhd wind , the conservation of angular momentum implies a slow drift of particles along the spiral magnetic field outside the alfvn critical point .
however , in some contexts , such as magnetar flares , the flow speed can approach the speed of light . similarly , in our collapse simulations we see large - amplitude motions on the magnetic field loops threading the neutron star , which suggest strong plasma heating .
the backreaction of the outflowing plasma on the reconnection of field lines is not taken into account .
even when the magnetic field has a tendency to reconnect through an x - point , reconnection will be _ delayed _ until the kinetic pressure @xmath305 of the outflowing pair - photon plasma ( with thermal energy density @xmath306 ) drops below the poynting flux that would flow toward the current sheet in the absence of plasma flow .
one requires @xmath307 where the coefficient on the right - hand - side is appropriate for fast x - point reconnection .
the magnetic field lines are stretched outward by the plasma flow beyond an alfvn radius @xcite @xmath308 given @xmath309 .
note that , at the peak of the outflow , the pressure of the pair - photon fluid is comparable to the pressure of the stretched magnetic field lines at @xmath310 , and outside this radius grows as @xmath311 with respect to the split - monopole field pressure .
for example , in a magnetar giant flare , the @xmath312 s width of the main gamma - ray pulse is comparable to the time for magnetic and elastic stresses to rearrange the stellar interior ; but it is @xmath214 times longer than the flow time out to the alfvn radius ( [ eq : ralf ] ) , and therefore much longer than the timescale for x - point reconnection at a speed @xmath313 . a magnetically - dominated plasma that is ejected from a collapsing magnetar ( or a nearby soft gamma repeater ) can be a much stronger source of synchrotron emission than a shocked plasma of comparable energy density .
existing calculations of radio afterglows of short grbs ( @xcite and references therein ) , as well as calculations of the radio afterglow of sgr giant flares @xcite , focus on synchrotron emission by a population of non - thermal electrons that are accelerated at a shock that leads the outflow . for the bulk of grbs , one infers efficiencies of conversion @xmath314 of bulk kinetic energy to non - thermal electrons and to magnetic fields downstream of the forward shock .
these moderate values of @xmath315 and @xmath316 are consistent with the broadband tails of radio x - ray emission that follow the brief , bright gamma - ray phase .
the synchrotron emission from a magnetically - dominated plasmoid will be proportionately much brighter , by up to a factor @xmath258 , for two reasons .
first , and most obviously , @xmath316 now approaches unity .
second , if the magnetic energy density dominates the thermal energy density , then a cascade process ( involving the creation of high - wavenumber alfvn modes ) preferentially heats the electrons @xcite .
if the plasma is very relativistic , damping is mainly due to charge - starvation of the waves , at wavenumbers where the amplitude of the fluctuating current density begins to exceed @xmath317 @xcite .
( a different damping mechanism operates at higher ion densities : the waves are landau - damped on the parallel motion of the electrons . ) in practice , the relative amplitudes of the bulk synchrotron emission and the shock emission will depend on the degree of disorder in the plasmoid magnetic field .
but for magnetar flares , indirect evidence that bulk synchrotron emission dominates comes from a rapid initial drop in radio flux that is consistent with the sudden compression of the plasmoid , followed by rapid adiabatic cooling @xcite .
magnetar outbursts involve more limited releases of energy that leave the original star intact .
they are triggered when the footpoints of a @xmath275 g magnetic field are strongly sheared by an elastic instability of a neutron star crust , thereby generating a hot plasma and an intense burst of gamma - rays @xcite .
the first , extremely bright , stage of a giant magnetar flare lasts only @xmath318 and bears a considerable resemblance to a ` classical ' gamma - ray burst , demonstrating high temperatures ( @xmath319 kev ) and a significant non - thermal component to the spectrum @xcite .
( this phase saturates almost all x - ray detectors , but was well - resolved by the geotail experiment @xcite in the 27 august 1998 and 27 december 2004 flares . )
the duration and luminosity are consistent with the internal rearrangement of the magnetic field in a neutron star , with a strength @xmath320 g based on considerations of magnetic field transport and global flare energetics @xcite , several times stronger than the standard dipole expression for the spindown - derived magnetic field .
the giant flares appear to involve the ejection of a plasmoid .
the combination of fast variability and extreme luminosity ( up to @xmath321 ergs s@xmath119 ) implies , through the usual arguments of gamma - ray opacity @xcite , that the emitting plasma has expanded to a much larger volume than that of the neutron star .
this expanding plasma is an excellent electrical conductor and must carry some of the stellar magnetic field with it .
this expected property of magnetar giant flares helps to explain two apparently contradictory phenomena .
a straightforward argument based on the theory of thermal fireballs ( e.g. @xcite ) shows that the expanding plasma is moving with a high lorentz factor at the radius ( @xmath322 cm ) where it becomes transparent to the gamma - rays @xcite .
the observation of variability on a timescale @xmath323 in the gamma - ray flux implies strong constraints on the baryon rest energy flux .
the advected electrons and ions must become transparent to the gamma - rays close enough to the magnetar that the differential light - travel time @xmath324 across the outflowing plasma is shorter than @xmath325 . given a total outflow luminosity @xmath326 , this implies that the ion rest energy contributes no more than a fraction @xmath327 of the energy flux .
on the other hand , radio monitoring detected transient emission in the weeks following both flares @xcite .
the emission following the 2004 flare was especially bright , as befitting the much greater energy of the burst , and could be tracked on the sky for more than a year .
the expansion of the radio source implies a transverse velocity @xmath328 @xcite .
after this , clear evidence is seen for a break in the radio light curve consistent with a transition from uniform expansion to a sedov - like phase .
we infer that the measured proper motion is probably the free expansion velocity , as corrected for relativistic aberration . in principle
it is possible for the measured transverse speed to be less than the speed of light , if the intrinsic motion @xmath329 is nearly luminal but the motion is directed away from the observer ( at some angle @xmath330 with respect to the line of sight to the magnetar ) : @xmath331 .
the observed relative brightness of both the radio and gamma - ray emission argues against this : one measures @xmath332 in 2004 vs. @xmath333 in 1998 , and a radio flux @xmath334 mjy vs. 0.3 @xmath335jy at @xmath336 ghz 1 week after the flare @xcite .
( a factor @xmath337 underestimate of the @xmath338 kpc distance to sgr 1806 - 20 appears unlikely given that the source position is in the galactic plane . ) the energy reservoir that powers the early , subluminal stage of the radio afterglow must be composed of something other than electrons and positrons emitted by the star , which would mainly have annihilated during the very brief fireball phase . a reconnected magnetic field is the most plausible delayed carrier of energy , especially given the strong limitations on the baryon flux during the @xmath339 gamma - ray pulse .
a closed loop of magnetic field carries a finite inertia , and so its center - of - mass frame will not accelerate with distance from the source as does that of a collimated particle beam .
the outflowing pair - photon fluid must overcome the tension of the magnetic field lines that are anchored in the star , and so comparable energy can be put into the stretched field , which is pulled into a split - monopole configuration .
the energy of the stretched field is concentrated close to the star , @xmath340 , although not as strongly concentrated as in a static dipole ( @xmath341 ) . in a static plasma ,
the time for reconnection at the alfvn radius @xmath342 is @xmath343 . comparing this expression with the measured e - folding time of @xmath344 for the tails of the giant flare pulses
@xcite suggests that reconnection is gated by the decrease in pair - photon pressure .
we conclude that , in a magnetar giant flare , most of the released magnetic energy tends to _ follow _ the pair - photon pulse , with the delay between the two components being due to continued shearing of the external magnetic as the interior of the magnetar adjusts on the internal alfvn time of @xmath345 .
understanding the global behavior of strongly gravitating , dynamical systems , containing dense matter coupled to ultra - strong magnetic fields , is of key importance for a thorough understanding of possible signals from them . in this work ,
we have presented a new approach to this end by combining the ideal mhd and force - free approximations in a suitable manner within general relativity .
we expect that this hybrid scheme represents real progress towards greater realism .
the stellar interior utilizes ideal mhd to faithfully model the neutron star without the disadvantages incumbent in the low density exterior .
likewise , the exterior solution uses the force - free approach and so captures the dynamics of the tenuous plasma .
the entire domain is described by a fully nonlinear and fully dynamic general relativity solution necessary for strong - field gravity . indeed , a key aspect of this approach is its generality .
it does not require a prescribed stationary stellar boundary , and can be applied , for example , to dynamical systems such as collapsing stars and non - vacuum compact binaries .
we have exploited this approach to study stellar collapse in both rotating and non - rotating cases .
our studies reveal a rich phenomenology in the magnetosphere as the collapse proceeds .
in particular , magnetic reconnection plays an important role by inducing strong electromagnetic emission as well as the infall of electromagnetic energy into the black hole , which in a short time loses all its ` hair ' .
when the star starts off rapidly rotating , the energy of the magnetospheric plasma grows significantly during the collapse .
as the star rotates faster , its magnetic field lines do not have time to re - adjust to the increased rotation , and are strongly wound up out to the initial light cylinder .
it is worth emphasizing that this conversion of dynamical energy into electromagnetic energy does not depend on resistive effects .
two issues of principle have arisen in performing these calculations .
first , we have shown conclusively that the force - free approximation to the evolution of a dilute , relativistic plasma inevitably leads to singularities .
these singularities are avoided in our calculations by continuously pruning the electric field . in the simulations that we have run
, this procedure appears to cause limited energy dissipation , but its necessity should be kept in mind .
the second issue of principle regards the dependence of the reconnection geometry on the resistivity model .
fluid pressure is responsible for slowing down the rate of reconnection unless an x - point geometry is able to form . by neglecting fluid pressure
, the force - free approximation clearly facilitates the formation of x - points . in spite of this
, we observe very limited numerical dissipation in current sheets .
although the plasma that is represented by our model magnetospheres is , in reality , strongly collisional , it should be kept in mind that fast x - point reconnection still occurs in collisional plasmas in the presence of hydromagnetic turbulence as has been demonstrated so far in weakly magnetized plasmas @xcite .
some further exploration of the reconnection geometry is possible in resistive mhd calculations of stellar collapse by varying the spatial dependence of the resistivity .
we have also discussed how the electromagnetic outbursts from collapsing magnetars may have interesting observational effects . in particular ,
our results are relevant to binary neutron star merger scenarios , in which a hypermassive neutron star forms and collapses to a black hole . collapse to a black hole can happen either promptly or after many dynamical times , depending on the masses involved and the equation of state describing the stars ( see , for instance , @xcite ) . although kelvin - helmholtz and magnetorotational instabilities will create ultrastrong magnetic fields ( @xmath346 g ) , we have suggested that the global field ( in particular , the amount of magnetic flux threading the merger remnant and eventual black hole ) could depend strongly on the lifetime of the remnant .
the formation of a distinct magnetosphere would require tapping only @xmath347 of the energy dissipated in shear layer at the remnant surface .
as such a star collapsed , the magnetosphere would qualitatively follow the behavior outlined here .
the electromagnetic output of the collapse could compete with the later emission from a blandford - znajek jet emanating from the black hole horizon , and source a powerful electromagnetic counterpart to the gravity wave signal ( e.g. @xcite ) .
in addition , the magnetic field dynamics that is revealed in our simulations has a number of interesting implications for gamma - ray bursts and magnetar flares , as discussed in secs .
[ sec : astrophysicsi ] and [ sec : magnetar ] . beyond the work analyzed here , our approach is readily applicable to other relevant systems and will be applied , in particular , to study binary neutron star systems @xcite , and black hole - neutron star systems ( see e.g. @xcite ) .
it is important to stress , however , that our approach is not free of ambiguities , in particular with respect to how and where the matching between the force - free and ideal mhd regions is implemented .
to the extent possible , we have tested the robustness of our results versus known solutions , which give us confidence in this approach . * _ acknowledgments : _ * it is a pleasure to thank p. goldreich for insights and discussions during the course of this work .
we also thank e. berti , p. cerda - duran , j. mckinney , r. oshaughnessy , e. ramirez - ruiz , e. quataert , a. spitkovsky and d. tsang as well as our long time collaborators m. anderson , e. hirschmann , p. motl , d. neilsen and o. reula , for discussions and comments .
this work was supported by the nsf ( phy-0969827 to long island university ) and nserc through discovery grants ( to ct and ll ) .
research at perimeter institute is supported through industry canada and by the province of ontario through the ministry of research & innovation .
computations were performed at teragrid and scinet .
details of the transition from the ideal mhd regime to the force - free regime merit particular attention , especially in our rotating collapse solutions .
since we match two different formulations of maxwell s equations coupled to conducting matter , it is important to understand how the choice of matching layer affects the exterior force - free solution .
of course , an unambiguous test can only be provided by a complete resistive mhd solution that can handle the strongly magnetized regions outside the star and , in particular , can follow the large changes in density and rotation that are encountered during the collapse . while work on this direction is in progress
, the simpler approach presented here allows us to obtain a first solution to the relativistic magnetosphere in this strongly dynamic situation .
the positioning of the matching zone is constrained by competing considerations . on the one hand , if it sits too close to the surface , then the interior mhd solution that sources the exterior force - free solution will be unrealistic , due to the density floor that is applied in the mhd atmosphere .
one thus might want to place the transition layer well within the stellar surface .
however , such a deep layer might underestimate the magnetic field strength at the base of the magnetosphere , and imply force - free behavior of the magnetic field where that approximation is not justified in particular , we have found that placing the transition zone at too low a density implies an unrealistically large toroidal magnetic field at the base of the force - free zone . recall that the region exterior to the star , with its tenuous plasma , will have a relatively large magnetization , and should be forced to co - rotate with the interior of the star . unless prohibitively high resolution is employed
, the atmosphere of the mhd solution has a large enough inertia that this condition can be violated .
instead , the magnetic field experiences a non - negligible ( and non - physical ) differential rotation .
this effect is easily exacerbated in a collapsing solution . within our current approach ,
we address these issues in two ways , by ( i ) adopting an initial configuration of the force - free fields and ideal mhd fields respecting such co - rotation up to the light cylinder and with a suitable differential rotation outside that radius ; and ( ii ) choosing a transition layer at an appropriate distance inside the stellar surface .
regarding the initial data , since the rotation induces a toroidal magnetic field we adopt initial data that has a toroidal magnetic defined as @xmath348 and also @xmath349 throughout domain .
we also choose the mhd atmosphere to co - rotate with the star for @xmath350 , and not to rotate otherwise .
both the location of the stellar surface and the density in the transition layer are dynamical .
it is important that the choice of transition density @xmath351 in eq .
( [ eq : kernel ] ) respects the increasing density of the collapsing star . in order to maintain a consistent depth of the transition layer ,
its position is adjusted as the star collapses by scaling the matching density @xmath351 in proportion to the peak density within the star : @xmath352 $ ] .
the transition layer is displayed at the beginning and near the end of the collapse in fig .
[ fig : collapse_rhoft ] . notice that with this conservative approach we are underestimating possible rotational effects .
the angular frequency @xmath63 of magnetic field lines anchored near the rotation axis of the star is illustrated in fig .
[ fig : collapse_omegaf_alongline ] . from the initial condition ( in which constancy of @xmath63 is enforced ) ,
a negative gradient in @xmath63 develops in the transition layer during the earliest stages of the collapse .
this negative gradient has then disappeared by the time that the stellar angular velocity has increased by @xmath353 . above the initial value @xmath354
( see fig .
[ fig : collapse_omegastar ] ) .
from then on , the qualitative behavior obtained is consistent with the expected one : during the collapse , the star transfers angular momentum to the magnetic field lines so that @xmath355 near its surface , propagating along the magnetic field lines with a speed @xmath356 .
the sharp negative gradient appearing in the last stages of the collapse reflects the strong radial gradient in @xmath357 that appears near the rotation axis , as well as the onset of strong general relativistic effects . as the event horizon arises ( prior to the formation of the apparent horizon at @xmath358 ) , it disconnects the interior from the exterior solution and causes @xmath63 to decrease , tending to the value @xmath229 expected by the blandford - znajek solution of a spinning black hole . | we study the gravitational collapse of a magnetized neutron star using a novel numerical approach able to capture both the dynamics of the star and the behavior of the surrounding plasma . in this approach ,
a fully general relativistic magnetohydrodynamics implementation models the collapse of the star and provides appropriate boundary conditions to a force - free model which describes the stellar exterior .
we validate this strategy by comparing with known results for the rotating monopole and aligned rotator solutions and then apply it to study both rotating and non - rotating stellar collapse scenarios , and contrast the behavior with what is obtained when employing the electrovacuum approximation outside the star .
the non - rotating electrovacuum collapse is shown to agree qualitatively with a newtonian model of the electromagnetic field outside a collapsing star .
we illustrate and discuss a fundamental difference between the force - free and electrovacuum solutions , involving the appearance of large zones of electric - dominated field in the vacuum case .
this provides a clear demonstration of how dissipative singularities appear generically in the non - linear time - evolution of force - free fluids . in both the rotating and non - rotating cases ,
our simulations indicate that the collapse induces a strong electromagnetic transient , which leaves behind an uncharged , unmagnetized kerr black hole . in the case of sub - millisecond rotation ,
the magnetic field experiences strong winding and the transient carries much more energy .
this result has important implications for models of gamma - ray bursts .
even when the neutron star is surrounded by an accretion torus ( as in binary merger and collapsar scenarios ) , a magnetosphere may emerge through a dynamo process operating in a surface shear layer .
when this rapidly rotating magnetar collapses to a black hole , the electromagnetic energy released can compete with the later output in a blandford - znajek jet .
much less electromagnetic energy is released by a massive magnetar that is ( initially ) gravitationally stable : its rotational energy is dissipated mainly by internal torques . a distinct plasmoid structure is seen in our non - rotating simulations , which will generate a radio transient with subluminal expansion , and greater synchrotron efficiency than is expected in shock models . closely related phenomena appear to be at work in the giant flares of galactic magnetars . |
in the phenomenon of multiband superconductivity ( mbsc ) , which was first treated theoretically in 1959 using bcs theory,@xcite distinct energy gaps open up on different sheets of the fermi surface in the superconducting state .
one of the well - studied multiband superconductors is the binary intermetallic compound mgb@xmath6,@xcite which has two bands and two superconducting gaps .
based on a quasiclassical theory , microscopic calculations of the electronic structure in the vortex state have been made in a two - band superconducting model.@xcite these suggest that at low applied fields the dominant contribution to the change of the total density of states ( dos ) comes from the small - gap band , which gives rise to spatially extended quasiparticles ( qps ) . at high fields these loosely bound states become delocalized , with the vortex core size determined by the more localized states associated with the larger gap .
the presence of delocalized qps modifies the dos , transport properties associated with the quasiparticles , and the spatial field distribution around a vortex.@xcite recently strong evidence for mbsc has been found in the filled skutterudite pros@xmath1sb@xmath2 from thermal transport measurements in the vortex state.@xcite pros@xmath1sb@xmath2 has attracted much attention since its superconductivity was discovered in 2002.@xcite it is the only known pr - based heavy - fermion superconductor ( @xmath3 k ) and exhibits a number of extraordinary properties.@xcite the pr@xmath5 ground state is a nonmagnetic @xmath7 singlet , which is separated from a @xmath8 first excited state ( tetrahedral notation@xcite ) by only @xmath910 k.@xcite a novel ordered phase appears at high fields and low temperatures .
there is evidence that time - reversal symmetry is broken in the superconducting state.@xcite radiofrequency ( rf ) inductive measurements of the magnetic penetration depth @xmath10 in the meissner state,@xcite thermal conductivity measurements in a rotated magnetic field,@xcite and flux - line lattice distortion@xcite all suggest nodes in the superconducting gap .
a double superconducting transition has been observed in specific heat measurements.@xcite however , recent specific heat and heat transport measurements@xcite on a highly homogeneous single crystal show only one transition peak in the specific heat and a fully - gapped fermi surface .
the latter result is corroborated by angle - dependent specific heat measurements on single crystals.@xcite previous transverse - field muon spin rotation ( tf-@xmath0sr ) measurements of @xmath10 in the vortex state of pros@xmath1sb@xmath2 found evidence for a bcs - like activated dependence at low temperature,@xcite suggesting the absence of gap nodes .
thus there are a number of open questions concerning the superconducting order parameter in pros@xmath1sb@xmath2 .
the present article reports a detailed tf-@xmath0sr study of pros@xmath1sb@xmath2 .
tf-@xmath0sr experiments@xcite have proved invaluable in characterizing both the superconducting and normal states of superconducting materials;@xcite in particular , a tf-@xmath0sr study@xcite of superconducting nbse@xmath11 clearly revealed effects of mbsc on the vortex - state field distribution in this compound .
we compare tf-@xmath0sr and rf inductive penetration depth measurements in superconducting pros@xmath1sb@xmath2 , and discuss the previously - reported@xcite discrepancy between these measurements in terms of extreme mbsc in pros@xmath1sb@xmath2 .
preliminary reports of parts of this work have been published.@xcite we have also carried out tf-@xmath0sr measurements of the knight shift in the normal state of pros@xmath1sb@xmath2 , which suggest that the perturbing effect of the positive - muon ( @xmath12 ) charge on the neighboring pr@xmath5 crystalline electric fields ( cef ) is negligibly small .
this indicates that the muon charge is unlikely to be the source of the discrepancy between the tf-@xmath0sr and rf results . a model calculation for the perturbed cef energy levels , based on the approach of kaplan , schenck , and co - workers,@xcite
is described .
et al._@xcite have reported a tf-@xmath0sr study of pros@xmath1sb@xmath2 , with emphasis on the knight shift in the superconducting state .
the remainder of this introduction contains three brief pedagogical sections : a description of the elements of the tf-@xmath0sr technique used in this study ( sec .
[ sect : tfmusr ] ) , a review of the important features of the @xmath12 knight shift ( sec .
[ sect : knight - intro ] ) , and an introduction to the cef model calculation ( sec .
[ sect : cef ] ) . after a description of the experimental procedure ( sec .
[ sect : exp ] ) , in sec .
[ sect : results ] we describe our experimental results in pros@xmath1sb@xmath2 , which include the temperature dependence of the tf-@xmath0sr relaxation rate and penetration depth , magnetic susceptibility data , and normal - state @xmath12 knight shift measurements .
the implications of these results for the nature of both the superconducting and the normal state are discussed in sec .
[ sect : discussion ] .
we summarize our results in sec . [
sect : conc ] . in the tf-@xmath0sr
technique@xcite spin - polarized positive muons ( @xmath12 ) are implanted in the sample and precess in a local magnetic field applied perpendicular to the initial @xmath12 polarization .
this precession is detected using the asymmetry of the @xmath12 beta decay ( the decay positron is emitted preferentially in the direction of the @xmath12 spin ) .
the distribution of @xmath12 precession frequencies directly reflects the distribution of magnetic fields in the sample .
thus tf-@xmath0sr can be used to measure the magnetic field distribution of the vortex lattice in a type - ii superconductor and the local magnetic susceptibility in the normal state . in the vortex lattice
each vortex possesses a normal - state - like core with radius of order of the superconducting coherence length @xmath13 , surrounded by a shielding supercurrent .
these supercurrents give rise to an inhomogeneous magnetic field that is periodic in the vortex lattice .
an analytical ginzburg - landau ( gl ) model for the spatial field profile @xmath14 ( ref . ) yields @xmath15 where the @xmath16 are the reciprocal lattice vectors of the vortex unit cell , @xmath17 is a modified bessel function , @xmath18 is the reduced field , and @xmath19,\ ] ] with @xmath13 the gl coherence length .
this result is valid in the limit @xmath20 , where @xmath21 is the smallest nonzero reciprocal lattice vector .
tf-@xmath0sr is a sensitive probe of this field distribution.@xcite the functional form of the @xmath12 spin relaxation , which depends on the field distribution , is fit to the functional form @xmath22 , where the frequency @xmath23 and phase @xmath24 describe the average @xmath12 precession and the relaxation function @xmath25 describes the loss of phase coherence due to the distribution of precession frequencies .
the relaxation rate associated with @xmath25 is a measure of the rms width @xmath26 of this distribution .
the expression@xcite @xmath27 where @xmath28 is the flux quantum , then gives a rough estimate of @xmath10 for a triangular vortex lattice in the london limit ( @xmath29 ) .
more accurate relations are available,@xcite but in the present case lead to negligible corrections to eq .
( [ eq : distri ] ) . to estimate the field distribution width without fitting a theoretical model , a gaussian distribution of local fields
is often assumed .
then the time dependence of the @xmath12 spin polarization is proportional to @xmath30 , where @xmath31 is the @xmath12 relaxation rate ; @xmath32 is the @xmath12 gyromagnetic ratio . it has been pointed out@xcite that this procedure is approximate at best , because the field distribution is not expected to be gaussian . in a more microscopic approach , a lineshape analysis program has been written to fit the gl model to tf-@xmath0sr data.@xcite the gl model is preferred over the london model because it self - consistently accounts for the vortex cores . in this fitting program
the fit function is the fourier transform of the magnetic - field distribution function @xmath33 $ ] , which can be obtained from eq .
( [ eq : distribution ] ) .
these fits utilize the entire form of the field distribution , not just its second moment .
they yield an effective ` @xmath0sr ' penetration depth @xmath34 , which becomes the true penetration depth only after extrapolation to @xmath35.@xcite results from such fits in pros@xmath1sb@xmath2 ( ref . )
are described below in sec .
[ sect : tf ] . in tf-@xmath0sr the total field at the @xmath12 site is given by the sum of the applied field @xmath36 , the internal field induced by the applied field , and the demagnetization and lorentz fields.@xcite the relative @xmath12 frequency shift @xmath37 where @xmath38 is the @xmath12 frequency in free space " ( i.e. , no condensed matter effects ) , must be corrected for the contribution of the demagnetization and lorentz fields @xmath39 to obtain the @xmath12 knight shift@xcite @xmath40 which contains the relevant information about the local magnetic susceptibility . in a paramagnetic metal
, @xmath41 originates from hyperfine fields produced by the field - induced polarization of conduction electrons and localized electronic moments .
the contribution from the conduction electrons is temperature independent and is usually very small , of the order of 100 ppm.@xcite the local moments , in the present work the pr@xmath5 ions , contribute to @xmath41 via two coupling mechanisms : ( 1 ) the dipolar interaction between the local moments and the @xmath12 , which may be described as a dipolar field at the @xmath12 interstitial site , and ( 2 ) the indirect rkky interaction,@xcite in which an additional spin polarization of the conduction electrons due to the local moments produces a hyperfine contact field at the interstitial @xmath12 .
both contributions are proportional to the local - moment susceptibility : @xmath42 in the principal - axis coordinate system of the dipolar tensor @xmath43 , where the @xmath44 are the diagonal elements of @xmath43 and @xmath45 is the contact hyperfine coupling constant , assumed isotropic . for a cubic lattice ( @xmath46 ) , only the hyperfine contact field at the interstitial @xmath12 site contributes to the average shift @xmath47 , since the contribution from the dipole - dipole interaction vanishes @xmath48 .
the local susceptibility is sensitive to change in the pr@xmath5 cef , and the @xmath12 charge may induce such a change . if so , the modified local susceptibility will be reflected in a breakdown of the proportionality of the @xmath12 knight shift to the measured bulk susceptibility @xmath49 since in this case @xmath50 .
this effect has been studied in detail by kaplan , schenck , and co - workers@xcite in pr - based compounds with singlet pr@xmath5 cef ground states . as noted above and discussed in sec .
[ sect : cef ] , our results in pros@xmath1sb@xmath2 show little if any such effect .
the crystal structure of pros@xmath1sb@xmath2 belongs to the @xmath51 space group , with pr@xmath5 ions at the points of a bcc unit cell . as an initial approximation
we assume the cef hamiltonian @xmath52 for the pr@xmath5 ions has @xmath53 ( cubic ) point group symmetry .
then @xmath54 where @xmath55 , @xmath56 , @xmath57 , and @xmath58 are the stevens operators for a given angular momentum quantum number @xmath59 , and the @xmath60s are parameters usually determined from experiment.@xcite @xmath52 splits the pr@xmath5 @xmath61 hunds - rule multiplet into a @xmath62 singlet , a @xmath63 doublet , and two triplets ( @xmath64 and @xmath65 ) .
pr@xmath5 ions in pros@xmath1sb@xmath2 actually have @xmath66 ( tetrahedral ) point group symmetry , however,@xcite for which an additional sixth - order term appears in the cef hamiltonian [ eq . ( [ eq : hami1 ] ) ] .
this term mixes the @xmath64 and @xmath65 triplet wave functions with each other@xcite but has a relatively small effect on the cef energies.@xcite if it is predominant , a @xmath64 or @xmath65 triplet will be the ground state , inconsistent with the singlet ground state in nonmagnetic pros@xmath1sb@xmath2 .
however , if this term is small , the change in the physical properties in zero magnetic field can be approximated by changing the parameters in the cubic cef hamiltonian [ eq . ( [ eq : hami1 ] ) ] slightly .
thus we use the cubic cef hamiltonian for simplicity . in the presence of an external magnetic field , the zeeman interaction mixes and
splits the cef energy levels .
the magnetic susceptibility is given by@xcite @xmath67\exp\left(-e_{n}^{(0)}/kt\right)}{\sum_{n}\exp\left(-e_{n}^{(0)}/kt\right)},\ ] ] where the @xmath68 are the unperturbed cubic cef levels , @xmath69 with @xmath24 the cef wave functions and @xmath70 the land g - factor , and @xmath71 in the molecular - field approximation , the measured magnetic susceptibility @xmath72 is given by @xmath73 , where @xmath74 is the molecular field parameter that describes exchange interactions between pr@xmath5 ions.@xcite equation ( [ eq : sus ] ) shows that the observed susceptibility is directly related to the local cef energy levels . as noted above , if we take into account possible muon - induced change of the cef , we must consider a modified cef hamiltonian to calculate the resultant change of the pr@xmath5 local susceptibility.@xcite in pros@xmath1sb@xmath2 the pr@xmath5 first excited state is separated from the singlet ground state by only @xmath910 k.@xcite it is particularly important to determine possible muon - induced changes in the local cef energy levels , since cef excitations are central to a number of theories of superconductivity in pros@xmath1sb@xmath2.@xcite changes in cef splitting might therefore affect local superconducting properties such as the vortex - state supercurrent .
tf-@xmath0sr experiments were carried out at the m15 and m20 channels at triumf , vancouver , canada , on a mosaic of oriented pros@xmath1sb@xmath2 crystals .
the crystals were mounted on a thin gaas backing , which rapidly depolarizes muons in transverse field and minimizes any spurious signal from muons that do not stop in the sample .
tf-@xmath0sr asymmetry data@xcite were taken for temperatures in the range 0.02250 k and @xmath75 between 10 mt and 1.0 t applied parallel to the @xmath76100@xmath77 axes of the crystals . for the @xmath12 knight shift measurements ,
the applied field was determined by measuring the precession frequency of muons that stopped in a small piece of silver foil included with the sample . for the superconducting - state tf-@xmath0sr measurements the magnetic field @xmath36 was applied in the normal state , and the sample was cooled in constant field to below @xmath78 . for pros@xmath1sb@xmath2 @xmath79 oe , @xcite
so that @xmath36 was significantly larger than @xmath80 and the sample was always in the vortex - lattice state .
the demagnetizing field @xmath81 , where @xmath82 is the demagnetization coefficient and @xmath83 in the superconducting state , adds to the field @xmath84 inside the sample , rendering it closer to the applied field . in sect .
[ sect : knight ] we estimate an effective @xmath85 , which yields an estimated variation of @xmath60 with @xmath80 ( i.e. , with temperature ) of only a few percent .
the strong bulk vortex pinning in pros@xmath86sb@xmath2 ( ref . )
also suppresses changes in @xmath60 .
thus modification of the vortex - state field distribution by flux expulsion , demagnetization effects , etc .
, seems unlikely to play an appreciable role in the experiments .
in this section we describe the temperature and field dependencies of the tf-@xmath0sr relaxation data in superconducting pros@xmath86sb@xmath2 , and compare the superconducting penetration depth @xmath4 obtained from these data with inductive measurements @xcite in the meissner state .
we concentrate on @xmath4 at low temperatures , where power - law behavior is evidence for gap nodes and temperature - independent ( or activated ) behavior signals a fully - gapped fermi surface .
the behavior of the data at higher temperatures is more complicated and will not be considered in detail .
furthermore , as noted above in sec .
[ sect : tfmusr ] , the effective tf-@xmath0sr penetration depth is expected to be field dependent , and approximates the true value only as @xmath88.@xcite thus we concentrate on the results at low temperatures and ( for the tf-@xmath0sr data ) low fields .
figure [ fig : lambda ] compares @xmath89 in in pros@xmath1sb@xmath2 below @xmath91 k. squares : @xmath34 from gl model fitting .
curve : fit to bcs low - temperature expression @xmath90 $ ] , @xmath91 , @xmath92 .
triangles : @xmath93 from rf inductive measurements@xcite in the meissner state ( see text ) . curve : fit to power law @xmath94 , @xmath95 .
inset : temperature dependence of @xmath34 up to @xmath96.@xcite squares : @xmath34 as in main panel .
circles : @xmath34 from gaussian fits and eq .
( [ eq : lambda]).,scaledwidth=45.0% ] pros@xmath1sb@xmath2 at low temperatures , obtained from gl model fits to tf-@xmath0sr data taken at @xmath97 mt , with the penetration depth increase @xmath98 in the meissner state obtained from rf inductive measurements.@xcite the rf data do not determine @xmath99 separately ; in fig .
[ fig : lambda ] the quantity plotted ( triangles ) is @xmath100 , which allows direct comparison of the temperature dependencies from the two experiments . it can be seen that @xmath89 is nearly constant below @xmath90.6 k , indicative of a fully gapped quasiparticle excitation spectrum . as previously noted@xcite the bcs low - temperature expression @xmath101 ( curve in fig .
[ fig : lambda ] ) gives a good fit to the tf-@xmath0sr data for @xmath102 , suggesting that the energy gap is isotropic .
the increase of @xmath103 ( triangles ) with increasing temperature is much stronger than the increase of @xmath89 and follows a power law @xmath94 , @xmath104 ( ref . ) .
this is the discrepancy between the measurements noted above ; a possible resolution is discussed in sec .
[ sect : mbsc ] . as a test of the sensitivity of @xmath34 to the data fitting procedure
, the functional form @xmath105 , with a damped gaussian envelope @xmath106 was also used to fit the tf-@xmath0sr data from both the normal and superconducting states . in the normal state , where this function provides a good fit ,
the gaussian relaxation arises from quasistatic sb nuclear dipolar fields .
the exponential damping increases with increasing field and is mainly due to normal - state susceptibility inhomogeneity ; dynamic fluctuations of hyperfine - enhanced @xmath107pr nuclear spins@xcite play a small role at low fields .
below @xmath96 the exponential rate @xmath108 was fixed at the normal - state value for each field , so that the temperature dependence of the gaussian rate @xmath109 reflects the effect of the superconducting state .
some such procedure is necessitated by the strong statistical correlation between @xmath108 and @xmath109 in eq .
( [ eq : dg ] ) ; the time constant and the shape of the relaxation function are influenced by both of these parameters , so that correlations between them can result from small systematic errors .
the principal justification for this _ ad hoc _ fixing of @xmath108 is the insensitivity of the superconducting - state results to details of the fitting function [ eq . ( [ eq : dg ] ) ] discussed below .
determination of the vortex - state field distribution width requires correction for the normal - state relaxation .
we take the superconducting - state gaussian rate @xmath110 to be given by @xmath111 , where @xmath112 is the normal - state rate.@xcite equation ( [ eq : distri ] ) relates the second moment @xmath113 of field distribution to @xmath10 in the london limit .
the second moment of the corresponding @xmath12 frequency distribution is @xmath114 , where @xmath32 is the @xmath12 gyromagnetic ratio .
then the estimated penetration depth @xmath10 from eq .
( [ eq : distri ] ) is @xmath115 as noted above the rms width @xmath110 of the best - fit gaussian is not necessarily @xmath116 , so that replacement of @xmath116 in eq .
( [ eq : lambda ] ) by @xmath110 is not completely justified .
nevertheless @xmath110 should scale with @xmath116 , and within its range of validity eq .
( [ eq : lambda ] ) should give the correct temperature dependence of @xmath10 .
this is because under these circumstances effects of nonzero @xmath13 are restricted to the high - field tail of field distribution , which is not heavily weighted in a gaussian fit ( cf .
fig . 1 of ref . ) .
pros@xmath1sb@xmath2 is a strongly type - ii superconductor ( gl @xmath117 , refs . and ) , and this picture should be applicable .
the temperature dependencies of @xmath34 obtained from gaussian fits and eq .
( [ eq : lambda ] ) ( circles ) and from gl model fits ( squares)@xcite are shown in the inset of fig .
[ fig : lambda ] for an applied field of 10 mt .
the results agree very well , giving strong evidence that @xmath34 is robust with respect to very different fitting strategies .
figure [ fig : rate ] gives the temperature dependence of the corrected spin relaxation rates @xmath118 in pros@xmath1sb@xmath2 .
( a ) : applied field @xmath119 mt ( circles ) .
( b ) : @xmath120 mt ( squares ) .
( c ) : @xmath121 mt ( open triangles ) .
solid triangles : 100 mt data from ref . .
curves : fits of the power law @xmath122 $ ] to the tf-@xmath0sr data .
diamonds : inverse - square meissner - state penetration depth @xmath123 from rf data ( ref . )
.,scaledwidth=45.0% ] superconducting - state @xmath12 spin relaxation rate @xmath124 , normalized to the zero - temperature value @xmath125 , for @xmath119 , 20 , and 100 mt .
the relaxation rates are well described by a power - law temperature dependence @xmath126 $ ] ( curves in fig .
[ fig : rate ] ) over the entire temperature range @xmath127 , with the parameter values given in table [ table : power - law ] .
.parameters from power - law fits @xmath128 $ ] to the corrected superconducting - state @xmath129 spin relaxation rate @xmath130 in pros@xmath1sb@xmath2 . [ cols="^,^,^,^",options="header " , ] we take @xmath45 to be the slope of @xmath41-@xmath72 in fig .
[ fig : pos_knight ] , since the tensor @xmath43 is traceless and the sum @xmath131 vanishes as can be seen from table [ table - a ] .
in order to obtain the values of @xmath132 , we must modify @xmath52 to produce the required level changes . in the presence of a muon ,
the symmetry of the neighboring pr@xmath5 ions is not the original cubic symmetry , but rather a very approximate tetragonal ( @xmath133 ) symmetry .
the general hamiltonian describing this tetragonal symmetry is@xcite @xmath134 which has five free parameters , compared to two in the original cubic hamiltonian . in principle , the muon may modify the charge distribution around the pr ions in every direction .
however , if we only consider changes occurring along the pr@xmath5-@xmath12 axis for simplicity , it has been shown that @xmath135 and @xmath136 remain at their cubic values.@xcite this reduces the number of the free parameters from five to three .
hence , using eq .
( [ eq : new knight ] ) , we can fit the @xmath12 knight shift by choosing appropriate values of @xmath137 , @xmath138 , and @xmath139 .
our theoretical fits in pros@xmath1sb@xmath2 are shown in fig .
[ fig : pos_knight_cef ] . knight shift in pros@xmath1sb@xmath2 .
circles : @xmath12 knight shift @xmath41 from fig .
[ fig : pos_knight ] .
error bars : " gaussian linewidth @xmath140 [ fig . [ fig : pos_fre](c ) ] .
dash - dot curve : calculated @xmath141 .
dashed curve : calculated @xmath142 .
dot curve : calculated @xmath143 .
solid curve : average shift @xmath144 .
dash - dot - dot curve : @xmath145 from linear fit ( fig .
[ fig : pos_knight ] ) , expected for no @xmath12 perturbing effect.,scaledwidth=45.0% ] the anisotropy is consistent with the new cef and eq .
( [ eq : new knight ] ) , decreases with increasing temperature , and is well within the spread @xmath140 of @xmath12 shifts ( shown as `` error bars '' in fig .
[ fig : pos_knight_cef ] ) .
the data are very well described by the average shift @xmath144 , which is almost the same as @xmath145 from fig .
[ fig : pos_knight ] that assumes no @xmath12 perturbing effect .
the fitted cef parameters are listed in table [ table - b ] under the columns with heading @xmath133 .
the changes in the parameters , which are very small and probably not statistically significant , give rise to a correspondingly small rearrangement of the local energy levels .
a comparison of bulk and locally perturbed energy level scheme of pros@xmath1sb@xmath2 is shown in fig .
[ fig : energy ] .
we see that the change of splitting between the ground state and first excited state energies is only 0.3 k , and conclude that the perturbing effect of the @xmath12 charge in pros@xmath1sb@xmath2 is negligibly small .
sb@xmath2 , compared to the local level scheme induced by @xmath12.,scaledwidth=45.0% ]
in this paper we have presented tf-@xmath0sr results in the first pr - based heavy - fermion superconductor pros@xmath1sb@xmath2 .
it is found experimentally that the effective penetration depth @xmath146 is temperature - independent in the vortex state for low temperatures , consistent with a nonzero gap for quasiparticle excitations .
in contrast , the temperature dependence of the penetration depth from rf inductive measurements suggests point nodes in the gap .
this discrepancy can be resolved in a scenario based on the recent discovery of two - band superconductivity in pros@xmath1sb@xmath2 .
the temperature dependence of the normal - state @xmath12 knight shift in pros@xmath86sb@xmath2 reveals a linear scaling of the knight shift with the bulk magnetic susceptibility at high temperatures .
a very small deviation from the linear relation appears in pros@xmath86sb@xmath2 below @xmath147 k. such deviations have been explained by @xmath12 induced modifications of the susceptibility of neighboring pr@xmath5 ions due to a change of the cef splitting .
our data indicate , however , that this modification is very small in pros@xmath86sb@xmath2 .
a model calculation based on cef theory and the associated perturbed electronic energy levels confirms the smallness of the @xmath12 perturbation effect .
therefore it is unlikely that the discrepancy between tf-@xmath0sr and rf inductive measurements is caused by the @xmath12 charge or , more generally , that the @xmath12 charge influences the role of the pr@xmath5 cef states in the superconductivity of pros@xmath86sb@xmath2 .
we are grateful for technical assistance from the triumf centre for molecular and materials science during the experiments .
this work was supported in part by the u.s .
nsf under grants 0422674 ( riverside ) , 0604015 ( los angeles ) and 0335173 ( san diego ) , by the canadian nserc and ciar ( burnaby ) , and by the u.s .
doe under grant de - fg-02 - 04er46105 ( san diego )
. work at los alamos was performed under the auspices of the u.s .
doe .
l. shu , d. e. maclaughlin , r. h. heffner , f. d. callaghan , j. e. sonier , g. d. morris , o. o. bernal , a. bosse , j. e. anderson , w. m. yuhasz , n. a. frederick , and m. b. maple , physica b * 374 - 375 * , 247 ( 2006 ) .
l. shu , d. e. maclaughlin , y. aoki , y. tunashima , y. yonezawa , s. sanada , d. kikuchi , h. sato , r. h. heffner , w. higemoto , k. ohishi , t. u. ito , o. o. bernal , a. d. hillier , r. kadono , a. koda , k. ishida , h. sugawara , n. a. frederick , w. m. yuhasz , t. a. sayles , t. yanagisawa , and m. b. maple , phys .
b * 76 * , 014527 ( 2007 ) . | transverse - field muon spin rotation ( tf-@xmath0sr ) experiments in the heavy - fermion superconductor pros@xmath1sb@xmath2 ( @xmath3 k ) suggest that the superconducting penetration depth @xmath4 is temperature - independent at low temperatures , consistent with a gapped quasiparticle excitation spectrum .
in contrast , radiofrequency ( rf ) inductive measurements yield a stronger temperature dependence of @xmath4 , indicative of point nodes in the gap .
muon knight shift measurements in the normal state of pros@xmath1sb@xmath2 suggest that the perturbing effect of the muon charge on the neighboring pr@xmath5 crystalline electric field is negligibly small , and therefore is unlikely to cause the difference between the tf-@xmath0sr and rf results .
the discrepancy appears to be related to multiband superconductivity in pros@xmath1sb@xmath2 . |
the proof below involves some elementary vector algebra and , crucially , use of the equations of motion , which can be assumed to be of the form @xmath73 where @xmath74 is the position of the orbiting body relative to the force center ( the orbital plane is taken the @xmath75-plane with the @xmath76-axis parallel to the angular momentum per unit mass ) . by definition , under polar reciprocation , the point @xmath10 with position vector @xmath77 ( on trajectory @xmath11 )
is mapped to the point @xmath18 ( on @xmath17 ) with position vector @xmath78 , where @xmath79 .
likewise , under polar reciprocation , @xmath18 is mapped to the point @xmath38 with position vector @xmath80 , where @xmath81 . substituting @xmath82 into the definition of @xmath83
, one finds that @xmath84/c , \ ] ] where @xmath85 ( @xmath86 ) . using eq .
( [ eq : eom ] ) and the fact that @xmath6 is a constant of the motion , @xmath87 can be simplified to @xmath88/c,\ ] ] and , hence , @xmath89 equations ( [ eq : rstar ] ) and ( [ eq : cstar ] ) imply that @xmath90 or @xmath91 ( as required ) .
\(1 ) it is not unusual to see the orbital plane ( or @xmath75-plane ) identified with the complex plane ( or @xmath76-plane ) via the ( obvious ) isomorphism @xmath92 .
formulate polar reciprocation as an operation in the complex plane . use this representation to prove polar reciprocation is involutory.@xcite + ( 2 ) the polar reciprocal @xmath0 of a known orbit @xmath1 of the _ kepler problem _ can be easily found . the claim that the images under polar reciprocation of just two points on @xmath1 suffice to fix @xmath0 .
+ b ) show that , with the @xmath93-axis aligned along @xmath94 , the velocity at points around @xmath1 is @xmath95 in the notation of this paper .
+ c ) by considering the _ periapse _ @xmath96 of @xmath1 ( where @xmath97 and @xmath98 ) , demonstrate that the energy @xmath99 of the orbiting body is related to the eccentricity @xmath46 of @xmath1 by @xmath100 where @xmath101 is the mass of the orbiting body .
( begin by finding expressions for @xmath102 and @xmath103 in terms of @xmath57 and @xmath46 . ) + d ) use eqs .
( [ eq : vandphi ] ) and ( [ eq : energy ] ) to prove that the inequality @xmath104 reduces to @xmath105 .
+ ( 3 ) the polar reciprocal of an elliptical orbit ( of the kepler problem ) must be a _
complete _ circle ( as opposed to only an arc of a circle ) because of the periodicity of the motion . show that this circle has radius @xmath106 where @xmath107 and @xmath108 are the distances ( from the force center @xmath2 ) of the _ periapse _ and _ apoapse _ , respectively , and that the distance of its center @xmath55 from @xmath2 is @xmath109 infer from eqs .
( [ eq : rhoe ] ) and ( [ eq : ostare ] ) an expression for the eccentricity of the orbit .
+ \(4 ) figure [ fig : prhyperbola ] depicts the polar reciprocal @xmath0 of the hyperbolic orbit @xmath1 in fig .
[ fig : hyperbola ] .
prove that @xmath110 is tangent to the circle in fig .
[ fig : prhyperbola ] of which @xmath0 is a part by showing that @xmath111 . by appealing to the `` tangent - secant theorem'',@xcite
deduce that @xmath112 .
hence , conclude that , in fig .
[ fig : prhyperbola ] , the circle has radius @xmath113 and @xmath114 use eqs .
( [ eq : rhoh ] ) and ( [ eq : oostarh ] ) to justify the assertion that results appropriate to parabolic orbits can be obtained by taking the limit @xmath115 , keeping @xmath116 fixed . + ( 5 ) many spacecraft maneuvers involve short firings of high - thrust rockets , which can be modeled as resulting in instantaneous changes @xmath117 in velocity with no change in position .
the important hohmann transfer ( depicted in fig .
[ fig : hto ] ) requires two such impulsive thrusts : the first ( at @xmath118 ) places the vehicle on the semi - elliptical trajectory @xmath119 , and the second ( at @xmath120 ) puts the vehicle into the circular orbit @xmath121.@xcite + a ) show that , provided @xmath117 is tangential to the spacecraft s orbit and the direction of motion of the spacecraft is not reversed , @xmath122 is unchanged by an impulsive thrust .
+ b ) find the polar reciprocal representation of the hohmann transfer in fig .
[ fig : hto ] , i.e. draw ( in one diagram ) the polar reciprocals of @xmath123 , @xmath121 , and @xmath119 .
+ c ) generalize the result of part b ) to the case of two counter - clockwise non - intersecting co - planar and _ co - axial _ elliptical orbits about the force center @xmath2 .
the transfer trajectory must connect the periapse of the inner orbit to the apoapse of the outer orbit .
99 h. abelson , a. disessa and l. rudolph ,
velocity space and the geometry of planetary orbits , " am .
j. phys .
* 43 * , 579 - 589 ( 1975 ) .
a. gonzlez - villanueva , h. n. nez - ypez , and a. l. salas - brito , in veolcity space the kepler orbits are circular , " eur .
j. phys .
* 17 * , 168 - 171 ( 1996 ) .
t. a. apostolatos , hodograph : a useful geometrical tool for solving some difficult problems in dynamics , " am .
j. phys .
* 71 * , 261 - 266 ( 2003 ) .
e. i. butikov , comment on ` eccentricity as a vector ' , " eur .
j. phys .
* 25 * , l41-l43 ( 2004 ) . c. i. mungan ,
another comment on ` eccentricity as a vector ' , " eur . j. phys .
* 26 * , l7-l9 ( 2005 ) .
d. l. goodstein and j. r. goodstein , _
feynman s lost lecture : the motion of planets around the sun _ ( w. w. norton , new york , 1996 ) .
g. w. griffiths , math .
intelligencer * 20 * ( 3 ) , 68 - 70 ( 1998 ) , review of ref . .
r. weinstock , math .
intelligencer * 21 * ( 3 ) , 71 - 73 ( 1999 ) , review of ref . .
d. derbes , reinventing the wheel : hodographic solutions to the kepler problems , " am .
j. phys .
* 69 * , 481 - 489 ( 2001 ) .
e. guillaumin - espaa , a. l. salas - brito , and h. n. nez - ypez , tracing a planet s orbit with a straight edge and a compass with the help of the hodograph and the hamilton vector , " am .
j. phys .
* 71 * , 585 - 589 ( 2003 ) .
d. w. tiberiis , comment on ` reinventing the wheel : hodographic solutions to kepler problems , ' by david derbes [ am .
j. phys .
69(4 ) , 481 - 489 ( 2001 ) ] , " am .
j. phys .
* 70 * , 79 ( 2002 ) .
i. newton , _ philosophiae naturalis principia mathematica _ ( university of california press , berkeley , ca , 1934 ) translation ( into english ) by motte ( revised by cajori ) .
d. chakerian , central force laws , hodographs , and polar reciprocals , " mathematics magazine * 74 * , 3 - 18 ( 2001 )
. trajectories for which @xmath124 have to be excluded , but these are one - dimensional and easily drawn once the initial conditions are known .
see d. v. anosov , a note on the kepler problem , " j. dynamical and control systems * 8 * , 413 - 442 ( 2002 ) .
equation ( [ eq : kstar ] ) is implicit in the existing literature on hodographs .
for example ,
( [ eq : kstar ] ) can be inferred from eq .
( 3 ) of ref . , which is a concise and extremely clear treatment of hodographs for the kepler problem .
algebraic considerations specific to the kepler problem are used to motivate the introduction of a _ sum vector _ , which is exactly the combination of vectors in eq .
( [ eq : kstar ] ) , in m. d. vivarelli , a configuration counterpart of the kepler problem hodograph , " celestial mechanics and dynamical astronomy * 68 * , 305 - 311 ( 1998 ) . h. s. m. coxeter , _ introduction to geometry _ , 2nd ed . ( john wiley & sons , new york , 1969 ) .
this result could also have been established directly by using the fact that @xmath125 to recast eq .
( [ eq : kstar ] ) as the relation @xmath126 , where @xmath101 is the mass of the orbiting particle and is the laplace - runge - lenz vector as defined in h. goldstein , _ classical mechanics _ , 2nd ed .
( addison - wesley , reading , ma , 1980 ) .
part d ) of problem 2 in appendix [ app : suggestedproblems ] addresses the origin of this angular restriction on @xmath0 for @xmath60 .
the hyperlink on this web page to _
last _ of these three operations appears between hyperlinks 29 and 30 ( and directs one to the web page @xmath70http://www.susqu.edu / brakke / constructions / inversiontool.htm@xmath71 ) ; the hyperlinks ( numbered 3 and 8) to the other two operations are more obvious .
a less ephemeral source for all of these constructions is e. c. wallace and s. f. west , _ roads to geometry _
( prentice - hall , upper saddle river , nj , 1998 ) , which deals with inversion of a point in the unit circle on p. 278 , and the other two constructions on pp . 191 - 2 .
p. t. tam , _ a physicist s guide to mathematica _
, 2nd ed .
( academic press , burlington , ma , 2008 ) . other applications of the complex variable formalism to central force problems appear in m. m. deliseo , `` the first - order orbital equation , '' am .
j. phys .
* 75 * , 352 - 355 ( 2007 ) . alternatively , euclid s proposition iii.36 on secants of a circle ( stated on page 8 in ref . ) .
a brief but usually careful introduction to the hohmann transfer of fig .
[ fig : hto ] is given in section 7.6 of r. d. gregory , _ classical mechanics _
( cambridge university press , cambridge , 2006 ) . | it is argued that , for motion in a central force field , polar reciprocals of trajectories are an elegant alternative to hodographs .
the principal advantage of polar reciprocals is that the transformation from a trajectory to its polar reciprocal is its own inverse .
the form of polar reciprocals @xmath0 of kepler problem orbits is established , and then the orbits @xmath1 themselves are shown to be conic sections using the fact that @xmath1 is the polar reciprocal of @xmath0 .
a geometrical construction is presented for the orbits of the kepler problem starting from their polar reciprocals .
no obscure knowledge of conics is required to demonstrate the validity of the method . unlike a graphical procedure suggested by feynman ( and amended by derbes ) , the algorithm based on polar reciprocals works without alteration for all three kinds of trajectories in the kepler problem ( elliptical , parabolic , and hyperbolic ) .
approximately , a hodograph is a plot of velocities along a trajectory ; more precisely , it is the locus of the tips of the velocity vectors after they have been parallelly transported until their tails are at the origin ( in velocity space ) .
there have been many articles on the pedagogic virtues of hodographs.@xcite _ _ feynman s lost lecture__@xcite contains , amongst other things , a recipe for drawing the elliptical path of a planet given its hodograph . the procedure , as reproduced in ref . , has some shortcomings,@xcite but , fortunately , these have been more than satisfactorily rectified by derbes.@xcite derbes notes that feynman s scheme also works for hyperbolic orbits and he is able to devise another construction for the exceptional case of parabolic orbits .
a completely different way of tracing all three kinds of orbits with the help of hodographs has been successfully developed by salas - brito and co - workers in a series of publications culminating in ref . .
the various geometrical methods of the previous paragraph are easily implemented , but , to prove their validity , a student would have to be acquainted with many properties of conics which no longer form part of most school curricula .
indeed , at least one reader of derbes article feels that his exposure to the hodograph left him `` disappointed by the trade - off of intricate calculus for obscure geometry''.@xcite there is , however , an alternative approach which requires only some elementary vector algebra and calculus for its justification .
the seed for this other construction can be traced back to a result of newton ( proposition i , corollary i on page 41 of ref . ) arising from the conservation of angular momentum in a central force field .
let be the position vector of a body relative to the center @xmath2 of the force field and let be the body s angular momentum per unit mass with respect to @xmath2 ; then , @xmath3 where @xmath4 is the component of perpendicular to the instantaneous velocity @xmath5 ( relative to @xmath2 ) : since @xmath6 is a constant of the motion , the speed @xmath7 of the body is inversely proportional to @xmath4 ( @xmath8 ) .
consider now the mapping @xmath9 illustrated in fig .
[ fig : defpr ] .
the point @xmath10 is on a typical trajectory @xmath11 of a body in an ( attractive ) central force field ( with center @xmath2 ) ; of course , @xmath11 is in the plane through @xmath2 perpendicular to ( and so are and ) . by construction ,
the vector @xmath12 ( in this plane ) is perpendicular to the velocity at @xmath10 and has magnitude @xmath13 . since @xmath14 [ from eq .
( [ eq : c ] ) ] and the direction of @xmath12 is obtained from that of @xmath15 by a clockwise rotation through @xmath16 ( see fig . [
fig : defpr ] ) , _ the curve @xmath17 _ in fig .
[ fig : defpr ] , which is the locus of points @xmath18 as @xmath10 varies over @xmath11 , is _ the hodograph of @xmath11 , rescaled by a factor of @xmath19 and rotated clockwise through 90@xmath20 _ ( see also section 3 of ref . ) .
following ref .
, i will term @xmath17 the _ polar reciprocal _ of @xmath11 .
it is the inverse of the mapping @xmath9 in fig .
[ fig : defpr ] which can be used to draw trajectories .
in fact , there is a pleasing symmetry .
the mapping @xmath9 is involutory , i.e. @xmath21 ( see appendix [ app : involution ] for an elementary proof ) , so trajectories in a central force field and their hodographs ( after rotation and rescaling as in the preceding paragraph ) are polar reciprocals of each other . _
the point - by - point determination of a trajectory from a rotated and rescaled hodograph involves exactly the steps depicted in fig .
[ fig : defpr]_.@xcite unlike the methods of refs . and , polar reciprocation is valid for _ any _ smooth hodograph associated with _ any _ central force field . be that as it may , i will now specialize to orbits of the kepler problem .
i will also use the identity @xmath22 which takes advantage of the fact that , for the trajectory @xmath11 depicted in fig .
[ fig : defpr ] , points perpendicularly out of the page , so that @xmath23 is parallel to @xmath12 and has magnitude @xmath24 . for an inverse - square force per unit mass of @xmath25 ( @xmath26 ) , the equation of motion for the position vector @xmath12 of a typical point @xmath18 on the polar reciprocal of an orbit ( see fig .
[ fig : defpr ] ) reads @xmath27 where @xmath28 and plane polar coordinates @xmath29 and @xmath30 have been adopted for the orbital plane ; the corresponding unit vectors are @xmath31 and @xmath32 with the origin of the coordinate system at the force center @xmath2 , and the azimuthal angle @xmath30 defined as in fig . [
fig : prk ] ( so that @xmath33 ) .
equation ( [ eq : tdopstar ] ) implies that @xmath34 is a ( vectorial ) constant of the motion , say @xmath35 ( drawn in fig .
[ fig : prk ] ) . setting @xmath36 , @xmath37 which demonstrates that the polar reciprocal @xmath0 of an orbit @xmath1 of the kepler problem is a circle or an arc of a circle ( as suggested by the plot in fig .
[ fig : prk]).@xcite the polar reciprocal of @xmath0 ( which would be the corresponding orbit @xmath1 ) comprises points like @xmath38 in fig .
[ fig : prk ] , the image ( under polar reciprocation ) of @xmath18 on @xmath0 . in terms of the angles in fig .
[ fig : prk ] , @xmath39^{-1}=(\mbox{\overrightharp{$ op$}}_*\cdot\widehat{e}_r)^{-1}$ ] or , evaluating @xmath40 with eq .
( [ eq : kstar ] ) , @xmath41 comparison of eq . (
[ eq : rsqd ] ) with the standard equation@xcite for a conic in polar coordinates confirms that @xmath38 is on a conic with focus @xmath2 , eccentricity @xmath42 , _ latus rectum _
@xmath43 , and directrix perpendicular to @xmath44 .
the angle @xmath30 can thus be identified as the _ true anomaly _ ( of celestial mechanics ) , and @xmath45 can be reinterpreted as a vector of magnitude equal to the eccentricity @xmath46 of the orbit @xmath1 , directed from the force center @xmath2 to the point on @xmath1 of closest approach ( i.e. the _ periapse _ ) .
in fact , to within a constant , @xmath45 is the laplace - runge - lenz vector.@xcite in the previous two paragraphs , it has been shown that it is easy to establish the character of the polar reciprocal @xmath0 of an orbit @xmath1 of the kepler problem and even easier to infer from @xmath0 that the orbit @xmath1 must be a conic section .
it is now possible to indicate how , as an alternative to the methods of refs . and
, polar reciprocation may be used , in principle , to draw orbits with a compass and ruler .
suppose that the orbiting body s velocity is given at some point @xmath47 on the orbit ( which need not coincide with an apse of the orbit ) ; let @xmath48 be the position vector of @xmath47 relative to @xmath2 , and let the associated velocity ( in the rest frame of @xmath2 ) be @xmath49 .
first , one must construct @xmath0 .
the following sequence of steps can be employed ( @xmath50 below denotes the image under polar reciprocation of @xmath47 ) : * calculate @xmath51 and then @xmath52 ; * add graphically the vectors @xmath53 and @xmath54 to find the position of @xmath55 relative to @xmath2 ; * determine the eccentricity @xmath56 ; * at @xmath55 , draw a circle of radius @xmath57 if @xmath58 or a circular arc on which the points @xmath18 are such that @xmath59 if @xmath60 ( cf .
fig .
[ fig : prhyperbola ] in appendix [ app : suggestedproblems]).@xcite from a point @xmath18 on @xmath0 , the corresponding point @xmath38 on @xmath1 can be determined by demanding that @xmath61 is parallel to @xmath62 and has magnitude @xmath63 .
well - known compass - and - ruler constructions suffice .
one can : * erect at @xmath18 a line @xmath64 perpendicular to the line through @xmath65 ; * draw through @xmath2 the line @xmath66 parallel to the line segment @xmath65 , and then ; * invert the point of intersection of @xmath67 and @xmath68 ( the point @xmath69 in fig .
[ fig : prk ] ) with respect to the unit circle centered on @xmath2 .
compass - and - ruler implementations ( with _ javasketchpad _ ) of all three of these procedures can be found at @xmath70http://www.susqu.edu / brakke / constructions / constructions.htm@xmath71 ( accessed july 2 , 2011).@xcite for students today , who have access to programs like _ mathematica _ , it may seem that there is little or no need for the geometrical construction under discussion ; it takes only a few keystrokes ( as in example 2.1.9 of ref . ) to generate a plot of a kepler problem orbit ; i have , nevertheless , witnessed the intellectual satisfaction students experience in being able to correctly anticipate the character and orientation of an orbit : these features can be deduced from initial conditions via steps ( i ) to ( iii ) of the preceding paragraph
. a more traditional alternative would be to use the laplace - runge - lenz vector , but i have found that even my most talented and dedicated junior - level students are somewhat mystified by this construct and disinclined to use it , when it is introduced in isolation ( as it usually is ) as a single - valued combination of dynamical variables which happens to be a constant of the motion .
my students have been more receptive to this `` exotic '' constant of the motion when it is presented within the context of a discussion of hodographs or polar reciprocals of the kepler problem .
other more mundane issues can be tackled .
some suggestions for problems are given in appendix [ app : suggestedproblems ] .
i would like to thank one of the anonymous referees of this paper for suggestions on improvements . |
minimizing energy consumption is one of the primary technical challenges in sensor networking .
many sensor applications such as habitat monitoring and industrial instrumentation envisage scenarios in which a large number of sensor nodes , powered by tiny batteries , will be actively deployed for months and even years . in many instances
, it may not be possible to replace these sensor nodes once they run out of energy because the sensor nodes could be inaccessible ( for example , embedded in concrete structures to sense stress levels ) .
replacing dead batteries in a sensor network consisting of a large number of nodes may also not be economically feasible .
there is a now a broad consensus that aggressive system level strategies impacting many layers of the protocol stack need to be devised to meet the lifetime requirement of extant and future sensor networks . in this paper
, we consider a single - hop sensor cluster .
nodes in the cluster periodically sample a field and transmit the data directly to a central location or base - station .
we are interested in minimizing the energy spent by these nodes in transmitting , with the objective of maximizing cluster lifetime .
sensor nodes also spend energy in receiving data , sensing / actuating and computation . the energy spent in
sensing / actuating represents a fixed cost that can be ignored .
the energy cost of receiving data can be easily incorporated in our optimization model .
we assume computation costs are negligible compared to radio communication costs .
this is debatable assumption in dense networks ; we intend to incorporate computation costs in future work .
we believe that our model is useful because many popular proposals recommend organizing a sensor network into clusters @xcite . here
each cluster elects a cluster - head ( which we call base - station ) .
nodes communicate only through their respective cluster - heads .
approaches that maximize cluster lifetime can be thought of as being complementary to network - wide approaches such as energy - efficient routing .
moreover , our model is applicable to scenarios where a roving base - station moves from one cluster to another , gathering data .
we define cluster lifetime as the time until the first node in the cluster runs out of energy .
while this is a somewhat pessimistic definition , we argue that a cluster will consist of relatively few nodes .
the failure of even one such node can have disastrous consequences on the cluster s performance ( for example , coverage ) .
this definition also has the benefit of being simple and popular @xcite .
other definitions proposed for network lifetime such as mean expiration time and time until a certain fraction of nodes fail are not appealing from a cluster viewpoint . somewhat to our surprise
, we find that analyzing the performance of this simple model is far from trivial .
the complexity arises from several competing system - level opportunities to be harnessed to reduce the energy consumed in radio transmission .
first , sensor data in a cluster is spatially and temporally correlated . in @xcite , slepian and wolf
show that it is possible to compress a set of correlated sources down to their joint entropy , without explicit communication between the sources .
this surprising existential result shows that it is enough for the sources to know the probability distribution of data generated .
recent advances @xcite in distributed source coding allow us to take advantage of data correlation to reduce the number of bits that need to be transmitted , with concomitant savings in energy .
second , it is also well known that channel coding @xcite can be used to reduce transmission energy by increasing transmission time . finally ,
sensor nodes are cooperative , unlike nodes in an ad hoc network that are often modeled as competitive .
this collaborative nature allow us to take full advantage of the first two opportunities for the purpose of maximizing cluster lifetime . motivated by our definition of cluster lifetime
, we pose the problem of maximizing lifetime as a max - min optimization problem subject to the constraint of successful data collection and limited energy supply at each node .
this turns out to be an extremely difficult optimization to solve .
consequently , we employ a notion of instantaneous decoding to shrink the problem space .
we show that the computational complexity of our model is determined by the relationship between energy consumption and transmission rate as well as model assumptions about path loss and initial energy reserves .
we provide some algorithms , heuristics , and insights for several scenarios . in some situations , our problem admits a greedy solution while in others , the problem is shown to be @xmath0-hard .
the chief contribution of the paper is to illustrate both the challenges and gains provided by source - channel coding and scheduling .
there is much related work in this area .
energy conscious networking strategies have been proposed by many researchers mainly at the mac @xcite and routing layer @xcite .
our study was motivated by previous research in @xcite , which explicitly incorporate aggregation costs in gathering sensor data . in @xcite ,
the authors consider the problem of correlated data gathering by a network with a sink node and a tree communication structure .
their goal is to minimize the total transmission ( energy ) cost of transporting information .
the first part of @xcite considers a model similar to ours , namely , that of several correlated nodes transmitting directly to a base station .
however , both @xcite and @xcite are interested in minimizing total energy expenditure , as opposed to maximizing network lifetime . in the latter case ,
the optimal solution is shown in both papers to be a greedy solution based on ordering sensors according to their distance ( which reflects data aggregation cost ) from the base station .
however , we show that this solution is not optimal for maximizing network lifetime
. an early version of our ideas appeared in @xcite .
this paper generalizes the work presented in @xcite and provides proofs of some key conjectures there . in section [ model ]
, we present our system model and describe our notion of instantaneous decoding . in section [ gc ]
, we consider a general channel model which allows us to consider the joint impact of cooperative nature of the sensor nodes and source and channel coding on system lifetime .
we prove that the both , the static and dynamic scheduling problems for the general channel model , are @xmath0-hard and the optimal dynamic scheduling strategy does better than optimal static scheduling strategy , in general .
we also provide the geometric interpretation of the optimal solutions and the solution search procedures . as a special case of this problem , in section [ srra ]
, we consider a scenario which allows us to neglect the impact of transmission time allocation .
this is similar to the scenario considered in @xcite and @xcite .
here we provide some key insights into the nature of the optimal solutions for both , static and dynamic scheduling , derived in @xcite .
we consider a network of @xmath1 battery operated sensor nodes strewn uniformly in a coverage area .
time is divided into slots or rounds . in each slot , sensors take samples of the coverage area and transmit the information directly to the base station .
we model the sensor readings at node @xmath2 by a discrete random process @xmath3 representing the sampled reading value at node @xmath2 in the @xmath4 time slot .
we assume that sensor readings in any time slot are spatially correlated .
we ignore temporal correlation by assuming that sensor readings in different time slots are independent .
however , temporal correlation can easily be incorporated in our work for data sources satisfying the asymptotic equipartition property ( aep ) .
the entropy of @xmath3 is denoted by @xmath5 .
initially , sensor node @xmath6 , has access to @xmath7 units of energy .
the wireless channel between sensor @xmath2 and the base station is described by a path loss factor @xmath8 , which captures various channel effects such as distance induced attenuation , shadowing , and multipath fading . for simplicity , we assume @xmath8 s to be constant .
this is reasonable for static networks and also in the scenarios where the path loss parameter varies slowly and can be accurately tracked .
the network operates in a time - division multiple access ( tdma ) mode . in each slot , every sensor is allotted a certain amount of time during which it communicates its data to the base station . the general problem is to find the optimal rate ( the number of bits to transmit ) and transmission times for each node , which maximize network lifetime .
both the rate and time allocation are constrained .
the rate allocation should fall within the slepian - wolf achievable rate region and the sum of transmission times should be less than the period of a time - slot ( which is taken to be unity ) . finding the optimal rate allocation is a computationally challenging problem as the slepian - wolf achievable rate region for @xmath1 nodes
is defined by @xmath9 constraints .
we simplify the problem by insisting that decoding at the base - station be instantaneous in the sense that once a particular node has been polled , the data generated at that node is recovered at the base - station before the next node is polled .
this reduces the rate allocation problem to finding the optimal scheduling order , albeit at some loss of optimality .
this loss of optimality occurs because our problem formulation assumes the separation between source and channel coding and it is well - known , @xcite , that the source - channel separation does not hold for the multi - access source - channel coding problem and slepian - wolf coding followed by channel coding is not optimal for the joint source - channel coding problem .
also , in general , turning a multiple - access channel into an array of orthogonal channels by using a suitable mac protocol ( tdma in our case ) is well - known to be a suboptimal strategy , in the sense that the set of rates that are achievable with orthogonal access is strictly contained in the ahlswede - liao capacity region @xcite .
however , despite this fundamental sub - optimality , we argue like @xcite that there are strong economic gains in the deployment of networks based on such technologies , due to the low complexity and cost of existing solutions , as well as available experience in the fabrication and operation of such systems .
let @xmath10 be the set of permutations of the set , @xmath11 .
the polling schedule followed by the network in any time slot corresponds to a permutation , @xmath12 .
let @xmath13 denote the @xmath14 node to be scheduled .
instantaneous decoding implies that the amount of data to be transmitted by node @xmath13 is the conditional entropy of the data source at node @xmath13 , given the data generated by all previously polled nodes .
we denote the amount of information generated by node @xmath13 by @xmath15 .
our aim is to find the scheduling strategy ( scheduling order and transmission time allocation ) that maximizes network lifetime .
in this section , we consider the general channel coding scenario where the transmission energy is the convex decreasing function of the transmission time .
for example , by inverting shannon s channel capacity formula for the awgn channel , it is straight - forward to show that transmission energy is a strictly decreasing convex function of transmission time @xcite .
other channel coding situations lead to a similar result .
in such a scenario , we not only have to find the optimal scheduling order , but also the optimum transmission times for each node .
we consider two kinds of schedules , namely , static and dynamic . in static scheduling
, the nodes follow the same fixed scheduling order in all time slots until the network dies . under dynamic scheduling , we allow nodes to collaborate further by allowing them to employ different schedules in different time slots .
more specifically , it is _ offline _ dynamic scheduling , where before the actual operation of network starts , the base - station has already computed the optimum set of schedules and the number of slots for which each schedule is used , rather than _ online _ dynamic scheduling , where only at the beginning of every polling slot , the base - station computes _ on fly _ the optimum schedule for that time - slot , based on its knowledge of the latest state of the network .
let @xmath16 be the energy required to transmit @xmath17 bits of information in @xmath18 units of time with path loss factor @xmath19 .
so , we can interpret @xmath20 as the energy required to transmit @xmath17 bits of information in @xmath18 units of time with unit path loss .
based on our discussion , we model the energy function @xmath20 as follows . 1 .
@xmath20 is a strictly _ decreasing _ continuous positive convex function in @xmath18 .
2 . @xmath21
unless stated otherwise , we assume @xmath22 to be the one that is obtained by inverting shannon s awgn capacity formula @xcite , that is : @xmath23 in static scheduling , each permutation , @xmath24 corresponds to a tdma schedule .
let @xmath25 denote the number of information bits transmitted per slot by node @xmath2 under the schedule @xmath24 .
let @xmath26 be the corresponding transmission time alloted to node @xmath2 and @xmath27 be the lifetime achievable by the system under the schedule @xmath28 .
note that lifetime is integer - valued but we will treat it as a real number .
the optimal static schedule is the solution to the following optimization problem .
@xmath29 however , using the `` channel aware '' algorithm proposed in @xcite for every schedule @xmath28 , we have the maximum lifetime and the corresponding transmission times allocation vector @xmath30 .
further , for this transmission time allocation vector , all the sensor nodes achieve the same lifetime .
so , the problem in reduces to the following optimization problem : @xmath31 before we analyze the solution of the optimal scheduling problem for the general case , it should be noted that given the cooperative nature of the sensor nodes , the nodes can collaborate with each other to a much greater extent by varying their transmission times .
for example , nearby nodes can finish their transmissions sooner , allowing far away nodes more time to transmit in order to improve system lifetime .
the structure of the problem in or is such that its computational complexity depends on the sensor node data correlation structure as well as the energy function .
the following three examples amply illustrate this .
_ example 1 : _ let us consider the following model for spatial correlation of the sensor data .
let @xmath32 be the random variable representing the sampled sensor reading at node @xmath33 and @xmath34 denote the number of bits that the node @xmath35 has to send to the base - station .
let us assume that each node @xmath35 has at most @xmath36 number of bits to send to the base - station , so @xmath37 .
however , due to the spatial correlation among sensor readings , each sensor may send less than @xmath36 number of bits .
let us define a data - correlation model as follows .
let @xmath38 denote the distance between nodes @xmath35 and @xmath39 .
let us define @xmath40 , the number of bits that the node @xmath35 has to send when the node @xmath39 has already sent its bits to the base - station , as follows : @xmath41 figure [ fig1 ] illustrates this for @xmath42 .
it should be noted that when @xmath43 , the data of the nodes @xmath35 and @xmath39 differ in at most @xmath40 least significant bits .
so , the node @xmath35 has to send , at most , @xmath44 least significant bits of its @xmath36 bit data .
also note the following property of the correlation model : @xmath45 however , the definition of the correlation model is not complete yet and we must give the expression for the number of bits transmitted by a node conditioned on more than one node already having transmitted their bits to the base - station .
there are several ways in which this quantity can be defined .
presently , let us consider the following definition of the conditional information : @xmath46 _ part 1 : _ here let us assume that the ratio of energy of a node and path - loss between base - station and the node is equal for all the nodes .
so , without the loss of generality , for every sensor @xmath2 , we can put @xmath47 .
this assumption is only to simplify the solution , yet it is not such an unrealistic assumption when we consider the scenarios such as one where given the equal energies of all the nodes , the distance of the base - station from any node is much more than the distance between any two nodes .
also , when we have the roving base - station for the data gathering , this assumption holds good .
so , using the energy consumption model of , this makes the time to transmit depend only on the number of bits that a node has to send to the base - station , so a sensor polling schedule that results in larger value of @xmath48 , will also result in the larger value of the sum of transmission times of all the nodes . a greedy scheme that assigns information bits to the nodes according to , gives the solution for the static scheduling problem in in @xmath49 .
start with any node as the first node of the schedule , then choose the next node in the schedule to be that node that minimizes the conditional number of bits .
however , given the definition of correlation model in and , this amounts to finding the nearest node .
so , the schedule that selects the nearest neighbor as the next node to be polled is the optimum schedule .
we call this algorithm : _
nearest neighbor next _ or
_ nnn_. for a desired value of network lifetime , the _ nnn _ schedule will give the smallest value of @xmath48 and the smallest sum of the transmission times , so using the `` channel aware '' algorithm proposed in @xcite , we can prove that this schedule maximizes the network lifetime .
it should be noted that it is the special property of this problem , due to the correlation model and assumption above , that the schedule that minimizes @xmath48 , also minimizes the sum of the transmission times , and subsequently maximizes the network lifetime . in general , it is not true that the schedule that minimizes @xmath48 also minimizes the sum of the transmission times . _
part 2 : _ without the assumption in _ part 1 _ , here we consider the general problem .
the following theorem proves that for the given spatial correlation model , the problem is @xmath0-hard . the static scheduling problem in is @xmath0-hard for the spatial correlation model of .
an arbitrary instance of `` shortest hamiltonian path '' problem can be reduced to this problem by following the same sequence of steps as in the proof in _
example 2_. _ example 2 : _ let us consider the spatial correlation model of @xcite , that is the one where the sensor data is modeled by gaussian random field .
thus , we assume that an @xmath1 dimensional _ jointly gaussian multivariate distribution _
@xmath50 models the spatial data @xmath51 of @xmath1 sensor nodes .
@xmath52 where @xmath53 is the ( positive definite ) covariance matrix of @xmath54 , and @xmath55 the mean vector .
the diagonal entries of @xmath53 are the variances @xmath56 .
the rest of @xmath57 depend on the distance between the nodes @xmath35 and @xmath39 : @xmath58 . without any loss of generality , here we use differential entropy instead of entropy , as we assume that the data at each sensor node is quantized with the same quantization step , and under such assumption , differential entropy differs from entropy by a constant .
[ gc_ss_thrm ] the static scheduling problem in is @xmath0-hard for the spatial correlation model of and energy function @xmath20 .
let us consider the decision version of this problem : does there exist a schedule @xmath28 , for which the network achieves the lifetime @xmath59 , with the following constraints ?
we will prove the @xmath0-hardness of this problem as follows .
we reduce an arbitrary instance of `` shortest hamiltonian path problem '' @xcite over euclidean and complete graph , which is well known to be @xmath0-complete , to some instance of the problem in . for the given instance of shp problem , interpret the edge cost between nodes @xmath35 and @xmath39 , as the spatial distance between the nodes @xmath35 and @xmath39 of our problem .
so , as we visit a node @xmath2 in the shp tour @xmath28 , we can compute the conditional entropy @xmath15 of that node using the knowledge of the model for spatial correlation among the sensor nodes as well as the history of the tour so far . with this computed conditional entropy value ,
using the first constraint of , we can compute the minimum time @xmath26 that this node needs to transmit @xmath15 bits of information to the base - station .
so , for every schedule @xmath28 , we can compute the sum of the minimum transmission times .
let us consider an euclidean , complete graph of @xmath61 nodes @xmath62 $ ] with symmetric edge costs .
let us consider two schedules @xmath63 and @xmath64 .
we have taken @xmath61 to only illustrate the main idea of the reduction , otherwise the approach is general enough to be applicable to the bigger networks .
let us assumes that the length of hamiltonian path @xmath65 for schedule @xmath63 is less than @xmath66 for the schedule @xmath64 .
now we are going to prove that the sum of transmission times for the schedule @xmath63 is less than that for the schedule @xmath64 .
for the spatial correlation model of interest , for every schedule , we can compute the conditional entropies of every node based on all the nodes visited previously @xcite .
for example , if the schedule is @xmath67 $ ] and @xmath68 denote their data samples , respectively , then : @xmath69 where @xmath70 and @xmath71 , denote the covariance matrices of @xmath68 and @xmath72 , respectively . assume that the transmission time @xmath73 of node @xmath74 $ ] is exponentially dependent on the entropy @xmath75 of the node ( this follows from the empirical results obtained after numerically solving the equation ) .
let us denote the transmission times of the nodes a , b , c , and d under schedule @xmath63 as @xmath76 and @xmath77 respectively . similarly , for the schedule @xmath64 , let the corresponding times be @xmath78 and @xmath79 respectively .
note that @xmath80 and @xmath81 .
now @xmath82 after substituting the values of @xmath83 , and @xmath79 and a little algebraic manipulation of the resulting expressions , we show that is true if ( with @xmath84 and @xmath85 as defined in @xcite ) : @xmath86 so , if a schedule has smaller hamiltonian path length , then the corresponding sum of the transmission times will be smaller too .
this implies that the solution of shp gives the smallest value of the sum of the transmission times .
so , for the schedule that gives shortest hamiltonian path , we can compute the sum of the transmission times and if this sum is less than @xmath87 , then we have at least one schedule that achieves the lifetime @xmath59 . in this section ,
we explore how network lifetime can be increased by employing multiple schedules .
instead of restricting the network to follow a single schedule , we allow the system to employ different schedules over time .
there are @xmath88 possible schedules to choose from .
let @xmath15 be the number of information bits generated per slot by node @xmath2 under the schedule @xmath28 , @xmath89 .
two or more schedules can collaborate by having the nodes use non - optimal transmit energies over two or more data - transmission slots to increase the lifetime of the network .
we have a total of @xmath88 schedules .
if only @xmath90 schedules are going to cooperate , then there are @xmath91 possible combination of the schedules .
let @xmath92 denote the number of time slots for which schedule @xmath93 is employed . the optimal network lifetime @xmath59 under dynamic scheduling is the solution to the following optimization problem @xmath94 \\ \pi_1 , \ldots , \pi_m \in \pi}}\sum\limits_{i = 1}^m \tau_{\pi_i } & \\ & \mbox{s
. t. } \sum\limits_{i=1}^m f(h_{\pi_i(k ) } , t_{\pi_i(k)})d_k \tau_{\pi_i } \le e_k , \forall 1 \le k \le n & \nonumber \\ & \sum\limits_{k=1}^{n } t_{\pi_i(k ) } = 1 , \forall 1 \le i \le m & \nonumber\end{aligned}\ ] ] specifically for @xmath95 , we have @xmath96 also note that if @xmath97 , then the problem in reduces to the static scheduling problem in .
so , the computational complexity of this problem can not be any less than that of the static scheduling problem , which is proven to be @xmath0-hard in theorem [ gc_ss_thrm ] . here
the question we are concerned with is that if the dynamic scheduling can indeed increase the network lifetime . in the following , we prove that even for the simplest case of the network of two nodes , it is indeed so .
[ gc_ds_thrm ] for @xmath98 , dynamic scheduling performs better than the optimal static scheduling . for the network of two nodes , let us consider two schedules @xmath99 where node @xmath87 is polled before node @xmath100 and @xmath101 , where the nodes are polled otherwise . now using our `` channel aware '' algorithm of the previous section , for a given polling schedule we can find the optimal allocation of the transmission times such that both the nodes spend same amount of energy , dying at the same time
let for schedule @xmath99 , this happens when the node 1 transmits for @xmath102 units of time and node 2 transmits for @xmath103 units of time .
similarly , for schedule @xmath101 , let the corresponding times be @xmath104 and @xmath105 .
let @xmath17 denote the entropy of first node polled in the schedule and @xmath106 denote the entropy of second node polled .
so , for schedule @xmath99 : @xmath107 and for schedule @xmath101 : @xmath108 . given the optimality of @xmath102 and @xmath104 for the schedules @xmath99 and @xmath101 respectively , we have for the energy consumptions of the nodes : @xmath109 if we assume the schedule @xmath99 to be the optimum static schedule , then the following holds true : @xmath110 let us consider the plot where the horizontal axis corresponds to the energy consumption @xmath111 of the node 1 and the vertical axis corresponds to the energy consumption @xmath112 of the node 2 . in this plot , we draw the energy consumption curves for both the schedules @xmath99 and @xmath101 , for different values of @xmath113 and @xmath114 respectively .
given the form of the energy consumption curves , it is easy to verify that these two curves corresponding to two different schedules , are convex and will intersect at one and only one point .
let us consider the `` equal energy line '' which passes through the pair of points @xmath115 and @xmath116 , so the equation of this lines is @xmath117 now let us also consider a line that passes through the point @xmath118 on the curve corresponding to the schedule @xmath99 with @xmath119 , and the point @xmath120 with @xmath121 on the curve for schedule @xmath101 .
the equation for such a line is @xmath122 now let us consider the point of intersection of these two lines .
at the point of intersection , we have : @xmath123 now , if we want to prove that with the dynamic scheduling we can perform better than the static scheduling , then we must prove that there exists at least one pair of values @xmath124 , for which the following holds @xmath125 substituting the expressions of @xmath126 and @xmath127 from in , and using the properties of the energy consumption function , we prove that the dynamic scheduling performs better than static scheduling for all @xmath128 such that @xmath129 this result implies that two schedule can cooperate to give longer network lifetime compared to optimum static schedule .
following subsection , generalizes this result .
let us consider the scenario where we have @xmath1 sensor nodes to poll and this polling can be done in @xmath88 ways .
let us consider an @xmath1 dimensional euclidean space , where an axis corresponds to the energy consumption of a node .
given that the energy consumption can only assume positive real values , we are only concerned with the first orthant of this @xmath1-dimensional space . for any given schedule , as the transmission time allocation to the different nodes changes , the corresponding energy consumption of the nodes changes .
so , the point defining the energy consumption of the nodes in this @xmath1 dimensional space describes an @xmath1-dimensional convex hypersurface . for @xmath88 possible schedules
, we have @xmath88 such hypersurfaces .
the computational complexity of the problem of finding the optimum static and dynamic schedules depends on the properties of the intersections of these hypersurfaces .
also , note that the general shape of these surfaces is determined by the energy consumption function and the model of the spatial correlation in the sensor data .
the optimal static schedule is the one whose energy hypersurface is intersected by the `` equal energy line '' closest to the origin .
further , the dynamic scheduling helps us achieve all the points on the convex hull @xmath130 of @xmath88 convex hypersurfaces , as those are all the points achievable with the cooperation of any number of schedules .
it is obvious that the network lifetime can not be increased by the cooperation of those schedules and their transmission time allocations , which give any point in the interior of this convex - hull , as there is always a point on the surface of the convex - hull that is closer to the origin and gives better network lifetime .
so , when two schedules cooperate , then the optimum transmission time allocation is the one that gives the line connecting the two points on the surface of the convex hull
. then the optimum network lifetime is achieved by some point on that line , specifically by the point on this line that is closest to the origin .
similarly , when three schedules cooperate , then the respective optimum transmission times allocation for the those three schedules is the one that gives the plane connecting the energy consumption points corresponding to the three schedules , on the surface of the convex hull and then the optimum network lifetime is achieved by some point on this plane . formally , if @xmath131 are the @xmath132 cooperating schedules , then the plane defined by the @xmath132 points on the @xmath132 corresponding hypersurfaces , that is closest to the origin , must belong to the convex hull @xmath130 .
for these @xmath132 schedules , the optimal lifetime of the network is obtained by that point on this plane that is closest to the origin . this is point
is obtained as the solution to the following optimization problem : @xmath133 with these optimum values of the parameters @xmath134 , we solve for the network lifetime as follows : @xmath135 the optimum network lifetime @xmath136 for the set of @xmath132 schedules , is obtained by solving above set of equations for all @xmath137 possible combinations of @xmath132 schedules , that is @xmath136 is obtained by : @xmath138 \\ \pi_1 , \ldots , \pi_m \in \pi } } l_{\pi_1\ldots\pi_m}\ ] ] further , to obtain the optimum value of network lifetime over all possible combinations of the schedules is obtained by : @xmath139 it should be noted that the equations - , essentially solve .
however , this alternative formulation of the problem in , helps us to prove following two important theorems : [ geothrm1 ] the optimum network lifetime for @xmath132 schedule cooperation is no worse than the optimum network lifetime for @xmath140 schedule cooperation .
that is @xmath141
. omitted for brevity .
[ geothrm2 ] when @xmath142 , then the cooperation among @xmath132 or more schedules does not improve the network lifetime anymore .
omitted for brevity .
in this section , we assume that transmission rate is linearly proportional to signal power . this assumption is motivated by shannon s awgn capacity formula which is approximately linear for low data rates .
the energy expended by a node to transmit @xmath143 units of information is given by @xmath144 , where @xmath19 is the suitably normalized path loss factor between the node and the base station .
the linear rate assumption implies , as shown below , that transmit energy is independent of transmission time .
hence , the optimal time allocation problem is trivial and we only need to find the optimal scheduling order . for the small data rates , the energy consumption function for the @xmath14 node under schedule @xmath28 reduces to @xmath145 . for example ,
by inverting shannon s awgn channel capacity formula , we get the following as the energy consumption function @xcite : @xmath146 for the small data rates , this gives for some constant @xmath147 @xmath148 under the `` small rate region approximation '' , the static scheduling problem in reduces to : @xmath149 the objective function represents the lifetime of node @xmath13 under the given static schedule @xmath28 . in @xcite , we describe a greedy static scheduling strategy , minimum cost next ( mcn ) , and prove its optimality .
the mcn schedule not only maximizes the minimum lifetime , but also maximizes all lifetimes from @xmath150 minimum lifetime to @xmath151 minimum lifetime .
this is desirable in the situations , where the network has to continue to operate even when one or more nodes die out .
also , note that the mcn solution is pareto - optimal . given an mcn schedule ,
no other schedule can help increase any node s lifetime without decreasing some other node s lifetime . in this section ,
we explore how network lifetime can be increased by employing multiple schedules under the small data rate approximation . under this assumption , as the general static scheduling problem in reduced to , the general dynamic scheduling problem in reduces to the following optimization problem that gives the optimal lifetime @xmath59
@xmath152 @xmath153 is the number of slots for which the schedule @xmath28 is used .
once more , the constraints ensure that the time assignment is feasible for each node with respect to its energy capability . also as in [ gc_ds ] ,
can be treated as a linear program . a dynamic schedule , @xmath154 , is given by the set @xmath155 . given @xmath88 variables , in general , there seems to be no easy way to solve to compute the optimal @xmath153 values .
however , in @xcite , we exploit the special nature of our problem to propose an algorithm , which we refer to as lifetime optimal clustering algorithm ( local ) , and prove its optimality . in this section
we discuss both the static and dynamic scheduling in `` small rate region approximation '' in the spirit of discussion of the nature of the solutions in [ geo_inter ] . under this approximation ,
the energy consumption is independent of the transmission time , so the flexibility to change the energy consumption by varying the transmission time is not available .
so , in the @xmath1-dimensional space , where each axis corresponds to the energy consumption of a node , for every schedule , we get a point in this space , rather than a hypersurface . for @xmath88 possible polling schedules ,
we get @xmath88 points . the optimal static schedule under this approximation
is given the mcn algorithm .
given the nature of the problem in , this corresponds to that schedule which gives a point closest to the `` equal energy line '' .
this is not difficult to see if one notes that optimal static schedule attempts to ` equalize ' the lifetimes / energy consumption of the nodes .
however , as noted above , the flexibility of varying the transmission times of the nodes to ` equalize ' their lifetimes is not available under this approximation , so the optimum static schedule does the best in providing a point closest to this `` equal energy line '' , if not at the line itself .
note that the point corresponding to the optimum static schedule may not be the closest to the origin .
similarly , the optimal dynamic scheduling under `` small rate region approximation '' corresponds to finding those schedules , the lines , planes , or the hyperplanes connecting which contain the point closest to the `` equal energy line '' .
when more than one such points are possible , then one that is closest to the origin is taken to be the point of operation of the network .
in this paper , we have considered the problem of maximizing the lifetime of a data gathering wireless network .
our contribution differs from previous research in two respects .
firstly , we proposed a combined source - channel coding framework to mitigate the energy cost of radio transmission . secondly , we have explicitly maximized network lifetime as opposed to other objective functions such as cumulative energy cost . to the best of our knowledge , both
these aspects have not been explored in the context of sensor networks previously . in our system model , nodes communicate directly to a base station in a time division multiplexed manner . with our notion of instantaneous decoding , we show that the network lifetime maximization problem reduces to finding an optimal scheduling strategy ( polling order and transmission time allocation ) .
we considered the general channel problem and proved that there the optimal static and dynamic scheduling problems are @xmath0-hard and the optimal dynamic scheduling strategy indeed does better than the optimal static scheduling strategy .
then we considered the scenario where the energy consumption is independent of transmission time . for both , the general channel problem and its approximated version , we provided the geometric interpretation of the optimal solutions and the solution search processes .
this paper assumed that source and channel coding is optimal , quantization is perfect and that a continuum of power levels can be employed .
network lifetime obtained under these assumptions is an upper limit to practically achievable performance
. it would be useful to consider the network lifetime problem with more realistic constraints .
finally , the system model has to be generalized to the multi - hop case . | we consider a single - hop data gathering sensor cluster consisting of a set of sensor nodes that need to transmit data periodically to a base - station .
we are interested in maximizing the lifetime of this network . even though the setting of our problem is very simple , it turns out that the solution is far from easy .
the complexity arises from several competing system - level opportunities that can be harnessed to reduce the energy consumed in radio transmission .
first , sensor data in a cluster is spatially and temporally correlated .
recent advances in distributed source coding allow us to take advantage of these correlations to reduce the number of bits that need to be transmitted , with concomitant savings in energy .
second , it is also well known that channel coding can be used to reduce transmission energy by increasing transmission time .
finally , sensor nodes are cooperative , unlike nodes in an ad hoc network that are often modeled as competitive .
this collaborative nature allows us to take full advantage of the first two opportunities for the purpose of maximizing cluster lifetime . in this paper
, we pose the problem of maximizing lifetime as a max - min optimization problem subject to the constraint of successful data collection and limited energy supply at each node .
this turns out to be an extremely difficult optimization to solve .
consequently , we employ a notion of instantaneous decoding to shrink the problem space .
we show that the computational complexity of our model is determined by the relationship between energy consumption and transmission rate as well as model assumptions about path loss and initial energy reserves .
we provide some algorithms , heuristics and insights for several scenarios . in some situations ,
our problem admits a greedy solution while in others , the problem is shown to be @xmath0-hard .
the chief contribution of the paper is to illustrate both the challenges and gains provided by source - channel coding and scheduling . |
hierarchical cold dark matter ( cdm ) cosmology has been largely successful in establishing the paradigm of structure formation and evolution on spatial scales of @xmath8 kpc , corresponding to individual halos , and beyond ( e.g. , cole et al . 2000 ; steinmetz & navarro 2002 ; spergel et al .
however , on smaller scales , a number of fundamental issues related to the origin of luminous galactic components and their structural parameters remain unresolved because of our limited understanding of dynamical processes , such as mass and angular momentum redistribution , star formation ( sf ) and the energy and momentum feedback from the stellar evolution onto the interstellar medium ( ism ) .
specifically , no concensus exists at present on formation of galactic disks , bulges , bars and supermassive black holes ( sbhs , e.g. , white & rees 1978 ; fall & efstathiou 1980 ; white 1996 ; ellis 2000 ) . because these structures are made from the baryonic matter
, either gas or stars or their mixture , a certain degree of dissipation must be involved .
detection of quasars at redshifts @xmath9 ( e.g. , fan et al .
2001 ; mathur , wilkes & gosh 2002 ; bertoldi et al .
2003 ) , when the age of the universe was less than a gigayear , rises intriguing questions about the origin of their host galaxies within the framework of the hierarchical clustering , and is important for the fate of the sbh bulge correlation , i.e. , @xmath10 , and the masses of the quasar host galaxies , at these @xmath11 ( e.g. , trenti & stiavelli 2007 ) . in this work
we investigate how various dissipative processes , including the sf and the feedback from stellar evolution onto the ism , affect the galactic disk and bulge formation and evolution . for this purpose
we follow a monolithic collapse of an initial perturbation in the dm and baryons within the hubble flow .
when exactly did the disks and other structural elements of disk galaxies , like bulges , bars and central sbhs , assemble ?
how much evolution has occured since ? both orbital and ground - based telescopes have began to tighten the constraints on these issues , but essential details still missing and controversies are abound .
first , within the hierarchical merging scenario , the so - called ` angular momentum catastrophe ' the angular momentum ( @xmath0 ) distribution of the collapsing baryonic matter does not match that of the observed disks in nearby universe ( e.g. , sommer - larsen , gotz & portinari 2002 ; maller & dekel 2002 ; van den bosch 2002b ; burkert & donghia 2004 ) .
this distribution is characterized by low and high @xmath0 tails which result both in the over - extended disks and very massive bulges .
in addition , gravitational torques between the collapsing baryons and the dm halos lead to disks with radial scalelengths substantially smaller than those observed . under the assumption of @xmath0 conservation during the collapse ,
the specific @xmath0 generated by the random work in the expanding universe is too small to explain the formation of bulgeless disks in the late type galaxies ( van den bosch , burkert & swaters 2001 ; van den bosch 2002a ) .
the second issue is that of the ` core catastrophe ' ( e.g. , moore et al .
1999 ) and relates to the dm central density distribution .
the numerical simulations of dm halo formation have shown that these attain a universal density profile which can be nicely fitted by a two - parameter family ( navarro et al .
1996 , hereafter nfw ) .
however , observations of disk galaxies and galaxy clusters indicate that the dm halos have flat density cores rather than nfw cusps ( e.g. , flores & primack 1994 ; moore 1994 ; burkert 1995 ; kravtsov et al . 1998 ; salucci & burkert 2000 ; boriello & salucci 2001 ; de blok & bosma 2002 ; sand et al .
the standard picture of slow adiabatic collapse of a smooth baryonic component in the centers of such halos can only lead to a larger central concentration , making matters worse .
el - zant , shlosman & hoffman ( 2001 ) and el - zant et al .
( 2004 ) have shown that a flattening of the central density cusps can be a natural outcome of energy deposition by the accreted clumpy baryons via a dynamical friction both in galaxies and in clusters of galaxies .
alternatively , weinberg & katz ( 2002 ) and holley - bockelmann , weinberg , katz ( 2005 ) have claimed that the cusps can be leveled through the action of the stellar bars , but this was strongly disputed by sellwood ( 2003 ) and mcmillan & dehnen ( 2005 ) .
third , abraham et al . (
1999 ) and van den bergh et al .
( 2002 ) have claimed that the fraction of stellar bars sharply decreases above the redshifts of @xmath12 , while the analysis of gems survey ( rix et al .
2004 ) of 1600 galaxies has shown that both the fraction of bars and their size and axial ratio distributions remain largely unchanged up to @xmath13 at least ( jogee et al . 2004 ) .
this latter evidence is supported by elmegreen et al .
( 2004 ) and sheth et al .
( 2003 ) based on much smaller samples of 186 and 4 galaxies respectively . finally , the dm halos resulting from cosmological simulations appear to be strongly triaxial , both in their prolateness and flatness , , where @xmath14 is the intermediate - to - major axis ratio and the flatness as @xmath15 with @xmath16 the minor - to - major axis ratio in the dm halo ] @xmath17 and @xmath18 ( e.g. , allgood et al . 2006 ) . on the other hand ,
halo shapes inferred from observations in the local universe are axisymmetric , although individual objects may exhibit some mild degree of prolateness ( e.g. , rix & zaritsky 1995 ; merrifield 2002 ) .
theoretical arguments supplemented by numerical simulations have shown that bars are incompatible with prolate halos ( el - zant & shlosman 2002 ; berentzen , shlosman & jogee 2006 ) . when taken in tandem with the plethora of stellar bars observed locally ( e.g. , marinova & jogee 2007 ; menendez - delmestre et al .
2007 ; knapen , shlosman & peletier 2000 ) and at the intermediate redshifts of @xmath19 , corresponding to the lookback time of 8 gyr ( jogee et al . 2004 ) , these results indicate strongly that the halo shapes have evolved only mildly during this time period . furthermore , dubinski ( 1994 ) and kazantzidis et al .
( 2004 ) have shown that the halos lose some of their triaxiality when baryons are added . in the next step , berentzen & shlosman ( 2006 )
have simulated the disk growth in an assembling dm halo within the @xmath20 mpc comoving box .
the initial conditions have been produced by means of the constrained realizations algorithm of hoffman & ribak ( 1991 ) .
the halo prolateness was found to be washed out within the disk radius , and the halo flatness to decrease substantially .
the strongly nonlinear process of galaxy formation coupled with a long list of dissipative processes provide a strong insentive for a numerical approach .
a number of numerical methods to follow up the disk formation have been attempted . in the models of immeli
2004 ) , when the gas in the proto - disk cools efficiently , the disk fragments and forms a number of massive clumps of stars and gas .
these clumps spiral in to the galactic center and merge there to produce a strong starburst , as also shown by shlosman & noguchi ( 1993 ) , leaving a central object which resembles a classical bulge .
the disk growth was simulated in a fixed spherically - symmetric external potential of the dm halo by adding the gas directly to the disk .
alternatively , samland & gerhard ( 2003 ) have modeled a disk and the associated dm halo growth by adopting the mass inflow rate from _ separate _ cosmological simulations , while preserving the spherical symmetry of the halo , using a chemo - dynamical code .
a number of issues have been investigated , among them the stellar population of the bulge .
a possibility of forming bars from baryonic spheroids immersed in triaxial dm halos was considered by gadotti & de souza ( 2003 ; see also el - zant et al .
2003 ) .
sommer - larsen , gtz & portinari ( 2003 ) analyzed the formation history of two galaxies with circular velocities comparable to the milky way ( mw ) and found gas accretion rates , and hence x - ray halo luminosities , at @xmath13 , 67 times larger than at @xmath6 for these objects .
more generally , it is found that the gas infall rates onto these disks are nearly exponentially declining with time , both for the total disk and for the `` solar cylinder . ''
this result supports the exponentially declining gas infall approximation often used in the chemical evolution models .
okamoto et al .
( 2005 ) has investigated the effect of the sf feedback on the morphological evolution of galaxies by assuming that the sf can proceed in two modes either in the high density or shock - triggered fashion .
the former mode leads to elliptical galaxies while the latter to disk ones .
the authors made a number of assumptions that require a further validation , e.g. , the sph kernel was fixed at the constant mass and not on the constant number of particles .
the simulations of governato et al .
( 2004 , 2007 ) form rotationally - supported disks with realistic exponential scalelengths and fit both the @xmath21 band and baryonic tully - fisher relations ( see also stinson et al .
2006 ; kaufmann et al .
an extended stellar disk forms inside the mw - sized halo immediately after the first merger .
the combination of the uv background and supernova ( sn ) feedback drastically reduces the number of visible satellites orbiting within this halo , bringing it in a fair agreement with observations .
formation of stellar populations in the mw - type galaxy were studied by scannapieco et al .
( 2006 ) using a combination of high - resolution @xmath22-body simulations with semi - analytical methods of metal enrichment .
substantial constraints on the formation of population iii stars have been obtained .
brook et al .
( 2007b ) have followed up with chemodynamical simulations , including the sf , concluding that surveys of low - metallicity stars in the galactic halo can directly constrain the properties of population iii stars . furthermore , the disk evolution was simulated by brook et al . ( 2007a ) , examining the possibility that high @xmath23 elements - to - iron abundance ratios detected in the thick - disk component of the milky way can be explained by sf during the gas - rich mergers .
all the above studies have brought a very useful insight into the formation of disk galaxies and have contributed to a better understanding of specific aspects of the problem . yet , as already mentioned , the problem is very complex and much more remains to be investigated , in particular regarding the dynamics of forming disks , the interaction between the various morphological components and subcomponents , as well as their origin and properties .
the present paper aims to advance of some of these open questions .
our emphasis is more on the dynamical aspects and includes discussions of the formation of bars , both inner and outer , made in such _ ab initio _ simulations .
our algorithm includes the sf , the feedback from the stellar evolution , etc .
we also performed a relatively large number of simulations , so as to gain insight on how the sf and energy feedback affect the structural properties of the galaxy components and subcomponents .
we use cosmological initial conditions and follow the hubble expansion and the subsequent collapse of a single perturbation in the gas and the dm in order to analyze the role of different processes in the formation of disk and bulge components .
we aim at understanding the role of the external cosmological and internal dissipative factors .
section 2 describes the details of our method of modeling the sf , our approach in determinig the shapes of spheroidal and disk components , and the initial conditions used for numerical simulations .
sections 3 and 4 present results of our simulations and model comparison .
discussion follows in the last section .
the numerical simulations have been performed with an updated version of the ftm-4.4 hybrid code which evolves the collisionless ( @xmath22-body ) and gaseous ( smooth particle hydrodynamics , sph ) components ( heller & shlosman 1994 ; heller 1995 ) , with @xmath24 and @xmath25 .
in addition , a number of models have been evolved using the parallel version , ftm-4.5 , of the code .
the gravitational forces were computed using the public version of the routine falcon ( dehnen 2002 ) , which is about ten times faster than optimally coded barnes & hut ( 1986 ) tree code .
the tolerance parameter @xmath26 is fixed at 0.55 .
the energy equation describing the internal state of the sph particles have been used to evolve a multiphase ism by calculating its heating and cooling rates and iterating for the temperature .
star formation was used to convert sph particles into collisionless particles which exert an energy feedback onto the surrounding gas .
the gravitational softening for the collisionless dm and stellar particles is @xmath27 pc , except for two models specified in table 1 . for the sph particles we use two options to soften the gravity
constant softening of @xmath28 pc or 250 pc , and a dynamic softening ( heller & shlosman 1994 ) with the minimum of @xmath29 pc .
a large number of tests has been performed to check the sensitivity of the results to the algorithm and its parameters .
a typical run with the dm only conserved the energy to within 1% and @xmath0 to within 0.1% .
we anticipate that the sf processes and the associated feedback from stellar evolution coupled with an overall dissipative and collisionless dynamics within the forming galaxy determine its evolutionary path .
strictly speaking , because the smallest spatial scales resolved here are substantially larger than scales which are relevant for an ` individual ' sf event , the underlying collapse should be treated witin the context of the two - fluid jeans instability ( zeldovich & novikov 1975 ) in a sheared medium .
we adopt the modified prescription for the sf from heller & shlosman ( 1994 ) that required the gas to be jeans unstable and subject to the feedback from ob stellar winds and supernovae ( sn ) , and introduced a number of statistical elements .
our main assumption is that the sf takes place in regions which are contracting , i.e. , @xmath30 and are jeans unstable , i.e. , @xmath31 which assures that the collapse would continue unhindered down to the smallest resolved spatial scales . under this condition
, the pressure gradients ( which are established over the sound crossing time @xmath32 ) will not be able to build up during the collapse time @xmath33 .
moreover , if the gas cooling time satisfies @xmath34 , this will ease the condition given by eq .
( 2 ) , which , for simplicity , can be replaced by @xmath35 for temperatures in excess of @xmath36 k ( e.g. , navarro & white 1993 ) .
this happens because the cooling function @xmath37 has a minimum at @xmath38 k. for larger densities than @xmath39 , the short cooling time @xmath40 will be guaranteed if the neutral ( cold ) gas will be at least moderately self - gravitating in the background density @xmath41 , @xmath42 here @xmath43 corresponds for the neutral gas ( i.e. , hi and hei ) , and @xmath44 combines the mass of the ionized ( hot ) gas , stars and dm , with @xmath45 being the fudge factor . for the standard model , @xmath46 ( see table 1 ) .
gas is converted into stars at a rate @xmath47 where the sf ( collapse ) timescale is taken as @xmath48 , with @xmath49 . in reality , the collapse time for molecular clouds is extended beyond the free - fall time , e.g. , because of an additional support from the mhd turbulence .
we have used @xmath50 for the standard model .
next , we introduce the probability that a gas particle of mass @xmath51 produces a stellar particle of mass @xmath52 during a given timestep of length @xmath53 of @xmath54,\ ] ] where @xmath55 and @xmath56 . here
@xmath57 represents the stellar mass which is produced in the sf event .
however , a stellar particle with a mass of only @xmath58 is created as @xmath59 of the mass is instantaneously recycled back to the parent gas particle , along with an increment in its metallicity by @xmath60 , where @xmath61 is the metals yield ( leitherer , robert & drissen 1992 ) .
the quantity @xmath62 is not allowed to be less than unity , in which case we set @xmath63 and @xmath64 , and the gas particle is removed from the simulation .
each gas particle is capable of creating four generations of stars with different metallicities and @xmath51 decreases accordingly after each sf event .
the mass of the stellar particle represents a cluster of stars .
the fraction and timing of massive stars that produce the ob line - driven winds and sn are calculated assuming the salpeter imf .
the thermal balance in the interstellar gas is calculated using the energy equation .
sources of heating and cooling include adiabatic , viscous and radiative processes . the balance of the specific internal energy along with the gas ionization fractions ( h , h+ , he , he+ , he++ , e ) and the mean molecular weight are computed as a function of density and temperature for an assumed optically thin primordial composition gas ( katz et al .
this allows for the computation of the neutral gas density @xmath43 required for the sf criteria of eqs .
( 13 ) . to model the energy feedback from the stellar evolution onto the gas , the newly formed stellar particles inject energy from sn and ob stellar winds into the @xmath65 surrounding gas particles .
the total rate of energy deposition by stars , @xmath66 , is given by , @xmath67 for @xmath68 , and @xmath69 for @xmath70 . here
@xmath7 is an energy thermalization parameter , @xmath71 is the age of a stellar particle and @xmath72 are the energy deposition rates per solar mass by massive stars and sn , respectively .
the metallicity of a stellar particle is given by @xmath73 in units of the solar metallicity ( maeder 1992 , 1993 ) .
the timestep of such ` active ' stellar particles ( and of all the gas particles ) is restricted to being smaller than @xmath74yrs in order to properly resolve the feedback timescale of @xmath75yrs .
the @xmath76 gas particles in the vicinity of the ` active ' stellar particles are treated in the following way .
their radiative cooling is temporarily disabled when receiving the energy from a stellar particle and when the condition in eq .
( 2 ) is still fullfilled .
the energy @xmath77 is thermalized and deposited in the gas in the form of a thermal energy , then converted to kinetic energy through the equations of motion .
this method is preferable over injecting a fraction of the stellar energy directly in the form of a kinetic energy this approach is ambiguous and the results are sensitive to the value of a timestep and to the particle number on the receiving end . for a comparison
, we have also run some models with an isothermal equation of state ( eos ) , implementing the feedback by directly over - pressuring the gas through an increase of the local sound speed over the isothermal value .
specifically the sound speed , @xmath78 , is increased by , @xmath79 to determine the intrinsic shape of the dm and stellar particle distributions , we remove any residual net velocity from the system , reject any unbound particles and iterate this procedure to locate the density center ( e.g. , aguilar & merritt 1990 ) , @xmath80 where @xmath81 is the local mass density at the position of each particle . in the reference frame defined by the so determined velocity centroid and density center , we compute the eigenvalues of the moment of inertia tensor for the mass within a specified radius . from these
we may determine the axes @xmath82 of a uniform spheroid with the same eigenvalues .
the ratios of these axes may then be used to characterize the shape of the system , as defined by @xmath83 @xmath84 where @xmath85 for the eigenvalues @xmath86 .
the initial conditions are those of a spherically - symmetric density enhancement in the einstein - de sitter universe .
we use an open cdm ( ocdm ) model with @xmath87 , @xmath88 , where @xmath89 is the current cosmological density parameter and @xmath90 is the hubble constant normalized by 100 @xmath91 .
models with @xmath87 and @xmath92 in @xmath93cdm universe have been run as well , but will be discussed elsewhere . because the collapse time of our dm halos happens at @xmath94 ( table 1 ) , the differences between the ocdm and @xmath95cdm models are insignificant .
we have also rerun our standard model n3 with the wmap3 baryon fraction of 17% ( model n41 ) and ( separately ) with 5 times larger number of the sph and collisionless particles ( model n42 ) the differences with the original model appear to be only quantitative .
lcccccccl * n3 * & 2 & 0.3 & 10 & 0.5 & 0.015 & 0.015 & & standard model : dm halo@xmath96baryons + * qn3 * & 2 & & & & 0.015 & & & as n3 but dm only + * n4 * & 2 & 0.3 & 10 & 0.5 & 0.015 & 0.015 & & enforced axisymmetric halo + * n5 * & 2 & 0.3 & 10 & 0.5 & 0.015 & 0.015 & isothermal & + * n6 * & 2 & 0.05 & 10 & 0.25 & 0.015 & 0.015 & & + * n8 * & 2 & 0.1 & 10 & 0.25 & 0.015 & 0.015 & & + * n14 * & 2 & 0.3 & 1 & 0.5 & 0.015 & 0.015 & & + * n15 * & 2 & 0.1 & 1 & 0.5 & 0.015 & 0.015 & & + * n16 * & 2 & 0.1 & 1 & 0.25 & 0.015 & 0.015 & & + * n17 * & 2 & 0.1 & 10 & 0.5 & 0.015 & 0.015 & & + * n18 * & 2 & 0.3 & 10 & 0.5 & 0.025 & 0.025 & & + * n19 * & 2 & 0.1 & 1 & 0.5 & 0.025 & 0.025 & & + * n20 * & 2 & 0.1 & 1 & 0.5 & 0.015 & 0.025 & & as n19 , with @xmath97 + * n21 * & 2 & 0.1 & 1 & 0.5 & 0.015 & 0.025 & & as n19 , with dynamic softening + & & & & & & & & in the gas + * n22 * & 2 & 0.1 & 1 & 0.5 & 0.015 & 0.025 & isothermal & as n21 + * n23 * & 2 & 0.01 & 1 & 0.5 & 0.015 & 0.025 & & as n21 , with @xmath98 + * n24 * & 2 & 0.01 & 1 & 0.5 & 0.015 & 0.025 & isothermal & as n22 , with @xmath99 + * n25 * & 2 & 0.05 & 1 & 0.5 & 0.015 & 0.025 & & as n21 , with @xmath100 + * n26 * & 2 & 0.05 & 1 & 0.5 & 0.015 & 0.025 & isothermal & as n25 + * n27 * & 2 & 0.3 & 1 & 0.05 & 0.015 & 0.015 & & as n14 , with @xmath101 + * n28 * & 2 & 0.3 & 10 & 0.5 & 0.015 & 0.050 & & as n3 + * n41 * & 2 & 0.3 & 10 & 0.5 & 0.015 & 0.015 & & as n3 , with wmap3 baryon + & & & & & & & & fraction + * n42 * & 2 & 0.3 & 10 & 0.5 & 0.015 & 0.015 & & as n3 , with 5 times more + & & & & & & & & baryon and dm particles + [ table : models ] the initial density profile corresponds to the average density around a @xmath102 peak in a gaussian random density field with power spectrum @xmath103 .
the simulations start at @xmath104 to ensure they begin in the linear regime .
the particles are initially moving outward with the hubble flow , with consecutive shells of mass stopping , turning around and falling back inward .
the model is parameterized by the filter mass @xmath105 , which defines the mass contained within a sphere of radius @xmath106 ( e.g. , thoul & weinberg 1995 ) .
the parameters @xmath105 and @xmath107 are determined by choosing the collapse redshift @xmath108 at which the shell @xmath107 reaches the center in a non - dissipative collapse and a circular velocity @xmath109 associated with the shell surrounding the characteristic filter mass @xmath105 if virialized at half its turn - around radius .
in addition , the model is characterized by the virial coefficient @xmath110 , where @xmath111 is the kinetic energy in random motions and @xmath112 is the potential energy , and a spin parameter @xmath113 , where @xmath0 , @xmath114 , and @xmath115 are respectively the total angular momentum , energy , and mass of the system . the initial angular velocity is taken as @xmath116 , where @xmath117 is the cylindrical radius and the central kpc has been softened .
the angular velocity has been normalized by the requirement that the initial @xmath118 for all models .
this distribution of the specific angular momentum complies with the universal profile of angular momentum in dm halos , based on the statistical sample of halos drawn from high - resolution numerical simulations ( bullock et al .
2001 ; see also fall & efstathiou 1980 ) .
for all the models we adopt the values of @xmath119 , @xmath120 and @xmath121 .
in addition , we have run models with @xmath122 and 6 , but those are not shown here .
the initial models consists of collisionless dm particles with the total mass of @xmath123 and of the gas , given by the sph particles , which have the same radial profiles as the dm . the initial gas comprises 10% of the total mass .
our results are presented in this and next sections .
we describe the evolution starting with qn3 , the pure dm model , and continue with models which include all components , as well as the sf and the feedback from stellar evolution the standard model n3 is followed by models from table 1 .
each subsection emphasizes a different aspect of evolution .
we provide a more specific comparison between models , based on parameter variation , in section 4 .
note that the specific times characterizing the model evolution are given for comparison only .
evolution of pure dm halos within the framework of a monolithic collapse is limited to a virialization and incomplete violent relaxation in the varying background gravitational field .
we find that the angular momentum @xmath0 incorporated in the initial conditions is essentially channeled into the internal circulation . in all models , including those with baryons , the halo forms a prolate triaxial figure and its minor axis is directed along the @xmath0-axis .
the halo _ figure _ tumbles with a pattern speed of @xmath124 during the collapse period of @xmath125 gyr and @xmath126 thereafter . this is more than two orders of magnitude lower than the pattern speeds of other prolate figures ,
i.e. , stellar or gaseous bars , forming within the growing disk , even _
after _ the bar secular slowdown .
overall the halo figure makes @xmath127 , or less than one full rotation , in a hubble time .
hence , for all practical reasons , we can consider that the halo figure is static with respect to any developing bars in the system .
the initial @xmath121 parameter remains constant for qn3 and shows an increase by @xmath128 for models with gas , over the hubble time .
the dm density profile in qn3 quickly achieves the nfw shape and the characteristic radius @xmath129 is @xmath130 kpc by the end of the run .
next , we determine the evolution of the dm halo shape in qn3 .
the inner 30 kpc halo turns around first and develops a triaxial structure between @xmath131 gyr .
1@xmath132 show the evolution of axial ratios , @xmath14 and @xmath16 , in qn3 . beyond @xmath133 kpc , the dm collapses over @xmath134 gyr and forms a strongly triaxial figure as well .
the initial triaxiality develops in response to a short - lived radial orbit instability ( e.g. , macmillan et al .
the dm halo in qn3 exhibits a clear increase in its triaxiality with radius , from the inner , @xmath135 kpc , to the outer , @xmath136 kpc , halo , both in its prolateness and flatness .
the former range is @xmath137 , while the latter lies within @xmath138 .
both parameters show a substantial increase into the outer halo . roughly speaking ,
the inner halo is assembled during the first gyr while the outer one during @xmath139 gyr .
2 shows the growth of the halo virial mass , @xmath140 which is computed by requiring that the mean halo density is equal to the critical density of the universe times @xmath141 , where @xmath141 is obtained from the top - hat model ( e.g. , romano - diaz et al .
2006 , 2007 ) .
the halo virial mass reaches @xmath142 by @xmath6 .
the halo growth rate agrees well with the fully cosmological simulations in ocdm universe ( e.g. , romano - diaz et al .
2007 ) . unlike samland & gerhard ( 2003 )
, we do not assume the law of halo growth and do not limit the shape of the halo to be spherically - symmetric .
next , we present models that include baryons and their associated processes .
the standard model , n3 , is treated as a representative one , and is compared to other models .
the @xmath95 parameter is growing from 0.05 by @xmath128 in all of these models .
we note that _ the very slow tumbling of the halo figure is not affected by the presence of the baryonic matter _ the angular momentum is channeled mainly into the internal circulation .
the overall trend in the evolution of a dm halo remains the same in the presence of baryons ( fig .
1@xmath143-@xmath144 ) . but its prolateness and flatness are lessened by @xmath145 compared to that in qn3 .
this means that the outer halo stays substantially more prolate and flattened than the inner one , as in qn3 .
the maximal prolateness and flatness of the outer halo are reached earlier in n3 and its subsequent reduction is stronger there over the next 23 gyr .
the inner 10 kpc also become more axisymmetric than in qn3 .
asymptotically , n3 is substantially less prolate than qn3 .
the washing out of the prolateness in the inner 20 kpc is clearly connected with the buildup of the disk over the first 23 gyr ( as shown already by berentzen & shlosman 2006 ) .
two factors play the major role in removing the prolatenes of the inner halo .
first , the baryonic ( density ) response to the halo ( equatorial ) prolateness is perpendicular to the halo s nearly non - rotating major axis . in this case
the inner inner lindblad resonance ( inner ilr ) , if it exists , moves to the center , while the outer ilr to the very large radii , outside the halo , securing this response everywhere within the halo .
this baryonic response dilutes the prolate potential of the dm ( more about this response in section 3.6 ) .
second , the disk sitting deep inside the dm potential , as well as numerous baryonic clumps forming _ prior _ to the disk assembly , scatter the dm particle orbits , further reducing its triaxiality .
the gravitational quadrupole of the disk , during its assembly , interacts with that of the prolate halo this induces a chaotic behavior in the halo and in the disk .
the result is a reduced prolateness and flatness in the halo and a more axisymmetric disk . for the stellar component ,
the inner @xmath139 kpc of n3 show a mild prolateness asymptotically , with the axial ratio of @xmath146 at @xmath147 kpc due to a mild bar .
the ability of the dm to maintain the non - negligible prolateness affects both the dynamical evolution of the galactic disk and facilitates the angular momentum redistribution in the disk - halo system , as we discuss in section 3.8 . because dissipative processes dominate the formation of a stellar disk , it is dramatically flatter than the dm . for n3 and other models with baryons discussed here
, the disk formation and evolution can be divided roughly into two stages : gas - dominated and star - dominated .
the former stage lasts over the first 23 gyr we observe that the gas dominates over the stars even in the central regions of the disk .
the latter stage lasts over the rest of the simulation time , during which the ratio of baryonic - to - dm matter within the central 10 kpc reaches @xmath148 and subsequently rises to @xmath149 . the inner few kpc , however , are baryon dominated .
the gas rotation curve flattens at about 290 @xmath150 .
the vertical thickness of the gas layer is large over the first 5 gyrs and then is gradually reduced .
the initially gas - dominated disks are not axisymmetric , and exhibit a range of morphologies : the @xmath151 and @xmath152 modes dominate them at various times , and occasionally they are affected by the off - center , @xmath153 mode . with time , the disks become more axisymmetric , but some stay oval for a long period of time .
+ even towards the end of the simulation in n3 , only a fraction of baryons resides within the central 10 kpc ( where most of the disk lies ) : @xmath154 in stars and @xmath155 in gas .
the total mass of baryons within this radius is @xmath156 , or 49% of the baryons in the simulation .
a similar baryon fraction , @xmath157 , is characteristic for all models . between a third and a half of all baryons
reside outside the sf region in the disk plane ( beyond 15 kpc ) and in the dm halo , i.e. , in the form of partially rotationally - supported and/or hot gas .
the rest of the gas is shock - heated to virial temperatures of @xmath158 k , and ionized .
the dm density profile for n3 can be fitted by the nfw law although the quality of the fit is lower than that of qn3 , and the final value of @xmath129 is @xmath159 kpc .
the inner few kpc and the dm cusp are dominated by the baryons at the end of all the runs with the gas .
early baryon inflow towards the center leads to the formation of a specific configuration near or at the dm density peak that resembles the `` cat s cradle '' pulled by a number of nearly radial ` strings ' ( filaments ) .
this is observed in all the models , except in n4 which has an enforced axisymmetric halo .
the source of this morphology is the background dm distribution within the inner @xmath160 kpc halo during the initial @xmath161 gyr ( fig . 3 , left ) .
the baryonic strings have one - to - one correspondence with the dm strings .
furthermore , the face - on disks show such strings attached to their edge , where the main baryonic inflow joins the disk .
the strings are not perfectly radial , and at the disk edge are tangent to it _ the cold gas inflow , therefore , joins the disk without being shocked , thus supplying a cold , unvirialized gas to the disk_. the typical symmetry resembles the irregular polygons , and ranges from triangles to octagons ( e.g. , fig . 3 ,
the central object an amorphous clumpy protodisk , has a typical size of 24 kpc and the radial strings attached .
the protodisk has an elongated shape and its figure initially nearly stagnates .
the baryonic blobs tend to form on these strings in the equatorial plane and fall along them onto the disk , visibly speeding up its figure rotation .
the feeding of the disk along the strings continues for some time during the first gyr .
models n23n26 with a small sf feedback , @xmath98 , exhibit larger clumps .
the early disk appears as a fat configuration with an equatorial sf and with flaring edges . by about @xmath161 gyr ,
the disk grows substantially , its shape changes rapidly between a less and a more prolate one , and with more or less prominent spiral structure .
dynamical friction appears to play a major role following the clump joining the disk similar to the behavior of the gas - rich disks analyzed by shlosman & noguchi ( 1993 ) . with the beginning of the collapse ,
the rate of baryonic inflow into the inner 20 kpc reaches its maximal rate within the first 1.5 gyr and then steadily falls off .
the sf rate shows a similar behavior ( except in n27 ) , with a peak at 1.51.8 gyr of @xmath162 for the n3 disk ( fig . 4 ) .
overall , the range of the sf rates is @xmath163 , and the peak sf rate can be delayed by 12 gyr compared to the standard model .
n27 differs and shows a flat sf rate for the first 67 gyr
. + the sf rates decay exponentially , in nearly all models ( except n27 ) , to @xmath164 , at @xmath139 gyr , and @xmath165 after @xmath147 gyr .
the latter rate is typically observed in the normal disk galaxies of the nearby universe . in all models ,
the proto - disk plane has been established and visible already at @xmath166 gyr , across the inner 12 kpc , while the stars exist only at the very center by that time .
thereafter , a small @xmath167 kpc gas disk becomes prominent and doubles in size to 5 kpc in another 0.5 gyr , with the sf at its midplane .
after @xmath161 gyr , a gaseous bar of @xmath167 kpc forms , supplemented by an ongoing sf , but collapses to the center and is damped shortly ( more about these bars in sections 3.53.7 ) .
by @xmath168 gyr the disks grow almost to 10 kpc with an intense sf the gas component remains puffed up significantly . by @xmath169 gyr ,
the gas layer starts to cool down .
the transient spiral structure becomes visible and is delineated by the sf regions . by @xmath170 gyr , the vertical structure of the stellar disks in most of the models , but not in all , looks alike .
we observe that the _ disk cools down and thins substantially when the sf rate drops below @xmath4 this is true for all the models , and typically happens after about 5 gyr . _
all the disks , except that of n27 , appear to form from _ inside out _ , and are gas - dominated initially .
the typical time for the disk buildup , i.e. , when it reached about 50% of its final mass at @xmath6 , is @xmath171 gyr .
the stellar rotation curve gradually extends to larger radii and steadily rises with time , while remaining flat in @xmath117 .
its peak moves in to smaller radii , then recedes somewhat ( fig .
( see also the comment on the bar pattern speeds evolution , reflecting the growing mass concentration in the disk , in section 3.6 . ) .
the gaseous component , in the inner few kpc , forms a high surface density ` thin ' disk embedded in the thicker ` corona . '
this corona extends to larger radii .
the extent of the stellar disk is limited to the narrow gas disk only .
while the relative details of this disk - corona morphology differ from model to model , the overall structure is robust .
models n19 to n26 , all show larger disks , some with a strikingly visible bulge accompanied by sf activity in the nuclear rings .
the central kpc of an assembling galaxy in n3 is dominated by the dm during the first @xmath161 gyr , and subsequently by the gas ( for a short period of time ) then by the newly formed stars .
the baryon / dm ratio tends to 0.8 within the disk radius at the end of the simulation .
this value , however , is hardly a standard .
it is affected by the feedback from the stellar evolution , by the critical density for sf and by the collapse time of unstable sf clouds .
+ + + early gas - dominated ( or gas - rich ) nuclear bars appear as a common feature after about 1 gyr in all of the models , albeit their longevity differs from model to model ( typically 45 gyr ) , e.g. , n3 in figs .
67 . their strength , measured by their axial ratio or by the amplitude of their @xmath151 mode , changes with time ,
and they become more star - dominated as the gas is channeled inward and progressively concentrates in the central @xmath8 pc .
the ( radial ) sizes of these nuclear bars change with time as well , from few 100 pc to about 2 kpc .
the central sf is clearly concentrated in these bars .
the stellar bulges are observed to dominate these bars after 2.53 gyr in some models .
the vertical extent of the gas layers within the bulges appears to be similar to elsewhere in the disk .
we note that nuclear _
gas _ bars , devoid of the stellar component , are observed in the central kpc of our models as well . _
these frequent gaseous nuclear bars survive for a limited time , depending on the gravitational coupling from the background potential , and collapse to the center_. nuclear rings appear prominently around the nuclear bars from time to time .
they seem to be connected to the large - scale disk by pair of gas spiral arms which are also visible in the sf ( the outer disks are visibly oval ) , e.g. , n19 in fig .
these rings become more prominent when the gas in the oval disk is channeled to the central kpc .
the gas nuclear bars , seen to develop inside the ring , shrink catastrophically and speed up , until the gas is dumped at the very center leaving a profound central hole in the cold gas distribution .
large - scale bars , @xmath172 kpc , have always developed very early , by @xmath173 gyr , e.g. , in n5 in all cases they have decayed in a few gyr .
in addition , bars have formed late in the evolution , after 5 gyr , e.g. , in n16 ( fig .
9 ) , in about 2/3 of the models .
the early bars appear in response to the prolateness in the dm halo ( and of the stellar disk consequently ) _ they do not develop as a result of a classical bar instability_. these bars are moderately strong , with @xmath174 , and seem to involve a large fraction of the stellar disk mass .
their shapes as well as the shapes of the outer disks depend on the mutual orientation between the dm halo , disk and bar major axes .
we use the @xmath151 mode amplitude , @xmath175 , to quantify the strength of this mode in the disk and the bar , as a function of their p.a . with the halo . applying this procedure to n3 , during the first gyr ,
shows that the disk @xmath175 integrated between 8 kpc
12 kpc displays its minima / maxima when the disk is oriented at @xmath176 to the halo , i.e. , when it leads / trails the dm halo major axis . during the second gyr , the @xmath175 has maxima when the disk is oriented at @xmath177 to the halo in about 1/3 of the rotations . in other cases ,
the @xmath175 appear to ` librate ' around p.a.@xmath178 . as the disk becomes less oval due to its interaction with the halo and the bar
, the residual @xmath175 always librates around p.a.@xmath178 .
the same analysis repeated for the bar in n3 shows that early in the evolution , the bar ` stagnates ' at p.a.@xmath179 to the halo . at later times , the bar sometimes appears stronger at this angle .
but in most cases , @xmath180 , @xmath175 appears irregular as a function of p.a .
hence the bar , as the least massive object in comparison with the disk and the halo , shows a much _ less _ regular behavior of its @xmath175 .
we also find that stellar bars forming late in the simulation are shorter , than the early bars , e.g. , in n16 , where the bar starts to grow anew after @xmath181 gyr and continues its slow growth until the end of the simulations .
in addition , early bars , large - scale and nuclear , are visibly dominated by the sf and are gas - rich . while their late large - scale counterparts appear to be gas - poor .
6@xmath182 exhibits the evolution of the bar pattern speed in n3 . due to the disk growth from inside - out
, it shows rather an unusual behavior it appears at a fixed , @xmath179 , angle to the halo and remains in this position for a large fraction of a gyr .
then the bar speeds up sharply , till @xmath171 gyr . by that time the central bar - hosting regions in the disk are mostly built and further addition of the mass to the disk happens well outside . in the models presented here ,
the growing stellar disks preserve their strong ellipticity ( prolateness ) over the first 12 gyr .
strong nuclear gaseous bars form and experience a distributed sf over their surface . in some models , these bars appear tumbling much faster than the ( oval ) disk figure .
a pair of grand - design spiral arms extend from the surrounding nuclear rings outward , across much of the stellar disks .
after few rotations , the nuclear gas bar collapses to the center ( e.g. , n18 and n26 models in figs .
10 , 11 ) .
+ + + these phenomena become progressively more pronounced along the sequence n16@xmath183n22 , excluding n17 and n20 .
n23n25 are more clumpy initially ( corresponding to shlosman & noguchi [ 1993 ] scenario with clumps spiraling in because of a dynamical friction ) and with no pronounced nuclear morphology later . the appearance of these clumps can be clearly correlated with a sharp decrease in the thermalization of stellar feedback energy , down to @xmath184 .
model n26 represents one of the clearest cases of nested bars having different pattern speeds ( fig .
the nuclear bar appears first within the central kpc at around 1 gyr as a result of the gas inflow from larger scales , where @xmath151 and @xmath152 are clearly visible and constantly evolve .
this nuclear bar can be observed both in gas ( and the associated sf ) and in stars . at @xmath185 gyr , the nuclear gaseous ring appears with an associated sf and is connected to the outer disk by a pair of grand - design spiral arms . at this time , the nuclear bar seems to have a different pattern speed from these arms and tumbles faster .
after @xmath3 gyr , a large - scale stellar bar becomes visible within the central 5 kpc .
the shape and the existence of the nuclear ring depends on the mutual alignment of the nested bars ( shlosman & heller 2002 ) . the ring is also visible in the edge - on disk as being dense and much thinner than the gas layer within the central kpc .
we also confirm that the strength of the small bar depends on the mutual orientation of the bars it is stronger when the bars are aligned and weaker when they are normal to each other ( heller , shlosman & englmaier 2001 ; shlosman & heller 2002 ; englmaier & shlosman 2004 ; debattista & shen 2006 ) .
we note that the nuclear bars are gas - rich in the beginning of this process and form stars vigorously .
gradually the gas in the bar is dumped on the center , and the bar dissolves after @xmath186 gyr .
the pattern speeds of the nuclear bars in all models increase over the first @xmath187 gyr and decline thereafter .
the stellar bar dissolves by @xmath188 gyr .
nuclear ring survives and becomes amorphous .
a specific case of evolution of nested bars has been analyzed by heller et al .
( 2007 ) .
the pure dm models conserve the total angular momentum @xmath0 with the precision of 0.1% . in models with baryons ,
@xmath0 is not conserved _ apriori _ because of the processes associated with the sf and feedback .
energy input from stars leads to pressure gradients , i.e. , hydrodynamical torques , etc .
we find that @xmath0 of all particles in models with sf is nevertheless a well preserved quantity across the models it increases by a small @xmath189 .
these processes , therefore , appear much less important than the gravitational torques for the @xmath0 balance . for baryons ,
@xmath0 decreases by about @xmath190 over the hubble time , while for the dm matter @xmath0 increases by about @xmath191 .
most of this decrease happens during the first 35 gyr . for stars , @xmath0 saturates after @xmath192 gyr , while @xmath0 in the gas continues to decrease by the end of the simulations about @xmath193 of baryonic @xmath0 is in the gas , depending on the model .
note that this involves the gas at large radii as well , in the regions devoid of stars .
this explains the large contribution to @xmath0 by the gas compared to its overall low mass fraction at the end . of a special interest
is the balance of @xmath0 within the central 10 kpc , @xmath194 , where most of the baryonic disk resides .
this angular momentum behaves in the following way ( e.g. , fig .
12 for n3 ) .
the peak of @xmath194 in the dm is achieved at @xmath195 gyr with the follow up sharp decrease by a factor of 22.5 .
the central 10 kpc remains dominated by @xmath194 contributed by the dm for the first @xmath3 gyr .
the rise in the baryonic @xmath194 is much slower ( than for the dm ) within this region and it reaches the maximum only at @xmath196 gyr , in some models declining about 25% thereafter .
the peak of @xmath194 for the dm has a similar value , within a factor @xmath197 , to that of the baryons within 10 kpc ( e.g. , fig .
even more interesting is _ the close similarity between the values of @xmath194 for the baryons and dm after the first ` splash ' over most of the evolution time _
this ` equipartition ' of @xmath194 tells that the angular momentum transfer from the disk to the halo is very efficient and saturates when both components have equal @xmath0 .
this behavior is observed for all models .
stellar @xmath198 mimics that of the baryons in total , being about 25% lower at all times . for the gaseous component
, @xmath198 reaches its maximum between 1.53 gyr , within a subsequent sharp decline ( by a factor of 2 ) and a slow decline thereafter .
the gas @xmath198 is always much smaller than in the dm ( by a factor of 2 ) . the evolution of gas @xmath0 mimics ( somewhat ) that of the dm overall .
the specific angular momentum @xmath199 of the gas is defined as @xmath200 , and similarly for stars and the dm . in models with baryons
we find that @xmath201 increases by about @xmath202 over the hubble time , most of this increase is dated by @xmath203 .
the baryons lose about @xmath204 of @xmath199 , mainly during the epoch @xmath205 .
however , @xmath199 of the gas and stellar particles _ separately _ increases ( nearly ) always monotonically .
initially , stars form at the very center with the minimal @xmath199 , then the sf propagates outward consuming the gas with higher @xmath199 . again , the specific angular momentum @xmath199 of the gas increases with time in all models because the sf first consumes particles with lowest @xmath199 . typically , the stellar @xmath199 saturates its growth after few gyr , while the gas continues to increase its @xmath199 . note ,
that the number of gas particles is not decreasing initially , but their mass is the algorithm is using the option of producing generations of stars from the same gas particles before removing the gas from the balance sheet ( e.g. , fig .
here we describe the model evolution based on the variation of specific parameters related to the sf processes and to the feedback from stellar evolution .
the isothermal eos has been used in n5 and in n22 , n24 and n26 .
compared to n3 , all the isothermal models develop stronger large - scale ( @xmath206 kpc size ) bars and stronger nuclear bars .
the large bars drive much more pronounced outer spiral arms , both in stars and in gas , and accumulate increasingly clumpy gas , especially during the first @xmath207 gyr .
( this is also true when comparing models with and without isothermal eos : n22 with n21 , and n24 with n23 . )
these large bars shrink radially ( arms persist in the gas ) to within the central kpc and largely dissolve due to the quadrupole interaction with the halos , and to a lesser degree due to the heating by a dynamical friction from the numerous clumps , just to develop sometime later on . compared to n3 ,
n21 and n23 , the bars in the isothermal models show a higher rate of sf .
the nuclear bars persist for a few gyrs , leaving an oval distortion thereafter . similarly to the large - scale bars , nuclear bars are ` resurrected ' from time to time . in all of the above models
a central opening of about 1 kpc in the gas appears , surrounded by a ring , by @xmath208 gyr .
gaseous bars driving two gaseous spirals develop within this hole , from time to time , and exhibit a faster pattern speed than the surrounding flocculent spiral structure .
by @xmath209 gyr , the hole grows to about 23 kpc in radius and more prominent in the models with the isothermal eos .
the comparison of pattern speeds shows a diverging evolution for the gas inside 1 kpc and 5 kpc , e.g. , late ( after 9 gyr ) speedup of the gaseous component in n26 , or a late gaseous bar in n22 .
overall , the isothermal models show morphological features that are sharper , including nuclear rings and bars , and their sf rates exceed that of the non - isothermal .
the spiral arms outside the nuclear rings have a grand - design appearance . during the maxima of the sf ,
the gas disk is more puffed up .
the edge - on gas disks exhibit a characteristic concave shape during the first 3 gyrs , and the sf is more pronounced in the mid - plane . in comparison ,
the n3 gas disk is more convex .
the outer massive spirals in the isothermal models appear to be responsible for these differences , which disappers by @xmath3 gyr .
the bulges in the isothermal models extends vertically beyond the disk after @xmath210 gyr , and , by the end of the simulations , are somewhat more pronounced .
the disks in non - isothermal models are smaller , e.g. , in n3 by about 20% , both in stars and gas .
typically , much more gas remains outside the n3 disk , both in the mid - plane and above the disk .
the disk shapes differ dramatically during the evolution between non - isothermal and isothermal models and the morphology of spiral arms looks different as well . in order to investigate the effect of a stellar feedback onto the gas , we have varied @xmath7 from 0.3 in n3 , to 0.1 in n17 .
additional relevant sequences in table 1 are the ` doublets ' n8 ( @xmath211)@xmath183 n6 ( 0.05 ) , n14 ( 0.3 ) @xmath183 n15 ( 0.1 ) , and the isothermal n22 ( 0.1 ) @xmath183 n24 ( 0.01 ) , and the ` triplet ' n21 ( 0.1 ) @xmath183 n25 ( 0.05 ) @xmath183 n23 ( 0.01 ) .
this parameter has a major impact on the evolving morphology in our models .
the sf starts earlier and tends to be more patchy , underlining the clumpy character of the gas , when @xmath7 is smaller .
the clumps spiral in toward the center because of the dynamical friction , especially around @xmath212 gyr .
models with a smaller feedback also show a more prominent morphological features , like bars , nuclear rings and spiral arms . in the first doublet sequence ,
the n17 evolution is similar to that in n3 , except the sf extends to larger radii .
the n3 disk is the smallest among the models listed above and the n23 disk is the largest . in the early stages , the central gas concentration and its central surface density
are observed to increase with decreasing @xmath7 .
the gas layer thins along the same sequence .
+ the stellar bulges form somewhat earlier in models with lower @xmath7 , in n23 by @xmath213 gyr , in n25 by @xmath214 gyr , and in n21 by @xmath215 gyr .
bulges in n17 and n3 form by @xmath215 gyr . by the end of the simulations ,
the largest bulge is found in models with the smallest @xmath7 , among the triplet and doublet parameter sequences given above , i.e. , in n23 ( fig . 14 ) and n24 .
the smallest bulges are found in n3 , n22 . such a bulge
is also expected in n21 . however , as can be seen in fig .
14 , the bulge of n25 is smaller than that of n21 .
the reason for this behavior lies in that n21 develops a strong late bar , while n25 exhibits only an oval distortion after the first 5 gyr .
n23 lacks a late bar totally .
stellar bars are known to heat up the central disk regions and inflate them vertically by increasing their velocity dispersions , both dynamically and secularly ( e.g. , athanassoula & misiriotis 2002 ; berentzen et al .
additional heating exerted by the bar in n21 contributes to the appearance of a more prominent bulge there .
we infer that two processes , at least , emerge as being responsible for the bulge and stellar bar evolution in our simulations both of these processes are ` guided ' by @xmath7 , the stellar evolution feedback onto the gas .
we observe that a decrease in @xmath7 alleviates the clumpiness in the gas , which in turn heats up the stellar component in the disk , therefore , delaying or damping the bar instability there .
this explains why n21 bulge does not ` fit ' in the sequence shown in fig .
we conclude , therefore , that by the end of the simulations there is a clear anti - correlation between the bulge prominence and @xmath7 .
we look for the effect of the critical gas density for the sf by comparing n14 ( @xmath217 ) with n27 ( 0.05 ) , and n15 ( 0.5 ) with n16 ( 0.25 ) .
the main difference observed along this sequence is the delayed evolution in n27 , where @xmath3 kpc bar develops much later compared to n3 and n14 , after @xmath196 gyr only .
it can be also seen in the sf colors .
the dominant sf in the central kpc of this model also appears very late , after @xmath171 gyr , and spreads out in @xmath117 and in @xmath11 ( low surface and volume densities in the gas ) .
it picks up , however , after @xmath139 gyr , and remains much higher than in other models .
the stellar disk is clearly much hotter and puffy in n27 .
the bulge develops late but becomes more pronounced than in n14 .
+ on the other hand , using a low value @xmath101 in n27 produces the smallest @xmath218 kpc disk with a very high surface density , enveloped in a low density stellar and gas material ( fig .
the sf is concentrated in the disk mid - plane . a bar - like stellar and gas response to the halo prolateness
is evident initially within the central @xmath3 kpc , but decays after 3 gyr .
this response forms at a fixed , nearly normal angle with the dm major axis .
hence , using a low density threshold for sf , in the presence of a substantial feedback , heats up the stellar and gas components in the growing disk .
n27 is probably the only model where the sf is distributed and is triggered at all radii @xmath219 kpc in the disk this disk does _ not _ appear to grow from inside out .
it produces a substantial bulge but the smallest stellar disk .
a number of models have a much shorter timescale for sf , @xmath221 instead of 10 ( see table 1 ) we compare n3 ( @xmath222 ) with n14 @xmath223 , n8 ( 10 ) with n16 ( 1 ) , and n17 ( 10 ) with n15 ( 1 ) .
early enough , by @xmath224 gyr , the sf in n14 is less concentrated toward the midplane and the stellar and gas disks are visibly thicker than in n3 .
the sf is also more profound in the n14 bars .
the gas surrounding the disk in n14 is more noticeable , to the extent that the gas disk flares beyond 5 kpc .
by @xmath225 gyr , a 3 kpc disk of higher surface density emerges in n14 while the vertical thickness of the disk is now comparable to n3 , the radial size of n14 disk is smaller . by @xmath226
gyr the bulge of n3 seems more pronounced , while toward the end the n14 bar is more gas - rich than in n3 .
n15 shows earlier and higher rate of sf and has an initial stellar , gas ( and sf ) bar .
this bar weakens and strengthens occasionally but after @xmath139 gyr grows monotonically . for a comparison
, n16 develops such a large - scale bar by 7 gyr .
the disk in n15 is more puffed up than in n17 and the grand - design spiral arms are much less visible .
grand - design spiral structure is strong between @xmath227 gyr apparently when the bar / disk are aligned at certain angles to the dm halo major axis .
the sf , which delineates the arms , follows them , while it is more diffuse in n17 .
the disk is thick and concave at this time , much more than in n17 .
compact gas clumps are present thereafter . by 3.5 gyr ,
the sf in n15 has subsided more than in n17 , the disk seems to be somewhat smaller and the bulge is more compact as well .
it also has less gas in the central kpc .
the gas layer starts to thin first in this model compared to n17 . towards the end of the simulations it is clear that the disk in n15 is overall thicker , and the bulge is standing out with respect to the face - on disk more than in n17
this bulge has an oval / barlike shape .
overall , models with smaller @xmath220 form somewhat smaller in size but more puffed up stellar disks . to enforce the halo axial symmetry during a continuous infall of material
, we have randomized the azimuths of 5% dm particles , every timestep .
therefore , no initial bars are either triggered or driven by the halo prolateness .
furthermore , this should lower the ability of the halo particles to be in resonance with the nonaxisymmetric features in the disk .
this , in turn , should decrease the efficiency of angular momentum flow from the disk to the halo and change the stellar bar evolution .
indeed , in model n4 , the initial collapse proceeds without the ` cat s cradle ' phase and no dm filaments are present .
consequently , the gas does not form the massive clumps .
the disk remains nearly axisymmetric at all times and no substantial spiral structure is visible in the outer part , unlike in in all other models .
some stronger oval distortion develops inside 12 kpc after @xmath196 gyr for some time and decays thereafter .
the gas bars are visible from time to time in this region .
this evolution is in a sharp contrast with n3 which displays an early and late bar activity on various spatial scales .
the sf and spiral morphology in n4 also differ substantialy .
+ models n18 and n19 have larger gravitational softening in gas and in stars @xmath229 pc instead of 150 pc in other models .
we first compare n18 with n3 .
the general evolution of n3 and n18 is similar but it differs in some details .
first , the former model shows sharper features ( shocks , spiral arms , sf regions ) .
second , the gas there is somewhat more clumpy and is able to concentrate more towards the center and the disk mid - plane .
already during the first gyr , n3 gas shows a larger surface density in the central kpc .
third , the bars that form in this model are less strong than in n18 ( figs . 6 , 16 ) .
while the n3 nuclear bar is gas poor ( efficiently channeling it to the center ) , the n18 bar remains gas - rich for a few gyrs .
two factors may reduce the bar strength here : larger gas and stellar concentration to the center and more clumpy gas that scatters bar stars .
this latter feature can also prevent the bar orbits to lock in the resonance with the dm halo orbits , thus resulting in a weaker bar . finally , the stellar bulge in n3 is less prominent than in n18 but is more centrally concentrated .
next , we compare n21 with n19 .
the stellar softening is the same but the gas softening is 150 pc in n21 instead of 250 pc in n19 . as in the previous pair ,
the morphological features are more detailed in n21 .
this is noticeable in the distribution of sf and gas in the strong spiral arms both are more clumpy , but not in stars .
as before , n21 develops a more massive gas concentration in the center , but the difference between the models is not as obvious , apparently because the stellar components have identical softening .
the nuclear rings appear ( vertically ) thinner in n21 and are much more long - lived . occasionally , they collapse to the center . in the long run ,
the difference emerges also in the stellar distribution as well the bulge is visibly larger in n19 and less centrally concentrated .
the bars in n19 are also more gas rich , but periodically this property is inverted between the models . on the other hand , the bars in n19 appear stronger .
hence , the gravitational softening emerges as an important parameter which drives the evolution in our models . especially in gas
, this parameter reflects the overall smoothness of the gas and can be related to a number of intrinsic processes in the ism , heating / cooling and the topology of the magnetic fields those affect the gas clumpiness .
we have simulated the formation of a two - component galactic disk within an assembling dm halo , including the accompanying sf and the stellar feedback onto the ism . for this purpose ,
we have followed the collapse of an isolated cosmological density perturbation with the spin parameter @xmath121 , from its linear regime at @xmath230 and up to @xmath6 .
our approach has several advantages .
we have improved the sf criterion , accounting not only for the contraction and the critical density of the sf region , but also for the background density .
each gaseous particle can make up to four generations of stars , with different metallicities , while an additional parameter allows to consider a delayed collapse of the molecular clouds due to mhd turbulence .
we consider the energy feedback both from stellar winds and from sn , and we are particularly careful in the way this energy is deposited in the gas . we have employed a reasonably high number of particles , allowing us to resolve a number of important features . in particular , our simulations are fully self - consistent , in order to avoid artifacts that could be introduced , e.g. , by a rigid halo ( e.g. , athanassoula 2002 ) .
this enabled us to study interesting dynamics linked to the halo shape .
we have voluntarily refrained from using a yet higher number of particles , in order to be able to run a large number of simulations , much more than is usual in similar studies . with our @xmath160 models
, we tried to gain insight on how the various parameters , and particularly the sf and the energy feedback , affect the structural properties of the bulge , the disk and its components , like bars and spirals .
of course , the available parameter space is very large and much further work and many more simulations are necessary before any thorough understanding is reached . yet , even with the number of simulations we have at our disposal , several clear trends emerge . in all models
the disk forms over a period of time of @xmath231 gyr , i.e. , it reaches 50% of its final mass over this time , although its equatorial plane is established very early .
the dm halo develops a characteristic triaxial shape , i.e. , prolate in the plane perpendicular to the original angular momentum , and flattened along the rotation axis . for pure dm models , the axial ratios of the inner ( @xmath232 kpc ) halos lie between @xmath233 and @xmath234 , and of the outer ( @xmath235 kpc ) halos between @xmath236 and @xmath237 . at the same time , we find that the halo figure tumbles very slowly , @xmath238 over the hubble time . _
the dm collapse pumps most of the angular momentum into the internal circulation within the halo , while the baryonic collapse is stopped by the centrifugal barrier in a flat disk .
_ thus the baryons , due to dissipation , collapse much more profoundly . the halo triaxiality decreases with time during the disk growth .
two processes contribute to this ( 1 ) the increasing central mass concentration in the model and the appearance of of numerous scattering centers ( blobs ) in the halo , and ( 2 ) the out - of phase response of the baryons in the disk to the halo shape .
the first effect reduces somewhat both the flatness and prolateness of the halo , even in the absence of a baryonic component , much more so in the presence of baryons .
the second effect results from the negligible tumbling of the halo figure under these circumstances the inner ilr ( if it exists ) and the outer ilr are pushed to the center and to large radii , respectively , and the disk acquires an elongation perpendicular to the major axis of the halo , thus diluting its potential in the equatorial plane .
this effect is model independent and is unrelated to dissipation .
the angular momentum ( @xmath0 ) flow within the evolving models has been always directed outwards the inner disk is losing @xmath0 to the outer disk and to the surrounding dm halo .
overall , the baryons lose their @xmath0 while the dm increases it in the process of evolution . on the average ,
the dm halo acquires about 2.5%3% of @xmath0 while the baryons lose about 25%30% of their original momentum .
the total @xmath0 is conserved to less than 0.5% even in the models with sf . the specific angular momentum ( @xmath199 )
behaves in a more complicated manner for the dm it is almost constant over the hubble time . for baryons
, @xmath199 decreases with time but each of the components separately , the gas and stars , increase their @xmath199 substantially over the time .
this happens because the first stars form in the center with a minimal @xmath199 , and the first gas consumed by the sf has small @xmath199 as well .
the disk evolution can be roughly divided into two stages : in the early stage , the baryons flow towards the forming small and amorphous disk along a number of radial string patterns this characteristic configuration forms in all models and resembles the cat s cradle .
it survives for about 1 gyr .
in the second stage , the disk periodically changes its shape from less to more oval and frequently develops a pair of grand - design spiral arms .
the disk shape and the strength of these arms depends on their relative angle with the major axis of the dm halo . in nearly all models , the disk grows from inside out and is gas - dominated initially .
about 1/3 to 1/2 of the original baryons remains outside the sf region of the disk or in the halo in form of hot gas at the end of the simulation ( see also sommer - larsen 2006 )
. the star formation ( sf ) is typically strongly concentrated toward the disk midplane .
the gaseous component in the disk is puffed up by the feedback from the stellar evolution .
we find that the gas layer narrows substantially when the sf rate there drops below @xmath4 which typically happens after the first 5 gyr .
the sf delineates various morphological features like early bars , spiral arms and nuclear rings .
late bars appear gas - poor and do not show much sf .
we find that parameters which characterize the disk morphology fit the observations of galactic disks in the nearby universe , e.g. , their radial and vertical scalelengths , shape parameters , bulge - to - disk ratios , etc .
the typical radial scalelength in the disk increases by a factor of 2 over the hubble time , ending within the range of @xmath239 kpc . during this time the disk grows by more than a factor of 10 in mass .
a caution should be taken when tackling the evolution of a collisionless component ( stars ) within the central few 100 pc , specifically the formation of bulges in disk galaxies .
insufficient timestep resolution can lead to being unable to resolve the orbits of individual stars this leads to energy non - conservation and overall heating in the region .
such heating has been detected in the central few 100 pc of our models .
it did not affect the large - scale evolution but could , in principle , modify the nuclear bar strength as well as increase the dispersion velocities in the disky component of the bulge .
we , therefore , postpone the discussion of the bulge formation , the bulge - disk decomposition and the evolution of characteristic scalelengths in our models to a separate publication where high - resolution timestep models are presented . essentially , the disks obtained in our simulations range from being bulge - dominated to nearly bulgeless
. models with larger feedback from the stellar evolution form _ smaller _ bulges and vice versa .
models with shorter timescale for sf formed ` fatter ' disks but not necessary more massive bulges .
moreover , reducing the critical density for sf has led to a distributed sf activity over the inner 15 kpc , as opposed to the inside out disk growth in other models , and to ` fat ' short ( small radial scalelength ) disks .
all models are characterized by an extensive bar - forming activity within the central few kpc .
this is especially true for the initial few gyr of the disk formation .
the early bar axial ratios vary in tandem with the mutual bar - halo orientation .
these bars appear to be gas - rich , channel their gas contents towards the central kpc and weaken substantially over this time . in some cases their strength was observed to revive , even after a prolonged period of time . in other cases , small nuclear bars developed in addition to the primary bars .
dynamically , we find that these nuclear bars either corotate or tumble differently than the large bars .
in all the latter cases , we find that this was preceded by a gas inflow to the region and that nuclear bars tumble substantially faster than their large - scale counterparts .
some of the nuclear bars appear not to have visible stellar components they quickly collapse to the center .
nuclear rings and spirals appear frequently over the hubble time and the former ones are robust features in the disk .
some models show large inflow rates of the gas into the central few kpc leading to the jeans instability in the gas and the formation of massive clumps .
these clumps experience dynamical friction within the disk and spiral in before being consumed by the sf process .
such a regime has been investigated by shlosman & noguchi ( 1993 ) who found that the bar growth is suppressed due to the disk being heated by the clumps ( see also immeli et al .
2004 ) . here
we confirm that indeed the bars have been substantially weakened by these clumps , or did not form in the extreme cases .
our main results indicate that it is possible to form disks which are in agreement with those in the local universe , e.g. , with observed sizes , scalelengths , shapes , mass distributions , etc .
furthermore , we obtain rotation curves that are flat and with the disk / halo contributions as observed ( e.g. , bosma , 2004 ) . the so - called angular momentum ` catastrophe ' discussed in section 1 is naturally avoided in this approach .
the baryons lose only about 25% of their angular momentum because both the original @xmath0 present and the feedback from the stellar evolution reduce their accretion rate onto the disk where most of the @xmath0 is lost .
we find that in all models the total angular momentum of baryons ( stars@xmath96gas ) within the disk region closely follows that of the dm there .
after the initial @xmath3 gyr , the baryons rapidly lose their @xmath0 that flows to the dm until ` equipartition . '
moreover , a full range of bulge - dominated to nearly bulgeless disks have been obtained as a result of the evolution over the hubble time .
the disks formed in our numerical simulations do not appear to suffer from being too compact and having small radial or large vertical scalelengths .
_ the growth of the disk can not be modeled realiably without resolving its radial scalelength _ an issue which is difficult to address at present in cosmological simulations and which is resolved in our approach . the problem of over - cooling when the gas radiates away the energy deposited by feedback from the stellar evolution ( ob stellar radiation - driven winds and sn ) , is also addressed here ( see section 2 ) .
bars provide the strongest impetus for an internal evolution in disk galaxies by imposing gravitational torques which trigger the mass and angular momentum redistributions within the central @xmath147 kpc . while prolate dm halos appear to be incompatible with large - scale stellar bars ( el - zant & shlosman 2002 ; berentzen et al .
2006 ) , the growth of disks acts to wash out the equatorial prolateness within , at least , the inner halos ( berentzen & shlosman 2006 ) .
the present work confirms this trend we observe a decrease in the halo flatness even well outside the disk , and elimination of its prolateness within the inner ( @xmath232 kpc ) halo and reduction in the outer halo . the disk shape and morphology
are affected by the halo shape during the first few gyr the disks appear oval in contradiction of those observed in the local universe ( e.g. , rix & zaritsky 1995 ) .
the interaction between the ovally - shaped disks and prolate halos provide for the spiral driving which clearly depends on the mutual disk - halo orientations . in summary
, we have explored different aspects of disk galaxy formation in the framework of the collapse of an isolated density pertubation embedded in the hubble flow .
this allows us to focus on specific processes normally unresolved in the cosmological simulations . on the other hand
, we lose the effect of the substructure on the model .
specifically , we have aimed at understanding the imprint of various feedback parameters describing the star formation on the basic charateristics of the disk - halo system .
one should consider this effort as an exploratory one , which requires a full implementation of realitic initial conditions in tandem with a higher numerical resolution .
we thank our colleagues , too numerous to list here , for interesting discussions on the subject of this work .
this research has been partially supported by nasa / ltsa 5 - 13063 , nasa / atp nag5 - 10823 , hst / ar-10284 ( is ) , and by nsf / ast 02 - 06251 ( ch , is ) and by anr blan06 - 2 - 134356 ( ea ) .
c.h and i.s .
are gratefull for the support from observatoire de marseille during their stay there . | we investigate the formation and subsequent co - evolution of galactic disks immersed in assembling live dark matter ( dm ) halos .
disk and halo structural components have been evolved from the cosmological initial conditions and represent the collapse of an isolated density perturbation .
the baryonic component includes gas ( which participates in star formation [ sf ] ) and stars .
the feedback from the stellar energy release onto the ism has been implemented , and so is the mass , momentum and energy balance between the stellar and gaseous components .
we find that ( 1 ) the growing _ triaxial _ halo figure tumbling is insignificant and the angular momentum ( @xmath0 ) is channeled into the internal circulation , while the baryonic collapse is stopped by the centrifugal barrier ; ( 2 ) density response of the ( disk ) baryonic component is out of phase with the dm , thus diluting the inner halo flatness and washing out its prolateness over the time period of the disk growth ; ( 3 ) the total @xmath0 is neathly conserved , even in models accounting for feedback from stellar evolution ; i.e. , the baryons lose @xmath1 and the dm acquires @xmath2 of their original @xmath0 ; baryons and dm within the disk radius have the same @xmath0 after the first @xmath3 gyr ; ( 4 ) the specific @xmath0 for the dm is nearly constant , while that for baryons is decreasing with time ; ( 5 ) early stage of disk formation resembles the cat s cradle a small amorphous disk fueled via radial string patterns ; followed by growing oval disk whose shape varies with its orientation to the halo major axis , and so is the strength of a pair of grand - design arms developed in the disk ; ( 6 ) the initially puffed up gas component in the disk thins when the sf rate drops below @xmath4 ; ( 7 ) about 40%-60% of the baryons remain outside the sf region of the disk or in the halo in the form of a hot gas by the end of the simulations ; ( 8) obtained rotation curves appear to be flat and account for the observed disk / halo contributions ( the halo virial mass is @xmath5 at @xmath6 ) ; ( 9 ) a range of bulge - dominated to bulgeless disks was obtained , depending on the energy thermalization parameter , @xmath7 , which characterizes the feedback from the stellar evolution smaller @xmath7 leads to a larger and earlier bulge ; lower density threshold for sf leads to a smaller , thicker disk ; gravitational softening in the gas has a substantial effect on various aspects of galaxy evolution and mimics a number of intrinsic processes within the interstellar medium ; ( 10 ) the models are characterized by an extensive bar - forming activity within the central few kpc whose properties vary with the bar - halo orientation ; ( 11 ) nuclear bars , dynamically coupled and decoupled form in response to the gas inflow along the primary bars , as shown by heller , shlosman & athanassoula . |
physics in 1d systems manifests a number of peculiar phenomena , such as spin - charge separation , conductance quantization,@xcite and anomalous low - temperature behavior in the presence of backscattering impurity.@xcite it is reasonable to expect that the more complex structures composed of crossed 1d systems , such as crossings and arrays , should exhibit some particular features as well .
although the transport properties of crossed 1d systems and their arrays have been thoroughly studied both theoretically@xcite and experimentally@xcite , the electronic structure of these systems is much less understood and the interpretation of existing experimental results is challenging .
recent scanning tunnelling microscopy ( stm ) experiments on a metallic carbon nanotube crossed with a semiconducting one@xcite have shown the existence of localized states at the crossing which are not due to disorder . however , these localized states do not appear systematically in all experiments , i.e. the effect is highly dependent on the nature of the carbon nanotubes ( metallic or semiconducting ) , of the barrier formed at the crossing , etc . aiming at clarifying this problem
, we present in this paper a detailed study of tunnelling effects between crossed 1d systems in the presence of potential barriers for massive quasiparticle excitations . because effects of electron - electron interactions can be reasonably incorporated in a random phase approximation ( rpa),@xcite we study a simpler model , accounting for electron - electron interactions only within hartree approximation .
the outline of this paper is the following : in section ii we introduce the model that we are going to use to describe the array of crossed nanowires . in section iii
we consider a particular case of free electrons and write down explicit solutions for the case of one and four crossings .
section iv contains formal general solution with additional details given for the case of a symmetric external potential .
we demonstrate the effect of tunnelling on the electronic structure of single crossings in section v and qualitatively discuss different possibilities depending on the external potential . section vi contains quantitative analysis and comparison with available experimental data of the electronic structure of single crossing for different values of parameters .
our conclusions and open questions are presented in section vii .
we consider a system composed of two layers of crossed quantum wires with interlayer coupling .
the upper layer has a set of parallel horizontal wires described by fermionic fields @xmath0 , whereas the lower layer contains only vertical parallel wires described by the fields @xmath1 .
the wires cross at the points @xmath2 , with @xmath3 and the distance between layers is @xmath4 , with @xmath5 , see fig.1 .
+ + the partition function of the system reads @xmath6 d[\psi^*_j ] d[\varphi_{i } ] d[\varphi_{i}^*]e^{-s/\hbar},\ ] ] with the total action given by @xmath7 the first term accounts for the kinetic energy and external potential @xmath8 , which can be different in each wire and may arise , e.g. , due to a lattice deformation , when one wire is built on top of another , @xmath9 where @xmath10 here , @xmath11 denotes the chemical potential in the upper ( @xmath12 ) or lower ( @xmath13 ) layer . the second term of eq .
( [ taction ] ) describes scattering at the crossings @xmath2 , @xmath14 where @xmath15\left ( \begin{array}{cc } u_{ij } & t_{ij}\\ t^*_{ij } & \tilde{u}_{ij } \end{array } \right ) \left [ \begin{array}{c } \psi_{j}(x_i,\tau)\\ \varphi_{i}(y_j,\tau ) \end{array } \right].\ ] ] notice that the matrix element @xmath16 describing intra - layer contact scattering can , in principle , be different from @xmath17 , but both must be real . on the other hand , the contact tunnelling ( inter - layer ) coefficient between the two crossed wires @xmath18 can be a complex number , since the only constraint is that the matrix above must be hermitian . the third term in eq .
( [ taction ] ) accounts for electron - electron interactions , @xmath19
we start by considering a very simplified case , namely , free electrons ( no electron - electron interaction , @xmath20 and no external potential , @xmath21 ) .
moreover , we assume @xmath22 and put @xmath23 . the interlayer tunnelling is assumed to be equal at each crossing point @xmath24 and to have a real and positive value .
in such a case , the partition function consists of only gaussian integrals .
we can then integrate out the quantum fluctuations , which reduces the problem to just solving the equations of motion . considering a real time evolution and performing a fourier transformation in the time variable ,
we are left with the following equations of motion for the fields : @xmath25 where @xmath26 denotes the electron mass and @xmath27 is the energy of an electron state .
firstly , we evaluate the solutions for the case of free electrons without tunnelling and then we investigate how the addition of tunnelling changes the results .
the solution for the free electron case consists of symmetric and antisymmetric normalized modes , @xmath28 respectively .
the corresponding momenta @xmath29 and @xmath30 depend on the boundary conditions : with open boundary conditions @xmath31 , @xmath32 and with periodic boundary conditions @xmath33 for a wire of length @xmath34 and @xmath35 integer . to find the solution for the case with tunnelling @xmath36
, we have to solve eqs .
( [ eom1 ] ) .
these equations are linear , therefore , the solution consists of a homogeneous and an inhomogeneous parts , @xmath37 which are @xmath38 @xmath39 imposing open boundary conditions , @xmath40 , we find @xmath41 writing the above equations in a matrix notation and inverting yields @xmath42 substituting explicitly the expression for @xmath43 given by eq .
( [ inh1 ] ) and using the mathematical identity @xmath44 leads , after simplifications , to the solution @xmath45 where , for open boundary conditions , @xmath46,\label{simplerep}\end{aligned}\ ] ] and the energy @xmath27 is related to @xmath47 as @xmath48 .
similar calculations can be performed for the case of periodic boundary conditions , yielding eq .
( [ fsol1 ] ) with @xmath49 in particular , for the simplest case of a single horizontal and a single vertical wires , with just one crossing at @xmath50 , the solution is : @xmath51 + by substituting @xmath52 , we find that at the crossing point @xmath53 the consistency condition requires that @xmath54 or @xmath55 the solution is even simpler if @xmath56 .
then , for open boundary conditions , the symmetric modes are @xmath57 and the antisymmetric modes are left unchanged in comparison with eqs .
( [ freesol ] ) . also , @xmath58 and the secular equation ( [ freespectr ] ) becomes @xmath59 ^ 2=1,\ ] ] which splits into two transcendental equations @xmath60 the first one describes the shifted values of scattering states energies , whereas the second equation has an additional bound state solution with @xmath61 , if @xmath62 .
the appearance of the bound state is exclusively due to the presence of tunnelling . for an electron in a wire of length @xmath63
nm the corresponding value is @xmath64 ev@xmath65 nm and for quasiparticles the value of @xmath66 is typically larger , inversely proportional to their effective mass . defining then @xmath67 and taking the thermodynamic limit @xmath68 , we find @xmath69 with the corresponding bound state energy @xmath70 and the wave function given by @xmath71 the factor @xmath72 instead of @xmath73 comes from the fact that now an electron can tunnel into the other wire , where its wavefunction @xmath74 .
( [ benergy ] ) and ( [ expsol ] ) hold for both open and periodic boundary conditions .
since the threshold value @xmath66 is quite small , the bound state should exist for a typical crossing with relatively good contact . however , the energy of the state is extremely small , @xmath75 ev if @xmath76 .
qualitatively similar results were found by numerical computation@xcite of the ground - state energy of an electron trapped at the intersection of a cross formed by two quantum wires of finite width . for the case of two wires in the upper and two in the lower layers , there are four crossings . in this case , the self consistent equations read @xmath77=m(x_1,x_2,e ) \left [ \begin{array}{c } \varphi_{1}(y_1)\\
\varphi_{1}(y_2)\\ \varphi_{2}(y_1)\\ \varphi_{2}(y_2 ) \end{array } \right]\ ] ] and @xmath78=m(y_1,y_2,e)\left [ \begin{array}{c } \psi_{1}(x_1)\\
\psi_{1}(x_2)\\ \psi_{2}(x_1)\\ \psi_{2}(x_2 ) \end{array } \right],\ ] ] where @xmath79 the secular equation then has the form @xmath80=0,\label{spectraleq}\ ] ] which yields a rather complicated transcendental equation ( @xmath81 is the identity matrix ) .
the spectral equation for bound states @xmath61 can be significantly simplified in the thermodynamic limit @xmath68 .
then , with @xmath82 , for both open and periodic boundary conditions , the matrix elements become @xmath83 and the secular equation in eq .
( [ spectraleq ] ) has 4 solutions with negative energy described by @xmath84 here , @xmath85 , @xmath86 , and @xmath87 ( notice the implicit dependence of @xmath88 and @xmath89 on @xmath27 ) .
the value of @xmath90 depends exponentially on the distance between the crossing points . in the limit
@xmath91 the value of @xmath92 , which correspond to four independent crossings with the bound state energy @xmath93 , the same value as we found in the previous case ( see eq.([benergy ] ) ) .
consider now a regular square lattice , with lattice constant @xmath94 .
then , one has @xmath95 and @xmath96 . from symmetry arguments ,
the wave functions should be @xmath97 and @xmath98 . after substituting them into eq .
( [ fsol1 ] ) and using eq .
( [ fundbound ] ) we find @xmath99,\\ \nonumber \varphi_l(y)=&-&t\psi_0(x_l)\frac{me^{i k_y l_y a}}{\hbar^2 \kappa}\left[\frac{\sinh(\kappa x - \kappa a l_y)e^{i k_y a } } { \cosh(\kappa a ) -\cos(k_y a ) } \right.\\ \nonumber & -&\left.\frac{\sinh(\kappa y - \kappa ( l_y+1)a)}{\cosh(\kappa a ) -\cos(k_y a ) } \right],\end{aligned}\ ] ] where @xmath100 , such that @xmath101 and @xmath102 .
therefore , @xmath103 and @xmath104 , with @xmath105 and @xmath106 related by @xmath107 thus , the spectral equation reads @xmath108[\cosh(\kappa a ) -\cos(k_y a ) ] } .\ ] ] by performing an analytic continuation @xmath82 in eq.([bandsol ] ) , we find an equations similar to the one obtained previously by kazymyrenko and douot@xcite when studying scattering states in a lattice .
the spectral equation describes a band formed by bound states with energies @xmath109 .
the momenta @xmath110 and @xmath111 run in the interval @xmath112 if @xmath113 or inside the region @xmath114 if @xmath115 .
similar results were calculated,@xcite estimated,@xcite and measured@xcite in the context of hybridization between vertical and horizontal stripe modes in high - tc superconductors .
now we consider a more general model , which takes into account the presence of an inhomogeneous potential @xmath8 arising from possible lattice deformations , and includes electron - electron interactions @xmath116 , which will be treated at a mean field level , within the hartree approximation @xmath117 .
each crossing @xmath2 is considered as a scattering point with tunnelling @xmath18 and scattering potential @xmath16 .
the corresponding equations of motion then read @xmath118\delta(x - x_l)&= & 0,\\\nonumber d_{iy}\varphi_{i}(x)+\sum_{l}[\tilde{u}_{il}\;\varphi_{i}(y_l)+t^*_{il}\psi_{l}(x_{i } ) ] \delta(y - y_l)&= & 0,\ ] ] where @xmath119 with @xmath120 .
this model is solved most easily through the green s function satisfying @xmath121 with @xmath122 and the corresponding open boundary conditions , @xmath123 or the periodic ones @xmath124 where the prime denotes the derivative with respect to @xmath125 .
note that we consider real time green s function for a particular wire ( not the whole system ) .
the solution to the model is @xmath126 g_j(x , x_l , e),\\ \varphi_{i}(y)&=&-\sum_{l}[\tilde{u}_{il}\;\varphi_{i}(y_l)+t^*_{il}\psi_{l}(x_{i } ) ] g_i(y , y_l , e),\label{gensol}\end{aligned}\ ] ] which we require to be normalized latexmath:[\[\sum_{l}\left(\int |\psi_{l}(x)|^2 dx+\int value of the functions at crossing points @xmath2 yields the equations @xmath128&=&0,\\\nonumber \sum_{l}[(\tilde{u}_{il}\ ; g_s(y_j , y_l , e)+\delta_{jl})\varphi_{i}(y_l)\\ + t^*_{il } g_i(y_j , y_l , e)\psi_{j}(x_{i } ) ] & = & 0.\label{cons}\end{aligned}\ ] ] to find nontrivial solutions for the fields @xmath0 and @xmath1 , the system of homogeneous equations in eq.([cons ] ) has to be linearly dependent and hence the solution is represented by the null space of the system .
this means that after writing the equations in a matrix form , the determinant of the matrix should be zero , thus leading to a spectral equation for @xmath27 . moreover , bound state solutions in the thermodynamic limit @xmath68 satisfy both open and periodic boundary conditions , since @xmath129 and @xmath130 . to understand better the dependence of the green s function @xmath131 on @xmath27 , we represent the function through the solutions of the homogenous equations , @xmath132 we omit the index @xmath133 in what follows for simplicity .
the most general and common representation , which holds for any static potential , reads as follows : @xmath134 here , the function @xmath135 is the solution of the homogenous equation @xmath136 and the spectrum @xmath137 is obtained by imposing the corresponding boundary conditions .
notice that in the present representation of @xmath138 the functions @xmath139 have to be orthonormal . by writing @xmath138 in the form given in eq .
( [ fsrep0 ] ) , the following identity arises @xmath140 the case @xmath141 for free electrons is illustrated in fig.[fig3 ] , where eq .
( [ nonintfs ] ) is plotted . if some external potential is present , @xmath142 has the same form but the positions of the poles are shifted and the corresponding values are different .
@xmath143 in units of @xmath144 versus @xmath27 in units of @xmath145 . ,
width=226 ] if no regularization is used , the calculations for @xmath146 must be performed in the finite size limit , otherwise with @xmath147 the energy distance between different modes vanishes and the poles situated on the real positive half axis merge to form a branch cut singularity .
this behavior can be readily seen on the example of eq .
( [ nonintfs ] ) , where can perform an analytic continuation , considering @xmath148 . then , in the limit @xmath147 , @xmath149 , and the function @xmath150 changes sign as one goes from the upper to the lower complex half plane for @xmath151 .
now we represent the green s function through the solutions of the homogenous equation @xmath152 this is a second order differential equation , therefore , it should have two linearly independent solutions , which we call @xmath153 and @xmath154 .
then the green s function is @xmath155 where the expressions for the coefficients @xmath156 ( functions of @xmath157 ) , are derived in the appendix a. in particular , for a symmetric potential @xmath158 , we can choose a symmetric @xmath159 and an antisymmetric @xmath160 solutions as linearly independent , i.e. , @xmath161 and @xmath162 .
thus we find @xmath163\ ] ] and @xmath164 to obtain the results in the thermodynamic limit @xmath165 , it is useful to rewrite @xmath166 using quantities which do not depend on @xmath167 explicitly .
for example , @xmath168 after substitution of eqs .
( [ freesol ] ) into eq.([fsrepf ] ) and simplification , for the case of noninteracting electrons we find @xmath169,\nonumber\end{aligned}\ ] ] which is the same expression as in the previous section ( see eq.([simplerep ] ) ) .
this is a posteriori justification of the use of the same letter @xmath138 in the first section .
the case of a harmonic potential is considered in appendix b.
now we apply our results including tunnelling and external potential to the simpler case of only two crossed wires , aiming to compare our findings with experiments . using the general solution given by eq .
( [ gensol ] ) , and considering @xmath170 , we can write @xmath171 g_1(x , x_0,e),\\ \nonumber \varphi(y)&=&-[\tilde{u}\;\varphi(y_0)+t\psi(x_{0 } ) ] g_2(y , y_0,e).\label{solsingle}\end{aligned}\ ] ] by substituting @xmath52 , we find that at the crossing point @xmath172\psi(x_{0})+t
g_1(x_0,x_0,e)\varphi(y_0)&=&0,\\ \nonumber [ 1+\tilde{u } g_2(y_0,y_0,e)]\varphi(y_0)+t g_2(y_0,y_0,e)\psi(x_{0})&=&0.\label{solsinglecross}\end{aligned}\ ] ] the consistency condition requires that @xmath173 or @xmath174[1+\tilde{u } g_2(y_0,y_0,e ) ] \\ \nonumber & -&t^2 g_1(x_0,x_0,e ) g_2(y_0,y_0,e).\end{aligned}\ ] ] the meaning of this equation becomes clearer in the symmetric case , when @xmath175 and @xmath176 . in this case
, it reduces to a quadratic equation , which bears two solutions , @xmath177 notice that they differ by the sign in front of the tunnelling amplitude @xmath178 , which is shifting the potential @xmath179 .
such symmetry effectively reduces the problem to 1d with effective potential @xmath180 .
hence , we have @xmath181 the shift of the energy levels in a wire due to the presence of the @xmath182 potential can be visualized with the help of the green s function expansion , where one has @xmath183 in the case with @xmath184 , the energies are exactly those of the poles and , therefore , remain unshifted .
however , since @xmath185 , the curve actually describes how the energies of the modes change as we keep increasing @xmath186 from @xmath187 if @xmath188 or decreasing @xmath186 from @xmath189 if @xmath190 . in the latter case
, we can run into the region with @xmath61 , which would correspond to the appearance of a bound state . nevertheless , to obtain an exact solution ,
it is more convenient to work with the expression for @xmath191 in terms of the wave functions , @xmath192 where we assumed @xmath193 for simplicity .
voltage versus length diagram , which shows the experimentally observed density of states .
notice the existence of two localized states in black .
( extracted from ref . )
.,width=226 ] now , we will compare our theoretical findings with experimental results .
we concentrate mostly on the analysis of a system consisting of two crossed single wall carbon nanotubes ( swnts ) : a metallic on top of a semiconducting ( ms ) one.@xcite in its unperturbed state , the band structure of a swnt can be understood by considering the electronic structure of graphene . due to its cylindrical shape
, the transverse momentum of one particle excitations in a swnt has to be quantized , whereas the longitudinal momentum may vary continuously . combining this condition with the assumption that the electronic structure is not very different from that of graphene , one finds two different situations , depending on the topology of the swnt : there are no gapless modes and the nanotube is semiconducting , or two gapless modes are present and
the nanotube is called metallic . analyzing the spectroscopic measurements performed along the metallic nanotube ( see fig .
[ figexp ] ) and comparing with the unperturbed electronic structure , one notices two main changes .
first , a small quasi gap opens around the fermi energy level @xmath194 between @xmath195 ev and @xmath196 ev in the spectrum of the massless modes ( corresponding to zero transverse momentum ) .
second , two peaks are visible at @xmath197 ev and @xmath198 ev in the region around the crossing , corresponding to localized states between the fermi energy and the van hove singularity at @xmath199 ev .
such states are not visible above the fermi energy , thus suggesting that the electron - hole symmetry is broken by the presence of some external potential
. the latter may appear due to lattice distortions and the formation of a schottky barrier at the contact between the nanotubes .
@xcite in the following , we show that if the potential is strong enough , localized states can form in the spectrum of the massive mode corresponding to the van hove singularity with energy @xmath200 . therefore , the observed localized states should have @xmath201 ev and @xmath202 ev . to incorporate in a more complete way the effects of the schottky barrier and lattice deformation , we assume @xmath203 to have a lorentzian shape , @xmath204 firstly , we study the influence of this potential alone on the electronic structure , i.e. we assume that there is no tunnelling @xmath205 , and no electron - electron interactions .
exact numerical solution of the schrodinger equation shows that an approximation of the potential in eq .
( [ potential ] ) by the harmonic one does not change the solution qualitatively .
therefore , we consider @xmath206 , which describes a harmonic oscillator with frequency @xmath207 and corresponding spectra @xmath208 for @xmath209 .
moreover , it is reasonable to assume that the strength of the barrier @xmath210 is of the same order as the energy of the bound states and that the potential is localized on the same length scale as the localized states .
hence , we take @xmath211 ev and @xmath212 nm . it follows then from our calculations that the difference between neighboring energy levels is quite small and there are many bound states present in the case when @xmath26 is the actual electron mass . however , assuming @xmath26 to be an effective electron mass , with @xmath213 , which is of the same order as the experimentally estimated values @xmath214@xcite and @xmath215,@xcite we find exactly two pronounced bound states : the first one has @xmath216 ev and is described by the symmetric wavefunction @xmath159 as shown in fig .
[ fig4 ] , @xmath159(nm@xmath217 ) versus @xmath218 ( nm ) , width=226 ] whereas the other has @xmath219 ev and is described by the antisymmetric wavefunction @xmath160 , see fig .
[ fig5 ] .
@xmath160(nm@xmath217 ) versus @xmath218 ( nm ) . , width=226 ] considering fig .
[ fig4 ] , we observe that the localization size of the state is around @xmath220 nm , which agrees well with the experimental data . on the other hand , the state shown in fig.[fig5 ]
has a zero value exactly at the crossing and is rather spread , a behavior which is not observed experimentally .
besides these two , a number of other states are also present in the vicinity of the van hove singularity with @xmath221 ev .
secondly , we take into account electron - electron interactions to consider other possibilities to obtain two pronounced bound states .
unfortunately , our approach only allows us to incorporate electron - electron interactions at the mean - field level by using the hartree selfconsistent approximation @xmath222 where @xmath223 is the electron density , given by @xmath224 here the summation @xmath47 goes over energy levels and @xmath225 is the fermi distribution . although it is known that in 1d systems quantum fluctuations play an extremely important role , we nevertheless start with the mean - field approximation as a first step to incorporate them in rpa .
moreover , we believe that their presence does not qualitatively change the obtained results . to render the numerical calculation simpler , we consider a delta - like interaction potential , which leads to @xmath226 by estimating the effective interaction strength @xmath227 from the luttinger liquid theory , we obtain that @xmath228 ev@xmath65 nm for @xmath229 cm / s.@xcite suppose that the lowest energy state with @xmath216 ev is occupied by an electron with a certain spin .
then , there is a possibility to add to the same state an electron with an opposite spin .
however , due to the repulsive coulomb interaction the energy of the two - electron state becomes @xmath219 ev for @xmath230 ev@xmath65 nm .
the corresponding self consistent solution is presented in fig .
[ fig4int ] .
@xmath159(nm@xmath217 ) versus @xmath218 ( nm ) , width=226 ] the state has the same shape as in fig .
[ fig4 ] , but is a bit more spread . by comparing the density of states ( dos ) distribution with scanning tunnelling spectroscopy ( sts ) data for the crossing,@xcite we observe that the inclusion of electron - electron interactions ( fig .
[ fig4int ] ) provides a much better agreement between theory and experiment for the @xmath219 ev bound state than in the previous case ( fig.[fig5 ] ) .
thirdly , we take into account tunnelling between the wires .
qualitatively , this leads to the splitting of energy levels and redistribution of charge density in the wires , thus effectively reducing the strength of electron - electron interactions . since we have no information about the electronic structure of the semiconducting nanotube , to make a quantitative estimation we assume that the effective mass is equal in both wires and that the potential is also the same .
in such a case , from symmetry arguments the electron density should be evenly distributed in both wires even for a very weak tunnelling .
therefore , the electron - electron interactions should be twice stronger than in the case without tunnelling , namely , @xmath231 ev@xmath65 nm to achieve the same energy value . moreover , if the tunnelling coefficient is large enough , the splitting of the energy levels becomes significant and detectable .
we can estimate the coefficient @xmath178 , if we assume that it has the same order for sm , metallic - metallic ( mm ) , and semiconducting - semiconducting ( ss ) nanotube junctions .
the ss and mm junctions have ohmic voltage - current dependance , characterized by the conductance @xmath232 .
moreover , we can estimate the transmission coefficient of the junction as @xmath233 , for @xmath234 . for mm junctions
experimental measurements@xcite typically yield @xmath235 , thus corresponding to @xmath236 ev@xmath65 nm . for example , for @xmath237 ev@xmath65 nm and @xmath238 ev in eq.([potential ] ) , without electron - electron interactions we find that the system has _ two bound states_. the lowest energy bound state with @xmath216 ev is shown in fig.[fig6 ] .
@xmath239(nm@xmath217 ) versus @xmath218 ( nm ) . , width=226 ] compared with fig .
[ fig4 ] , the state has a peak exactly at the crossing , corresponding to a local increase of the dos .
the other bound state with @xmath219 ev is shown in fig .
[ fig7 ] .
@xmath240(nm@xmath217 ) versus @xmath218 ( nm ) . ,
width=226 ] contrary to the previous case , the state has a deep at the crossing , corresponding to a local decrease of the dos .
however , these local change in dos is too small to be observable in the present experimental data .
if we now include electron - electron interactions with @xmath230 ev@xmath65 nm and add a second electron with different spin to the system , we find that the new state has @xmath241 ev and acquires the shape shown in fig .
[ fig8 ] .
@xmath239(nm@xmath217 ) versus @xmath218 ( nm ) . ,
width=226 ] the last result suggests that there are yet other possible interpretations of the experimental results .
firstly , if the potential in the metallic swnt is significantly decreased due to screening effects but a schottky barrier in the semiconducting swnt can reach considerable values , sufficient for the formation of the bound states , then the latter are also going to be present in the metallic swnt due to tunnelling between swnts .
secondly , there is still a possibility to find a bound state existing purely due to tunnelling , i.e. , without external potential , as was shown in eq .
( [ expsol ] ) , and a second bound state may arise with different energy due to coulumb repulsion between electrons with different spins
. however , this is most probably not the case we have in the experiments , because due to electron - hole symmetry such states would exist also above the fermi energy , a result which is not observed experimentally .
we presented several possibilities to explain the observed localized states at the crossing of metallic and semiconducting nanotubes.@xcite all of them require the existence of an external potential in the metallic and/or semiconducting swnt to break the electron - hole symmetry , since the localized states were seen only below the fermi energy .
most probably , such a potential comes from a schottky barrier and the effect of lattice distortions is minimal , since such localized states were , up to now , observed only for ms crossings and not for mm or ss ones .
moreover , the effective mass of quasiparticle excitations should be of order @xmath213 , where @xmath242 is the actual electron mass , to generate only a few bound states localized on a region of approximately @xmath220 nm with energy of order of @xmath243 ev .
the best agreement with the experimental data is obtained by assuming that the second bound state has a different energy due to the coulumb repulsion between electrons with different spins .
the role of tunnelling in the observed electronic structure is not clear and allows for many interpretations . to avoid such ambiguity ,
the electronic structure of the semiconducting nanotube should be measured as well . moreover , to be sure that the available sts measurements indeed represent the electronic structure of the nanotube and are free of artifacts introduced by the stm tip @xcite several measurements with different tip height should be performed .
we are very grateful to s. g. lemay for useful discussions . here
we consider the green s function as a function of one variable @xmath125 and fix @xmath157 for a moment .
since @xmath138 is the green s function , we require it to satisfy proper boundary conditions @xmath244 , be continuous @xmath245 , and also @xmath246 . substituting eq .
( [ fsrepf ] ) into the above requirements one finds @xmath247 where @xmath248 multiplying the eq .
( [ acoef ] ) by the matrix @xmath249 we find @xmath250\\ \psi_{2}(-l)[\psi_{2}(l)\psi_{1}(x_2)-\psi_{1}(l)\psi_{2}(x_2)]\\ -\psi_{1}(l)[\psi_{2}(-l)\psi_{1}(x_2)-\psi_{1}(-l)\psi_{2}(x_2)]\\ -\psi_{1}(-l)[\psi_{2}(l)\psi_{1}(x_2)-\psi_{1}(l)\psi_{2}(x_2 ) ] \end{array } \right),\ ] ] where @xmath251^{-1}.\ ] ] the wronskian @xmath252 is nonzero for linearly independent functions and its value does not depends on the point @xmath157 .
suppose that eq .
( [ srdng ] ) has a solution @xmath253 which is neither symmetric nor antisymmetric .
thus , for symmetric potentials @xmath254 is also a solution and both of them are linearly independent .
furthermore , we can then compose a symmetric @xmath255 and an antisymmetric @xmath256 solutions .
in particular , for a harmonic potential @xmath257 , one can find such a solution @xmath258 where @xmath259 is the hermite polynomial for integer @xmath260 .
it follows then that @xmath261 and @xmath262 moreover , in the thermodynamic limit @xmath147 , @xmath263 the eq .
( [ harmfsol ] ) approaches asymptotically the expression for free fermions , as @xmath264 for @xmath61 , @xmath265 10 m. bttiker , y. imry , r. landauer , and s. pinhas , phys .
b * 31 * , 6207 ( 1985 ) .
c. l. kane and m. p. a. fisher , phys.rev . lett . *
68 * , 1220 ( 1992 ) .
a. komnik and r. egger , phys .
lett.*80 * , 2881 ( 1997 ) .
b. gao , a. komnik , r. egger , d. c. glattli , and a. bachtold , phys .
lett.*92 * , 216804 ( 2004 ) .
m. s. fuhrer , j. nygard , l. shih , m. forero , young - gui yoon , m. s. c. mazzoni , hyoung joon choi , jisoon ihm , steven g. louie , a. zettl , and paul l. mceuen , science * 288 * , 494 ( 2000 ) . h. w. ch .
postma , m. de jonge , zhen yao , and c.dekker , phys .
b * 62 * , 10653 ( 2000 ) .
j. w. janssen , s. g. lemay , l. p. kouwenhoven , and c. dekker , phys .
b * 65 * , 115423 ( 2002 ) .
i. e. dzyaloshinskii and a. i. larkin , zh .
fiz.*65 * , 411 ( 1973 ) [ so .
jetp * 38 * , 202 ( 1974 ) ] .
s. das sarma and e. h. hwang , phys .
b * 54 * , 1936 ( 1996 ) .
r. l. schult , d. g. ravenhall , and h. w. wyld , phys .
b * 39 * , 5476 ( 1989 ) .
j. p. carini , j. t. londergan , k. mullen , and d. p. murdock , phys .
b * 46 * , 15538 ( 1992 ) .
k. kazymyrenko , b. douot , phys .
b * 71 * , 075110 ( 2005 ) .
h. dickinson and s. doniach , phys .
b * 47 * , 11447 ( 1993 ) .
a. h. castro neto and f. guinea , phys .
lett.*80 * , 4040 ( 1998 ) .
x. j. zhou , p. bogdanov , s. a. kellar , t. noda , h. eisaki , s. uchida , z.hussain , and z.x .
shen , science * 286 * , 268 ( 1999 ) .
a. odintsov , phys .
lett.*85 * , 150 ( 2000 ) .
a. odintsov and hideo yoshioka , phys .
b * 59 * , 10457 ( 1999 ) .
jarillo - herrero , s. sapmaz , c. dekker , l. p. kouwenhoven , and h. s. j. van der zant , nature * 427 * , 389 ( 2004 ) .
m. radosavljevic , j. appenzeller , ph .
avouris , and j. knoch , appl .
lett . * 84 * , 3693 ( 2004 ) .
s. g. lemay , j. w. janssen , m. van den hout , m. mooij , m. j. bronikowski , p. a. willis , r. e. smalley , l. p. kouwenhoven , and c.dekker , nature ( london ) * 412 * , 617 ( 2001 ) .
b. j. leroy , i. heller , v. k. pahilwani , c.dekker , and s. g. lemay , nano lett . * 7 * , 2937 ( 2007 ) . | we study a set of crossed 1d systems , which are coupled with each other via tunnelling at the crossings .
we begin with the simplest case with no electron - electron interactions and find that besides the expected level splitting , bound states can emerge .
next , we include an external potential and electron - electron interactions , which are treated within the hartree approximation .
then , we write down a formal general solution to the problem , giving additional details for the case of a symmetric external potential . concentrating on the case of a single crossing
, we were able to explain recent experinents on crossed metallic and semiconducting nanotubes [ j. w. janssen , s. g. lemay , l. p. kouwenhoven , and c. dekker , phys .
rev .
b * 65 * , 115423 ( 2002 ) ] , which showed the presence of localized states in the region of crossing . |
we start by recalling the operation of the scanning tunneling microscope ( stm ) to measure the ldos , @xmath3 , at frequency @xmath2 , in some region of a surface . for a typical measurement , the tip distance
is held constant while the sample bias , @xmath4 , is changed , leading to a steady - state current given by @xcite @xmath5 the ldos is found from the differential conductance , @xmath6 it is well known that this expression neglects the voltage dependence of the electronic transmission from tip to sample .
interpretational issues notwithstanding , this limits the accurate extraction of the native electronic density of states to the linear response regime , as not all changes in the current are due to changes in the ldos . in particular
, when the system has strong many - body interactions , such as a poorly screened electronic impurity , the applied bias will disturb the natural state of the surface by disrupting the nearby electron density . here
, we suggest that these effects can be limited by an alternative setup . rather than increasing an applied bias , to interrogate the many - body system , @xmath1
, we propose to use a narrow band probe , @xmath0 , that scans in energy . in a paradigmatic , one - dimensional setup ,
the current into a narrow reservoir of bandwidth @xmath7 at a frequency offset @xmath2 is proportional to the ldos ( how many particles at a given frequency ) , @xmath8 so long as @xmath9 is small relative to variations of @xmath10 . both empty and
filled states can be probed in this way by changing the filling of @xmath0 .
this setup requires tunability of @xmath0 : its chemical potential , occupation , bandwidth / hopping , and contact magnitude / location with @xmath1 need to be adjustable without compromising its non - interacting behavior . in other words , this setup is naturally suited to the capabilities provided by ultra - cold atoms in artificial lattice potentials .
a schematic of a magnetically trapped implementation of this setup is shown in fig .
[ fig : expdiagram ] . states using an initially empty @xmath0and ( * b * ) unoccupied @xmath1 states using an initially filled @xmath0in a 1d lattice as a function of @xmath2 , the shift in onsite frequency of @xmath0 , and the time @xmath11 . here , the total lattice length , @xmath12 , is @xmath13 , the bandwidth of the probe is @xmath14 s@xmath15 , and the dotted lines indicate the region averaged for the current , @xmath16 ms , @xmath17 ms@xmath18 $ ] ( to minimize the influence of transient currents and edge effects ) .
( * c * ) the ldos for @xmath1 determined from the average current and eq .
( displayed with an offset for clarity ) .
the shaded regions represent the error calculated using the standard deviation of the current in the region @xmath19 combined with a broadening error of size 2@xmath9 .
the fermi level , @xmath20 , is at the crossover point of the two measurements .
( * d * ) integrated error of the ldos versus total lattice size and ( * e * ) versus averaging time , @xmath21 , ( blue solid line is @xmath22 , green dashed line is @xmath23 ) . the baseline error ( when @xmath24 , the dotted line in ( * d * ) and ( * e * ) ) is set by the finite @xmath25 , which broadens the actual ldos ( see eq . in the methods ) . for short lattices and times ,
the error in the ldos is already small .
thus , even modest cold - atom systems or numerical many - body calculations can effectively reconstruct the ldos .
[ fig : avecurrnonint ] ] ms@xmath15 , @xmath26 ms@xmath15 , @xmath27 ms@xmath15 ( from top to bottom ) , and total size @xmath22 .
( * b * ) the resulting ldos for @xmath28 ms@xmath15 , @xmath26 ms@xmath15 , @xmath27 ms@xmath29 , displayed with an offset for clarity , using the same averaging procedure as performed as in fig .
[ fig : avecurrnonint ] , with the shaded regions indicating the standard deviation of the current combined with the probe broadening error .
note that a steady - state still forms as the system ceases to be a true insulator once it is connected to the non - interacting probe . as @xmath30 increases , a gap forms between the occupied and unoccupied bands , in addition to causing a pronounced broadening of both of the bands .
the dots on the figure indicate the extent of the bands as found using a mean - field approximation ( see eq . in the methods ) .
[ fig : avecurrintmott],title="fig : " ] ms@xmath15 , @xmath26 ms@xmath15 , @xmath27 ms@xmath15 ( from top to bottom ) , and total size @xmath22 .
( * b * ) the resulting ldos for @xmath28 ms@xmath15 , @xmath26 ms@xmath15 , @xmath27 ms@xmath29 , displayed with an offset for clarity , using the same averaging procedure as performed as in fig .
[ fig : avecurrnonint ] , with the shaded regions indicating the standard deviation of the current combined with the probe broadening error .
note that a steady - state still forms as the system ceases to be a true insulator once it is connected to the non - interacting probe . as @xmath30 increases , a gap forms between the occupied and unoccupied bands , in addition to causing a pronounced broadening of both of the bands .
the dots on the figure indicate the extent of the bands as found using a mean - field approximation ( see eq . in the methods ) .
[ fig : avecurrintmott],title="fig : " ] as the interacting system and @xmath0 as the probe .
the ldos calculation was repeated for each site @xmath31 .
( * b * ) the time - dependent current shows the formation of a steady - state flow into @xmath0 when attached to the center of @xmath1 .
( * c * ) @xmath1 has an inhomogeneous interaction profile : a linear decrease from @xmath32 ms@xmath15 on one end to @xmath33 ms@xmath15 on the other .
( * d * ) the ldos for occupied and unoccupied states as a function of frequency offset @xmath2 and contact lattice position @xmath34 .
the weakly - interacting region of the system allows more particles to occupy lower frequency states , while the more strongly - interacting side forces the open states well beyond the fermi level . in addition
, a superimposed even - odd effect is visible , which is due to the finite size of the lattice , creating oscillations from the boundary .
[ fig : avecurrintinhomo],title="fig : " ] as the interacting system and @xmath0 as the probe .
the ldos calculation was repeated for each site @xmath31 .
( * b * ) the time - dependent current shows the formation of a steady - state flow into @xmath0 when attached to the center of @xmath1 .
( * c * ) @xmath1 has an inhomogeneous interaction profile : a linear decrease from @xmath32 ms@xmath15 on one end to @xmath33 ms@xmath15 on the other .
( * d * ) the ldos for occupied and unoccupied states as a function of frequency offset @xmath2 and contact lattice position @xmath34 .
the weakly - interacting region of the system allows more particles to occupy lower frequency states , while the more strongly - interacting side forces the open states well beyond the fermi level .
in addition , a superimposed even - odd effect is visible , which is due to the finite size of the lattice , creating oscillations from the boundary .
[ fig : avecurrintinhomo],title="fig : " ] the hamiltonian , @xmath35 , for this setup is @xmath36 where @xmath9 and @xmath2 are the hopping and _ relative _ onsite frequencies , respectively ( i.e. , the trapping frequencies for a cold - atom system , which can be controlled via magnetic potentials ) and @xmath37 ( @xmath38 ) are the creation ( annihilation ) operators of site @xmath34 .
@xmath39 is the many - body hamiltonian of @xmath1 .
@xmath40 is the contact hamiltonian between @xmath1 and @xmath0 , which we also take to having hopping frequency @xmath9 , which gives a weak contact to a narrow band .
a laser sheet prevents contact between @xmath1 and @xmath0 while they are loaded with atoms and cooled to their independent ground states .
one can also excite @xmath1 by a quench or some other process ; the operational definition of the ldos also extends its applicability to nonequilibrium conditions , in addition to many - body systems . at @xmath41 ,
the sheet is removed .
then the composite system evolves , giving rise to the current @xmath42 , which yields the ldos at frequency @xmath2 according to eq . .
the measurement is repeated for each @xmath2 of interest .
( blue x s ) below the fermi level gives the total particle occupation , which matches that found from a direct measurement of @xmath43 , the occupation number , from the numerical calculation of the ground state ( green line ) .
the formation of a steady state current then provides not only a measure of the occupied and unoccupied states but also yields the real space occupation .
the error bars represent the propagated error from the broadening and standard deviation as in the previous figures .
[ fig : occupation ] ] as noted above , the experimental realization of this approach requires tunability of the hopping frequencies ( the probe bandwidth ) and the onsite frequencies ( the probe offset ) , among other parameters .
the trap depth and spacing control the hopping and this can be tuned via the magnitude of the trapping potential .
the onsite frequency is related to the ground state of each well , which can be modified by controlling the shape of the potential in each well .
this potential can be created by atom chips with magnetic fields generated by patterned circuits of magnetic materials @xcite or through spatial light modulation ( slm ) techniques @xcite , as illustrated in fig .
[ fig : expdiagram ] . in @xmath1 ,
the interactions can be controlled by using a feshbach resonance @xcite to alter the scattering length , allowing for simulation of different types of many - body systems , such as spatially inhomogeneous or varying strength interactions ( all the way down to the non - interacting limit ) . by engineering a series of magnetic potentials with differing onsite energies and contact locations
, one can measure the ldos as a function of frequency at any location within @xmath1 . with this method
, we examine a system of interest with hamiltonian @xmath44 where @xmath45 is the characteristic tunneling frequency , @xmath46 is the number operator , and @xmath47 is the interaction frequency .
the two components may refer to the spins of electrons or internal states of ultracold atoms .
when inter - spin interactions are present , @xmath48 is expanded in a similar manner to include spin . for numerical calculations
, we use @xmath49 ms@xmath15 , a typical cold - atom tunneling frequency @xcite .
we first examine a spin - polarized , non - interacting system @xmath1 : @xmath50 for all @xmath34 , which can be both solved exactly for the current @xcite and the result compared to the exact ldos ( see the methods ) .
figure [ fig : avecurrnonint]c shows the reconstruction of the ldos using a finite lattice and time average .
there is quantitative agreement between the exact ldos and that from the energy - resolved probe ( using eq . ) .
the dominant source of error in the ldos is the finite probe bandwidth @xmath9 , not the length of the lattice or the time of the average ( fig .
[ fig : avecurrnonint]d and e ) .
this demonstrates that cold - atom systems or many - body simulations are well suited to implement this method , even though they are limited to finite lengths and times . as the total lattice length , @xmath51 , the averaging time , @xmath52 , and the probe bandwidth , @xmath53 ( in this order )
, the exact ldos would be recovered .
we now apply the same approach to an interacting system with a constant @xmath54 .
figure [ fig : avecurrintmott ] shows the ldos of a mott - insulator like state . as the interaction strength increases , the band splits and a gap forms between the occupied and unoccupied bands , as is typical for a mott insulator .
however , the fermi frequency , @xmath55 , and the occupied band is shifted to higher frequency .
the observations also agree with predictions from green s function calculations in the hopping - only and interaction - only limits ( see the methods ) . in ref .
@xcite , it was found that there is a filling - dependent , conducting - to - nonconducting transition as a function of @xmath30 for interacting - induced transport : an inhomogeneous quench in @xmath30 , where the interaction strength is taken from @xmath56 to a finite value for half the lattice , drives particles from that half of the lattice to the other ( non - interacting ) half so long as @xmath30 is not too strong and the filling not too large .
figure [ fig : avecurrintmott]b demonstrates that it is this shift of the occupied bands to higher frequency that aligns occupied states in the interacting side to open states in the non - interacting side , allowing particles to flow . as @xmath30 is increased further , eventually only the tail of the occupied band is aligned with open states , thus giving a decreasing but nonzero current solely due to many - body interactions .
when the spatial dependence of the density of states is of interest , the probe can be used in a regime analogous to a `` scanning - mode '' : the probe can be coupled to the system at any point along its length through a right angle construction , as shown in fig .
[ fig : avecurrintinhomo]a .
examining a system with spatially inhomogeneous interactions
e.g . , a linear decrease in the interaction strength @xmath30 , fig .
[ fig : avecurrintinhomo]b can determine both how the particle density shifts in space and energy .
figure [ fig : avecurrintinhomo]c shows the occupied and unoccupied ldos of this inhomogeneous lattice as a function of position .
the spatial decrease of the interactions forces particles to the region with small interactions , where at the very end the lattice has an ldos similar to a non - interacting system . just near the non - interacting boundary , however , a large peak in occupied density of states forms , i.e. , states pinned well below the fermi level . on the interacting side , the number of particles is small with an ldos just below the fermi level .
a superimposed even - odd effect is visible , which is due to finite lattice effects , creating oscillations away from the boundaries .
there is a direct correspondence between the particle occupation and the integrated occupied ldos , as shown in fig .
[ fig : occupation ] .
we conclude by noting that the density of states is a central concept in our description of matter . we have provided an operational definition of the ldos for many - body systems , applicable to in- and out - of - equilibrium cases , fermionic or bosonic systems , etc .
the core principle is that for a current to flow whether in a steady state or not into an empty , narrow probe band , there must be occupied states at that energy ( similarly for a full narrow band and unoccupied states ) .
the measurement is resolved in both energy and space , as well potentially in other characteristics ( e.g. , spin - resolved ) .
we demonstrate that a cold - atom setup will allow for the measurement of this operational definition : the many - body ldos can be extracted with minimal disturbance to the system . while cold - atom systems allow for tunability , issues still can arise regarding , e.g. , the orbital character of the local states and higher energy excitations . unlike solid - state systems , however , the effect of these issues can be reduced in this setup . a related approach , where tunable cold - atom systems are used to more controllably i.e . , with less disturbance to the native state
implement eqs . and , is also beneficial , but the narrow band setup we propose minimizes the total current flowing and all other disturbances , more so than that of the stm . the energy - resolved atomic scanning probe will complement the quantum - gas microscope @xcite , which resolves the spatial location of atoms . moreover , there is also recent interest in using quantum gases to simulate the physics of the early universe @xcite , for which the energy - resolved probe can help shed light on the nature of high - energy excitations and symmetry breaking .
the ability to probe both experimentally and numerically the undisturbed density and occupation of states as a function of energy and position illuminates the structure of many - body systems and will help advance the theoretical description of underlying physical processes .
the non - interacting lattice is exactly solvable for both the current and the ldos .
for the current , we need the retarded green s function for a semi - infinite , one - dimensional lattice with onsite frequency @xmath2 and hopping frequency @xmath57 , which is @xcite @xmath58 .\ ] ] using this expression , the current for the infinite lattice , @xmath51 , is given by the landauer formula @xcite , @xmath59 t(\omega),\ ] ] where @xmath60 are the initial particle distributions ( fermi - dirac distributions or completely filled / empty ) in @xmath61 and @xmath62 is the transmission coefficient @xmath63 { \operatorname{re}}\left [ \sqrt{4 \omega_{\mathcal{s}}^2 - \omega^2 } \right ] } { \left| \mu + \im \sqrt{4 \omega_{\mathcal{p}}^2 - ( \omega-\mu)^2 } + \im \sqrt{4 \omega_{\mathcal{s}}^2 - \omega^2 } \right|^2 } .\ ] ] implementing the setup with @xmath64 gives the reconstructed ldos as simply a sum of the occupied and unoccupied states , and also directly yields the fermi level , @xmath20 . in the main text ,
the system of interest @xmath1 is half filled and in its zero temperature ground state . here , we show , for simplicity , how the ldos of a fully filled non - interacting system can be mapped out .
the current in this case is @xmath65 when @xmath66 .
thus , @xmath67 for @xmath68 .
. gives the exact ldos , @xmath69 , for a non - interacting @xmath1 .
the remaining terms in the full expression broaden the reconstructed ldos by approximately the probe bandwidth , @xmath7 , which thus has to be small enough to discriminate features in the ldos of @xmath1 and , when using a cold - atom setup , the bandwidth should be large enough to get an appreciable current .
the local density of state ( ldos ) is defined as @xcite @xmath70 here , @xmath71 is the @xmath43-th eigenfunction of the full hamiltonian with eigenvalue @xmath72 ( in units of @xmath73 ) . for systems in the thermodynamic limit
, the ldos is related to the real - space green s function by @xcite @xmath74.\ ] ] these two equations give the same expression for a noninteracting semi - infinite lattice .
the energy dispersion is @xmath75 $ ] and @xmath76 $ ] with @xmath77 .
thus , @xmath78 @xmath79 here , @xmath80 is the value when @xmath81 is satisfied .
the green s function is similar to eq . .
explicitly , @xmath82 , so @xmath83 .
both agree with eq . and the landauer formula .
the green s function for the hubbard model in the mean - field approximation is @xcite @xmath84 here , we focus on the 1d case with @xmath85 and chemical potential @xmath86 .
@xmath87 denotes the chosen spin species , and @xmath43 is the filling factor .
the poles of the green s function lead to the energy dispersion , which has the form @xmath88 with @xmath89 giving the two bands : @xmath90 the retarded green s function in real space is @xmath91 . since the green s function is for a uniform , translational invariant system , we may choose @xmath92 . by using the identity @xmath93 with @xmath94 denoting the cauchy principle integral
, the ldos for one spin component is @xmath95 .
\notag\end{aligned}\ ] ] here , @xmath96 satisfies @xmath97 .
one may view the prefactors as the weight of wavefunctions at @xmath98 and the delta functions pick the frequencies that matches the two energy bands . in the isolated - site limit
where @xmath99 , the green s function is exactly solvable and on a selected site it is @xmath100 $ ] . by analytic continuation @xmath101 , we obtain @xmath102 .
then , @xmath103\delta(\omega+\mu_{\mathcal{s}})+(n/2)\delta(\omega+\mu_{\mathcal{s}}-u)\}$ ] .
thus , there are two peaks at @xmath104 and @xmath105 .
away from the isolated - site limit , the two peaks broaden into two bands separated by @xmath30 , and this agrees with our observation shown in fig . [ fig : avecurrintmott ] . in order to extract the ldos from real - time measurements on finite lattices
, the current must be averaged over a finite time , i.e. , the current will be in a quasi - steady state @xcite .
the estimate of the ldos is thus @xmath106 as @xmath107 , this will converge to the true steady state . as we demonstrate , already for very small lattices an accurate ldos can be obtained and thus only modest resources are needed . the numerical calculations are performed as follows : for the non - interacting system , the current was found from the numerical integration of the equations of motion @xcite .
the transient current when the probe is placed in contact with the system is damped on the characteristic tunneling time and the recurrence time is proportional to the lattice size , which determines both the lower and upper limits to the time region @xmath19 given a finite lattice length .
we use @xmath108 ms , @xmath109 ms@xmath18 $ ] , where @xmath12 is the lattice length . we can also define an exact error for the non - interacting case , @xmath110 ^ 2 \ ; d\mu}}{\sqrt{\int { d_{\text{ex}}}(\mu)^2 \ ; d\mu } } , \ ] ] where @xmath111 is the exact ldos . the uncertainty due to probe broadening is due to contributions to the current from anywhere within the probe bandwidth of @xmath112 , which results in an error , @xmath113 , of @xmath114\ ] ] for the positive @xmath2 side and , @xmath115 , @xmath116\ ] ] for the negative side .
this is combined with the standard deviation , @xmath117 , from the time - dependent current for a total error of @xmath118 for the interacting systems , we perform time - dependent , density matrix renormalization group calculations @xcite within the itensor tensor product library @xcite . in all simulations ,
the time step was reduced until the calculation converged with respect to energy conservation and the maximum matrix bond dimension was allowed to increase without bound .
the energy cutoff was set to @xmath119 .
the averaging is done in the region @xmath108 ms , @xmath109 ms@xmath18 $ ] , as with the non - interacting case .
daniel gruss acknowledges support under the cooperative research agreement between the university of maryland and the national institute of standards and technology center for nanoscale science and technology , award 70nanb10h193 , through the university of maryland .
massimiliano di ventra acknowledges support from the doe grant no .
de - fg02 - 05er46204 .
m.z . proposed the project , d.g . and c.c.c .
performed analytical calculations and d.g . performed numerical calculations
all authors helped to clarify the ideas and wrote the manuscript .
correspondence and requests for material should be addressed to m.z .
( [email protected] ) .
the authors declare no competing financial interests . | the density of states is a concept that is ubiquitous in classical and quantum physics , since it quantifies the energy distribution of states available in a system .
spectroscopic means allow its measurement over the entirety of a system s energy spectrum , but do not generally provide spatial resolution .
moreover , scanning probes measure the density of states locally at the position of the probe tip , but typically do not have access to the whole spectrum . here
, we show how the _ local _ density of states over the _ whole _
energy spectrum can be measured in atomic gases .
this probe provides a simple , yet quantitative operational definition of a local density of states for both interacting and non - interacting systems as the rate at which particles can be siphoned from the system of interest by a narrow energy band of non - interacting states .
we demonstrate that ultra - cold atomic lattices are a natural platform for implementing this concept and visualize the energy and spatial dependence of the atom density in inhomogeneous , interacting lattices .
ultra - cold atomic lattices are increasingly employed in the simulation of many - body phenomena @xcite . additionally , the scanning tunneling and atomic force microscopes @xcite are arguably the most versatile instruments for probing materials , molecules , structures , and devices . from imaging surface defects to probing local electronic structure to testing novel devices
@xcite , they have given the means to comprehend structures and behavior at their most basic level .
increasingly , tunable atomic - based systems that mimic , e.g. , materials and electronics have been used to probe physical behavior in a way not possible with traditional techniques . in this direction , recent experimental work demonstrated the expansion of an atomic cloud @xcite , transport of quantum correlations and phase coherence @xcite , and the presence of quantum effects in neutral matter transport @xcite .
this approach holds great promise in advancing our understanding of the natural world , especially many - body and out - of - equilibrium behavior . in this article
, we give the local density of states ( ldos ) an operational definition in terms of the rate at which particles can be distilled from a many - body system at a specific energy .
the basic premise is that for a system to sustain a current into a narrow band reservoir , it must have a finite density of states at that energy .
we show how to implement this approach and apply it to many - body systems . as we seek to harness and control quantum behavior for a new generation of technologies , such as quantum computers @xcite , spintronic devices @xcite , and beyond
, this approach to quantifying physical phenomena will yield an unprecedented level of detail and help to determine the underlying theoretical description of many strongly interacting systems .
, is placed in contact with a non - interacting , narrow band probe , @xmath0 .
( * a * ) the cold - atom traps are fabricated via a controlled geometry of patterned magnets , which determine the spatially - inhomogeneous coupling rates and trapping frequencies . a laser sheet ( red )
is initially present to prevent the flow of atoms from @xmath1 to @xmath0 until a measurement is performed .
( * b * ) this system is modeled by a 1d lattice with inhomogeneous hopping rates and onsite frequencies in @xmath1 and @xmath0 .
( * c * ) a representation of the setup in frequency space shows the narrower range of frequencies and offset , @xmath2 , of the probe .
[ fig : expdiagram ] ] |
over the last few years , observations of galaxies at @xmath4 have opened up a new window into the reionization epoch , turning it into the latest frontier both for observational and theoretical studies of galaxy formation .
planned @xcite and existing wide - area narrowband searches for @xmath0 objects such as the subaru deep field @xcite , the large area lyman - alpha survey @xcite , the chandra deep field - south @xcite , and the hubble ultra deep field @xcite are now combining with lyman - alpha dropout searches @xcite , targeted searches near lensing caustics in galaxy clusters @xcite and occasionally serendipity @xcite to uncover star - forming galaxies from the reionization epoch in significant numbers .
a question immediately raised by this new stream of observations is , what are the physical properties of these early galaxies ?
optimally , one would determine properties such as the stellar mass , star formation rate , and metallicity directly from high - quality spectra , but at present this is infeasible for such faint systems .
hence properties must be inferred from photometry alone , occasionally augmented by emission line information .
this requires making some poorly constrained choices for the intrinsic galaxy properties . a commonly applied method known as spectral energy distribution ( sed ) fitting
involves generating an ensemble of population synthesis models under a range of assumptions for the intrinsic nature of the object , and then finding the set of assumptions that best reproduces a given galaxy s observed photometry ( e.g. * ? ? ? * ; * ? ? ?
the physical properties that yield the lowest @xmath5 model are then forwarded as the most probable values , sometimes with little attention to statistical uniqueness or robustness ( see * ? ? ?
* for a nice exploration of such issues )
. amongst the various assumptions used in sed fitting , the one that is often least well specified and produces the widest range in final answers is the galaxy s star formation history ( sfh ) . with no prior information ,
common practice is to use simple sfhs with one free parameter such as constant , single - burst , or exponentially - decaying , which in aggregate are assumed to span the range of possible sfhs for a given galaxy . indeed , in most cases all one - parameter sfhs yield plausible results , though the parameters obtained and quality of fits in each case can vary significantly . if it were possible to narrow the allowed range of sfhs through independent considerations , physical parameters could in principle be more precisely determined .
one approach for constraining sfhs a priori is to incorporate information from currently favored hierarchical structure formation models .
as we will discuss in this paper , hydrodynamic simulations tend to produce a relatively narrow range of star formation histories for early galaxies .
their galaxies sfhs tend to follow a generic _ form _ at these early epochs , best characterized as a constantly - rising sfh .
this form is broadly independent of cosmology , feedback assumptions , or other ancillary factors , and is furthermore distinct from any one - parameter models commonly used today .
a primary aim of this paper is to test whether this relatively generic sfh form is consistent with observations , and if so , what the implication are for the physical properties of high - redshift galaxies . despite impressive recent successes in understanding cosmology and large - scale structure in our universe ( e.g. * ? ? ?
* ; * ? ? ?
* ) , many uncertainties remain in our understanding of galaxy formation .
several recent papers have tested models of high-@xmath6 galaxy formation by comparing them to observed bulk properties such as luminosity functions at rest - frame uv and ly@xmath7 wavelengths .
these comparisons have shown that such models are broadly successful at reproducing observations , under reasonable assumptions for poorly constrained parameters such as dust extinction @xcite . while this broad success is encouraging , it is subject to some ambiguousness in interpretation , because the properties of individual galaxies are not being compared in detail .
one could envision situations in which a model reproduces an ensemble property of galaxies but not the detailed spectra of individual objects . as an example
, it was forwarded by @xcite that lyman break galaxies at @xmath8 are actually merger - driven starbursts , in contrast to many other models predicting them to be large quiescent objects . despite quite different sfhs , both models reproduced many of the same bulk properties such as number densities and clustering statistics . for @xmath9 galaxies where statistics are currently poor
, such degeneracies can hamper interpretations of bulk comparisons of observations to models .
a complementary set of constraints on galaxy formation models may be obtained by comparing models to the individual spectra of observed galaxies . in practice , for high-@xmath6 systems ,
photometry over a reasonably wide set of bands must substitute for detailed spectra .
such comparisons of models to data would move towards more precise and statistically robust analyses that do not rely on having a large ensemble of objects .
this last aspect is critical , because the very earliest observed objects that may provide the greatest constraints on models will in practice always be few in number and detected only at the limits of current technology . in short ,
what is desireable would be a tool to compare models and observations of high - redshift galaxies that ( 1 ) employs reasonably generic predictions of current galaxy formation models ; ( 2 ) provides a quantitative and robust statistical assessment of how well such models reproduce observations ; ( 3 ) yields information on the physical properties of galaxies under various assumptions ; ( 4 ) obtains such information based solely on observed photometry ; and ( 5 ) does all this on a galaxy - by - galaxy basis rather than relying on having a large statistical sample of observed galaxies . in this paper
we introduce such a tool , called spoc ( simulated photometry - derived observational constraints ) .
spoc takes as its input the photometry ( with errors ) of a single observed galaxy along with an ensemble of model spectra drawn either from simulations or generated using one - parameter sfhs .
the output is probability distributions of physical parameters derived using a bayesian formalism , along with goodness - of - fit measures for any given model .
the probability distributions give quantitative constraints on the galaxy s physical properties , while the goodness - of - fit can be used to discriminate between models and determine whether a given model ( be it simulated or one - parameter sfhs ) is able to provide an acceptable fit to that galaxy s photometry . after introducing and testing spoc , we apply it to a sample of six @xmath2 galaxies from the literature that have published near - infrared photometry .
we show that in five of six cases , the simulated galaxies fit observations at least as well as one - parameter sfhs . since there is no guarantee that simulations produce galaxy sfhs that actually occur in nature , the fact that good fits are possible shows that the existence of the majority of observed @xmath0 galaxies is straightforwardly accommodated in current galaxy formation models .
however , in one case , we find that simulated galaxies provides a much poorer fit than can be obtained with one - parameter sfhs , as burstier sfhs provide a much better fit than can be obtained from any simulated galaxies . at face value
, this implies that our simulations can not yet accommodate the full range of observed galaxies , and that some physical process may be missing , although we will explore alternate interpretations . for each galaxy we also present the best - fit physical parameters , with uncertainties , obtained using each model sfh .
the simulations provide significantly tighter constraints than the full range of one - parameter sfhs , as expected based on their relatively small range of sfhs produced .
these values can therefore be regarded as predictions of our simulations that may be tested against future observations .
[ sec : spoc ] introduces spoc , detailing our bayesian formalism and discussing systematic uncertainties .
[ sec : models ] presents the simulations and the one - parameter models that will be used as the template library for spoc .
[ sec : performance ] discusses what drives the inferred physical properties in the context of our simulations , and shows that spoc accurately recovers the physical properties of simulated galaxies .
[ sec : kesr ] explores the best - fit parameters of one observed reionization - epoch galaxy in detail , and compares with results from traditional one - parameter sfh models .
[ sec : othergals ] repeats the previous comparison for a larger set of observed galaxies , highlighting the variety of interesting results that spoc obtains .
finally , in [ sec : summary ] we present our conclusions .
pedagogical explanations of sed fitting techniques have been presented elsewhere @xcite , so we refer the reader there for more detailed discussion of those aspects . here
we provide some basic insights and notes . clearly , the amount of physical information that can be inferred from available data depends on the quantity and quality of the data . for some high-@xmath6 galaxies , only narrow - band photometry and rest - frame ultraviolet ( uv ) spectroscopy
are available ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . for others , an emission - line measurement and 13
rest - uv broad bands are available ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
studies employing the lyman dropout technique in the optical must further contend with the possible presence of low - redshift interlopers @xcite and large uncertainties from dust extinction .
nevertheless , some interesting constraints can be placed on the underlying physical properties of the sources from solely rest - uv data @xcite . with the addition of rest - frame optical data , e.g. from _
spitzer s _ infrared array camera ( irac ) , it becomes possible to obtain simultaneous constraints for the stellar mass , star formation rate ( sfr ) , dust extinction , and redshift using spectral energy distribution ( sed ) fitting techniques @xcite .
the uncertainties inherent in such analyses primarily stem from a poor constraint on the age of the galaxy s stellar population , because the relationship between age and the strength of the telltale balmer break depends on the form of the assumed sfh ( @xcite ; figure [ fig : degeneracy ] ) .
this age uncertainty propagates via a host of degeneracies into increased uncertainties in the inferred stellar mass , sfr , metallicity , and dust extinction , if no priors are assumed on these quantities .
additional uncertainties arise from the unknown form of the appropriate template sed ( e.g. * ? ? ?
* ) and the treatment of stellar evolution assumed by the chosen population synthesis models ( see e.g. * ? ? ?
still , sed fitting offers the most promising approach for determining the physical properties of individual high-@xmath6 galaxies . given this
, how can one employ simulations to improve constraints on sed fitting ?
one can view a numerical simulation as producing a monte carlo sampling of parameter space such that the frequency with which a given set of physical parameters ought to occur is proportional to the number of galaxies in the simulation that are characterized by that set of parameters .
in essence , numerically - simulated galaxies provide implicit priors " for sed fitting , i.e. solutions that are _ a priori _ weighted more heavily because they occur more frequently .
the underlying assumption is that simulated galaxy sfhs represent those occuring in nature .
this is by no means guaranteed , and indeed whether spoc provides an acceptable fit to a given galaxy constitutes a stringent test of the simulation , because a galaxy s spectrum encodes information about its full sfh .
this is the manner in which spoccan provide a test of galaxy formation models based on individual systems .
we now summarize the bayesian statistical method employed in spoc .
our goal is to constrain the stellar mass , sfr , mean stellar metallicity , age , dust extinction , and redshift ( @xmath10 , @xmath11 , @xmath12 , @xmath13 , @xmath14 , and @xmath6 , respectively ) based on available measurements @xmath15 . according to bayes theorem ,
the probability @xmath16 that the measurements @xmath15 correspond to a galaxy with the intrinsic physical parameters @xmath17 ( where a hat indicates a particular value of a parameter ) is given by @xmath18 the prior @xmath19 indicates the relative _ a priori _ probability that a randomly selected galaxy has this particular combination of parameters , and the likelihood @xmath20 indicates the probability of obtaining the measurements @xmath15 for a galaxy characterized by the parameters @xmath21 ; for a given model galaxy and data set @xmath15 this is assumed to be proportional to @xmath22 .
any information regarding the expected distributions of physical properties of the observable galaxies ( such as the stellar mass function ) or relationships between these properties ( such as a mass - metallicity relation ) can be taken into account via a contribution to the prior , and will generally give rise to more precise and possibly more accurate constraints . in this work , we assume uniform priors on @xmath6 and @xmath14 , and we do not assume any dependence between @xmath14 and the other intrinsic physical properties .
we introduce an additional prior @xmath23 to account for any other priors .
for example , when matching observed galaxies against model galaxies derived from the outputs of two cosmological simulations that span different comoving volumes , @xmath23 represents the ratio of the simulation volumes .
after several applications of the product rule , we obtain @xmath24 this is the fundamental equation that spoc evaluates .
generically , one would use equation [ bayes2 ] by beginning with a set of models that uniformly samples the relevant parameter space and then guessing the form of the prior @xmath25 , which now encodes the assumed distribution of intrinsic physical properties of galaxies as a function of redshift . in the high - redshift literature , where little is known about the intrinsic physical properties of the galaxies , it is common to neglect priors altogether ( or , equivalently , to choose the model with the lowest @xmath5 ) or even to introduce them accidentally by not sampling parameter space uniformly .
the difference between this work and that of previous authors is that we account for this prior implicitly by using numerically simulated galaxies as the model set . to see how this works ,
consider how one would use equation [ bayes2 ] in practice .
for simplicity , suppose that we wished to constrain a galaxy s stellar mass and that the mass could only fall within one of two ranges .
if we omitted priors and assumed that the models sample stellar mass uniformly , then the probability that the galaxy s mass falls within a given range would be given by @xmath26 , where the sum is taken over all models whose mass lies within that range and the normalization @xmath27 is chosen so that the sum taken over all models in both ranges equals unity . if we believed that galaxies with masses in one range were , say , twice as common ( and therefore _ a priori _ twice as likely to be the right answer ) as galaxies with masses in the other range , we could account for this via an explicit prior by changing the sum to @xmath28 where @xmath29 for models in the more common range and @xmath30 for models in the less common range ( with @xmath27 of course rescaled ) .
it is clear that an equivalent method to employing this explicit prior would be to generate twice as many models in the more common range , resulting in twice the probability of selecting one of these models .
generalizing this idea , one can view simulated galaxies as a monte carlo sampling of parameter space that naturally produces more models with parameters that are more commonly found .
hence by taking a set of simulated galaxies , generating a library by resampling this set with parameters having uniform priors ( namely , @xmath31 and @xmath6 ) , and using that library to discretely sample the probability distribution in the right - hand side of equation [ bayes2 ] , one can solve equation [ bayes2 ] effectively incorporating the implicit priors given by the simulated galaxies .
this is in essence the spocalgorithm .
a major difficulty with the spoc approach is that there is no guarantee that the simulation predicts the correct distribution of intrinsic properties of galaxies ; in bayesian terms , the priors could be wrong . on some level
this is bound to be the case as we do not account for every process that could _ in principle _ affect galaxies at this epoch ; indeed , no model currently does .
however , our goal is to determine whether our treatment is _ sufficient _ to account for current observations .
if not , then the failures indicate needed improvements to the model .
if our treatment can account for current observations , then the constraints that we derive may be regarded as physically - motivated predictions , subject to verification when more constraining data become available .
the two greatest uncertainties for the input physics present in our current simulations are ( 1 ) numerical resolution manifested either as an inability to account for physical processes that occur on scales that are too small or too rapid ( e.g. merger - induced starburst ) or as a lack of numerical convergence and ( 2 ) the prescription for superwind feedback . in
[ sec : resolution ] we use a simple convergence test to argue that our results do not suffer from numerical resolution limitations . as to our treatment for outflows
, we can estimate the extent of any resulting systematics by comparing results from our three different outflow simulations .
while this does not span the full range of possible feedback mechanisms , the fact that ( as we show in [ sec : winddust ] ) most of the best - fit parameters are insensitive to the choice of wind prescription suggests that outflows do not noticeably alter typical sfhs at a given stellar mass . on the other hand ,
if there are significant physical processes affecting galaxy seds that are not accounted for by our simulation or population synthesis models , then our simulated galaxies may fail to reproduce the observed spectra , or they may mistakenly model nonstellar contributions to the observed sed as starlight . among the possibilities here
are active galactic nuclei ( agn ) , incorrectly modeled thermally pulsating asymptotic giant branch ( tp - agb ) stars @xcite , emission lines , and an inappropriate treatment of dust or igm absorption .
we will argue in [ sec : othergals ] that significant agn contamination is unlikely for the high - redshift objects we will consider here .
the contribution of tp - agb stars is also unlikely to be important partly because we do not model measurements from bands redder than @xmath32 in the rest - frame , and partly because at @xmath33 less than half of the existing stellar mass is more than 200 myr old ( table [ table : many_objs ] ) .
emission lines and incorrectly - modelled igm absorption could in principle affect our results at the 10% level @xcite .
these effects are expected to be similar for the various sfhs investigated because we use the same population synthesis models to model the stellar continuum in each case . regarding dust ,
we have found , in agreement with @xcite , that our results are relatively insensitive to the form of the dust law that we consider ( see [ sec : consistency ] ) .
thus , for the preliminary study in this paper we ignore all of these effects .
another possible problem is that galaxy classes that are rare in reality are likely to be rare in the simulations .
accordingly , if the comoving volume from which the catalog of simulated comparison galaxies is drawn is sufficiently small that a simulated analogue to an observed rare object is neither expected nor found , that object can not directly constrain the model .
for example , our simulations produce no galaxies massive enough at @xmath34 to fit hudf - jd2 , the putative @xmath35 object at @xmath36 reported by @xcite .
although this particular object is likely to be at a lower redshift @xcite , it does illustrate limitations imposed by simulation volume , which could also impact constraints on rare classes such as sub - millimeter galaxies ( e.g. * ? ? ? * ) ( alternatively ,
if such objects are indeed common at @xmath37 then they represent a challenge to our simulations ) . in principle
one could work around this issue by running larger - volume simulations or by deriving the priors from the simulations and then resampling parameter space by hand .
in lieu of these approaches , the simulations utilized must have comoving volumes comparable to the effective volume of the survey in which the object was found .
it may appear overly ambitious to attempt to constrain 6 ( or more ) seemingly independent parameters for a galaxy for which fewer than 6 measurements are available .
however , cosmological simulations allow us to do this because they generically predict that galaxies intrinsic physical parameters are manifestly _ not _ independent ; there are tight predicted correlations between , for example , stellar mass on the one hand and star formation rate and metallicity on the other @xcite . in summary , using simulated galaxies to estimate physical properties is only valid when the dominant emission mechanism is star formation , and when other uncertainties can be carefully analyzed and shown to be negligible . for galaxies at high redshift , such as the ones we consider in this paper , this is believed ( but not guaranteed ) to be true .
however , in the general case these issues must be considered carefully . in turn , the goodness of fit enables constraints to be placed on simulations of galaxy formation , and can highlight missing physics that may be required in order to explain the observed properties of galaxies .
we draw our simulated galaxies from cosmological hydrodynamic simulations run with gadget-2 @xcite , including our improvements as described in ( * ? ? ?
* hereafter od06 ) .
this code uses an entropy - conservative formulation of smoothed particle hydrodynamics ( sph ) along with a tree - particle - mesh algorithm for handling gravity .
heating is included via a spatially uniform photoionizing background @xcite , which is an acceptable approximation for the galaxies that are observed at high redshift owing to the fact that they form in highly overdense regions that undergo local reionization at @xmath38 @xcite .
all gas particles are allowed to cool under the assumption of ionization equilibrium , and metal - enriched particles may additionally cool via metal lines .
cool gas particles are allowed to develop a multi - phase interstellar medium via a subresolution multi - phase model that tracks condensation and evaporation following @xcite .
stars are formed from cool , dense gas using a recipe that reproduces the @xcite relation ; see @xcite for details .
the metallicity of star - forming gas particles grows in proportion to the sfr under the instantaneous recycling approximation .
stars inherit the metallicity of the parent gas particle , and from then on can not be further enriched .
cosmological hydrodynamic simulations that do not include kinetic feedback from star formation invariably overproduce stars ( e.g. , * ? ? ?
* ; * ? ? ?
* od06 ) . because superwinds can affect the physical properties of the simulated galaxies ( e.g. * ? ? ?
* ) , we consider model galaxies from simulations with three different superwind schemes : ( 1 ) a no wind " model that omits superwind feedback ; ( 2 ) a constant wind " ( cw ) model in which all the particles entering into superwinds are expelled at 484 km / s out of star forming regions and a constant mass loading factor ( i.e. the ratio of the rate of matter expelled to the sfr ) of 2 is assumed ( as in the runs of * ? ? ?
* ) ; and ( 3 ) the momentum - driven wind " ( vzw ) model of od06 , in which the imparted velocity is proportional to the local velocity dispersion ( computed from the potential ) and the mass loading factor is inversely proportional to the velocity dispersion @xcite , as inferred from observations of local starbursts @xcite .
this selection is meant to bracket plausible models in order to expose any related systematic uncertainties ; however , owing to the range of successes in comparison with igm metal - line observations obtained by od06 for the vzw model , we focus on this model when the conclusions from the different wind models are broadly similar .
all of our wind models were tested in simulations that assumed the old " wmap - concordant cosmology @xcite having @xmath39 , @xmath40 , @xmath41 , @xmath42 , and @xmath43 .
each of our simulations has @xmath44 particles , with parameters as given in od06 .
we only employ the @xmath45 and @xmath46 simulations from od06 , as the @xmath47 runs did not have any galaxies large enough to be observable at @xmath9 .
an additional set of simulations ( the jvzw " model ) were run using our preferred wind model with the 3rd - year wmap cosmology @xcite , namely @xmath48 , @xmath49 , @xmath50 , @xmath51 , and @xmath52 . due to an error in the initial conditions generation ,
the power spectrum index was set to @xmath53 rather than the currently - favored @xmath54 ; however , this has little impact on our results as we will show that they are insensitive to such differences in cosmology .
there is a slight change in the wind model for jvzw versus vzw , in that jvzw has a smaller mass loading factor by a factor of two - thirds compared to vzw ( in the terminology of od06 , @xmath55 km / s ) in order to compensate for the lower collapse fraction at high redshift in the new cosmology .
in addition to @xmath45 and @xmath46 box sizes , we also run a @xmath56 box with the jvzw model to sample the bright end of the mass function in order to better constrain some observed galaxies that we will consider in [ sec : othergals ] .
we found that model galaxies from the jvzw simulations have bulk properties that are similar to that from vzw .
for this paper we will compute all luminosity distances assuming the new 3rd - year wmap cosmology .
we identify galaxies using spline kernel interpolative denmax ( see @xcite for a full description ) .
we only consider galaxies with stellar masses exceeding 64 star particles , which represents a converged sample in terms of both stellar mass and star formation history @xcite . according to this criterion
, our @xmath45 simulation volumes resolve galaxies with stellar mass @xmath57 . for this work ,
the most important output of the simulations is the set of sfhs corresponding to the resolved galaxies in each simulation at the various redshift outputs .
we obtain the rest - frame spectrum for each star formation event in a given galaxy at the time of observation by interpolating to the correct metallicity and age within the @xcite models , assuming a chabrier imf . summing these up
, we obtain the galaxy s intrinsic rest frame spectral energy distribution ( sed ) .
we consider the following prescriptions for dust reddening : the @xcite starburst dust screen , the @xcite embedded star formation law , the @xcite small magellanic cloud bar law , and the @xcite milky way law .
we account for igm absorption bluewards of rest - frame ly@xmath7 using the @xcite prescription
. the @xcite law may be less appropriate for @xmath9 than at lower redshifts because the universe is completing reionization at this epoch .
indeed , @xcite found that they were able to improve the quality of their fits to the seds of two reionization - epoch galaxies by simply doubling the optical depth predicted by @xcite .
however , they also found that the best - fit derived parameters are relatively insensitive to the igm treatment .
thus , for simplicity we retain the @xcite treatment without modification .
-0.0 in to date , efforts to use sed - fitting to infer the physical properties of high - redshift galaxies have generally employed some combination of constant , exponentially decaying , and single - burst star formation histories in order to span the presumed range of possibilities . in general
, it has been found that the stellar mass , sfr , and redshift of a galaxy can be fairly well - constrained via this technique while the age , metallicity , and dust extinction can not .
much of the gain in precision that results from using simulated galaxies in sed - fitting results from the relatively small range of sfhs that actually occur in the simulations .
for example , the solid black curves in figure [ fig : sfh ] show the sfhs of the 3 galaxies from the vzw simulation that yield the best fits to the @xmath58 galaxy abell 2218 kesr , which we will discuss extensively in
[ sec : kesr ] .
the sfhs have been sampled in 20-myr bins and smoothed with a 100-myr tophat in order to make the plot more readable .
all 3 galaxies begin forming stars at @xmath59 and exhibit a sfr that is generally rising .
an examination of simulated sfhs at these redshifts shows that steadily rising sfhs are typical . for this reason we consider a constantly - rising model sfh in this work ; as we will see , the constantly - rising model reproduces most closely the constraints obtained from the simulated galaxies ( figure [ fig : degeneracy ] ) .
this model has to our knowledge not been investigated before . in order to facilitate comparison with much of the available sed - fitting work that is available in the literature ,
we investigate three one - parameter model sfhs for each galaxy in addition to simulated sfhs , as described below : * _ exponentially decaying sfr _ we generate models with sfr proportional to @xmath60 . we use four values of @xmath61 logarithmically spaced between 10 and 795 myr , roughly the age of the universe for our most distant object . each of these sfhs is sampled at 23 ages @xmath13 evenly spaced between 10 and 1000 myr . * _ constant sfr _ we generate models that have been forming stars at a constant rate @xmath11 for @xmath13 myr .
for @xmath13 we sample 41 ages that lie between 10 and 1000 myr , and for @xmath11 we sample 45 sfrs that lie between 0.2 and 30.0 @xmath62 . * _ constantly rising sfr _ in the constantly rising sfh , a galaxy s sfr is proportional to its age . while a rising sfh can clearly not be maintained for all galaxies until low redshifts , it arises fairly generically for high - redshift galaxies in hydrodynamic simulations @xcite .
we generate models in which each galaxy s sfr has been rising at a constant rate for @xmath13 myr , where for @xmath13 we have sampled 41 ages that lie between 10 and 1000 myr . for each star formation history ,
we have generated models with masses in the range @xmath63 $ ] and metallicities @xmath64 .
these sfhs are then put through the spoc formalism , in order to determine the probability distribution of physical properties . during the fitting
, we require that the oldest star of a given model is not older than the age of the universe at the model s redshift .
we begin by testing that spoc recovers the ( known ) properties of simulated galaxies .
this serves to both test the algorithm and quantify its intrinsic uncertainties . to do so
, we take the 73 galaxies that are resolved by our vzw simulation at @xmath65 , and determine how accurately we can recover their intrinsic physical properties using model galaxies from the @xmath66 and @xmath67 outputs as inputs to spoc . while the model and sample galaxies are not strictly independent in this test ( all but the least massive galaxies at @xmath65 correspond to at least one ancestor in the @xmath67 output and descendant in the @xmath66 output ) , the galaxies are evolving rapidly enough that these populations are effectively independent .
the test - case and model galaxies are compared in 6 bands from @xmath68-band to irac 4.5@xmath69 m ( the same ones applied to abell 2218 kesr in [ sec : kesr ] ) , where we assume a 0.15 magnitude uncertainty in each band .
the test - case galaxies are reddened with a fiducial dust extinction @xmath70 via the @xcite law .
we apply spoc to these test - cases using each of the different extinction curves mentioned in [ sec : simulations ] in order to investigate the systematic uncertainties resulting from our ignorance of the appropriate extinction curve for high - redshift galaxies . during the fitting , redshift space
is sampled by perturbing each model galaxy over a grid extending to @xmath71 so that we sample the range @xmath72 $ ] ; @xmath14 is sampled over the range @xmath73 $ ] .
spoc constrains six quantities : @xmath74 , sfr , @xmath31 , @xmath75 , age , and redshift .
the definitions of @xmath74 and redshift are self - evident .
@xmath31 is defined in terms of the @xcite reddening presciption . for the purposes of this work , metallicity @xmath75
is defined as the mean mass fraction of metals in the galaxy s stars ; this is useful in determining what metallicity to choose during population synthesis modeling .
although metallicity is not the dominant factor in determining a galaxy s sed , the fact that the vzw model reproduces the mass - metallicity relation of star - forming galaxies at @xmath76 @xcite as well as for the host galaxy of grb050904 , which is located at @xmath77 @xcite , leads us to believe that this model s predictions for the metallicities of observed reionization - epoch galaxies are plausible ( finlator et al .
2007 , in prep . ) .
we define a galaxy s age as the mass - weighted mean age of its star particles ; this is more meaningful than the more commonly - used age of the oldest star , which is both difficult to constrain observationally and difficult to predict owing to the stochastic nature of our simulations star formation prescription .
we define a galaxy s sfr as the average over the last 100 myr leading up to the epoch of observation ; if none of a galaxy s stellar mass is older than 100 myr then the age of the oldest star is used .
this metric is found to correlate more tightly with rest - frame uv flux than averages over a shorter time - baseline for the numerically simulated sfhs .
-0.0 in 0.0 in first we consider the case in which the test - case and model galaxies are both reddened via the @xcite extinction curve .
for this case , the points in figure [ fig : consistency_frac ] show how the fractional error in the six inferred properties varies with stellar mass and the solid black histogram gives their combined distribution .
the dotted lines indicate the mean formal @xmath78 uncertainties ; these are computed directly from the probability densities that are returned by spoc rather than from the scatter in the points . in general , the recovered physical parameters lie within 50% and @xmath79 of the correct values , suggesting that our sed - fitting technique is indeed self - consistent . the fact that the formal uncertainties are at least as large as the scatter ( and , in some cases , are somewhat larger ) suggests that the formal uncertainties are sufficiently conservative .
the most accurately ( and precisely ) recovered parameter is redshift .
the high accuracy in this case owes to the fact that the @xmath80 , @xmath81 , and @xmath82 fluxes tightly contrain the position of the lyman break , which itself results from the @xcite prescription for igm absorption .
stellar mass is recovered with 20% accuracy , owing primarily to the fact that the rest - frame optical flux is generally dominated by numerous long - lived , low - mass stars whose mass - to - light ratio is relatively insensitive to age and dust extinction .
additionally , as we will show in figure [ fig : consist_compare_models ] , the lack of a significant systematic offset in the recovered stellar masses owes to the similarity between the sfhs of the test - case and model galaxies .
metallicity is also accurately recovered .
this is expected given that there is a tight mass - metallicity relation in the simulations ( the @xmath78 scatter is 15% ) that does not vary strongly with redshift @xcite , and the fact that the test galaxies and models came from the same simulations . without this implicit prior
, metallicity can not be tightly constrained from broadband photometry @xcite . turning to sfr
, we expect a reasonably accurately inferred sfr given the tight correlation between sfr and stellar mass that the simulated galaxies obey @xcite ; in other words , if the redshift is known and the stellar mass can be constrained from the rest - frame optical flux , then the sfr is already constrained to within a factor of two regardless of the rest - frame uv flux .
figure [ fig : consistency_frac ] bears this out . in detail ,
sfr is somewhat less accurately recovered than stellar mass owing to the degeneracies with age and @xmath14in fact , a close inspection reveals that galaxies with underestimated sfr have overestimated @xmath14 and vice - versa .
age is accurately recovered owing largely to the small range of sfhs that occur in our simulations .
just as only a small range of metallicities remains available once the stellar mass is constrained , a relatively small range of ages is available once the redshift and stellar mass are constrained ( figure [ fig : sfh ] ) . if we relax the assumption that we know the correct form of the dust extinction curve , we find systematic effects up to the @xmath1 level . in figure
[ fig : consist_compare_models ] , the dotted red , short - dashed blue , and long - dashed magenta histograms correspond to the cases where the models were reddened with the @xcite , smc bar @xcite , and @xcite laws while the test - cases were reddened with the @xcite law as before .
metallicity and age are not strongly affected because these are tightly constrained by the combination of stellar mass and redshift .
in contrast , @xmath14 , @xmath10 , and sfr are underestimated for the other curves by up to 60% while the photometric redshifts are systematically off by up to 2% , with the the smc law yielding the largest underestimates .
these discrepancies owe to the varying slopes of the extinction curves : steeper extinction curves require less overall dust ( i.e. , lower @xmath14 ) and redder rest - frame uv colors ( i.e. , lower sfr , and thus lower stellar mass in our simulations ) in order to match a given observed rest - frame uv color . similarly , photometric redshifts are systematically off because the extra suppression of rest - frame uv flux that results from an overly steep extinction curve can be partially cancelled out by underestimating the galaxy s redshift . in summary , stellar mass , metallicity , age , and sfr
can simultaneously be recovered by spoc when numerically simulated models are used owing to the existence of implicit priors on these parameters . any remaining discrepancy between the observed and model uv fluxes
is minimized by the choice of @xmath14 , which is also relatively accurately recovered .
thus , our sed - fitting technique is indeed self - consistent . however ,
if the slope of the assumed dust extinction curve is incorrect then the resulting best estimates of the physical parameters may be off by up to @xmath83 while the photometric redshift may be off by up to @xmath84 .
these systematic uncertainties are generic to studies of high - redshift galaxies that employ sed - fitting and are unrelated to uncertainties that result from our ignorance of the correct form of high - redshift sfhs . since it is the latter aspect that we are currently trying to constrain , we do not further consider sed - fitting errors .
-0.0 in 0.0 in it is reassuring but not terribly surprising that spoc can accurately recover the physical properties of the galaxies that it uses as templates .
a more interesting question is how well spoc can recover galaxy properties using a different sfh than that of the input galaxy , as this illustrates the variations in inferred physical parameters among various assumed sfhs . to address this
, we have fit the test - case galaxies that were used in [
sec : consistency ] using model sets generated from constant , decaying , and rising sfhs as described in [ sec : sfhs ] .
figure [ fig : consist_compare_models ] gives the distributions of fractional errors in the inferred values of stellar mass , sfr , age , and redshift that result when using the different model sets .
the vzw case is simply a vertically - binned histogram from figure [ fig : consistency_frac ] .
generally , the one - parameter models yield stellar mass and age results that are within 50% of the correct values .
the errors for these quantities are generally distributed with slightly larger scatter than the errors from the vzw models and show systematic discrepancies up to the 40% level .
the sfrs are overestimated systematically by 50200% with significantly more scatter than returned by the vzw models ; this is clearly the quantity that is most dependent on the assumed sfh .
the vzw models systematically underestimate redshift by 0.2% while the one - parameter models are low by 0.5% ; the scatters are comparable for all of the models .
we briefly discuss results specific to each one - parameter model in turn .
when considering all test - case galaxies together , the constant - sfr models tend to underestimate the age and stellar mass by 40% and 10% respectively while overestimating the sfr by a median factor of 3 , the largest discrepancy among the sfhs that we consider .
when we split the sample into massive " and low - mass " galaxies at @xmath85 , we find that the constant models tend to overestimate the ages of massive " galaxies by @xmath86% while underestimating the ages of low - mass galaxies by @xmath87% . in order to match the rest - frame optical measurements
, the constant - sfr models then overestimate the sfr for the low - mass and massive galaxies by 100200% and 0100% , respectively , with 50% scatter in each case .
stellar masses are underestimated by 20% for the low - mass galaxies and overestimated by roughly the same amount for massive galaxies .
this illustrates that uncertainties in parameter recovery are not only dependent on the assumed sfh , but also on the mass .
the decaying models tend to reproduce the stellar mass and age with systematic errors of roughly 10% and scatter comparable to the scatter from the vzw models .
these successes are somewhat surprising because the simulated sfhs look nothing like the decaying case .
conversely , the sfrs are higher by a median factor of 2.8 , only slightly better than the constant model . the fact that the sfr could be dramatically overestimated while the age , stellar mass , and dust reddening ( not shown ) are recovered accurately probably owes to our use of 100-myr average sfrs . when considering all of the stellar mass that has formed in the last 100 myr , a larger fraction of the o - stars will have evolved off of the main sequence for decaying or constant sfhs than for rising models that show the same 100 myr average sfr , leading to the result that models with differing sfrs nonetheless produce similar uv luminosities .
the broad success of the decaying model despite the input sfhs looking nothing like the decaying case shows that sed fitting can yield interpretations consistent with monolithic collapse models even though the true sfh may be quite different .
the rising models tend to underestimate the age by 1040% and overestimate the sfr by 40% , though with significant scatter .
both of these offsets are compensated by overestimated @xmath14 in such a way that the stellar masses are recovered quite accurately , with @xmath88% systematic offset and scatter comparable to what is achieved via the numerically - simulated models .
overall , this model probably recovers the true parameters most faithfully among the one - parameter models , though it is not a dramatic improvement over the others . in summary
, we have shown that spoc can self - consistently recover the physical properties of the model galaxies that we use in fitting observed high - redshift galaxies .
further , simple one - parameter models are able to recover stellar mass and age to within 50% accuracy and sfr to within a factor of three , although there are systematic offsets at a comparable level .
all models yield photometric redshifts with better than 1% accuracy although none of them outperform the numerical models .
-0.0 in -0.0 in the spoc library of simulated galaxies that we employ is accumulated from simulations at different volumes and resolutions ( e.g. in the jvzw case , we use the @xmath89 runs ) . in @xcite
, we showed that , down to the adopted stellar mass resolution limit , the physical properties of galaxies are similar at overlapping mass scales between the various simulations . to reiterate this point in a way that is more relevant to the current work , figure [ fig : catalog_resolution ]
shows how the rest - frame uv - optical color and mean stellar metallicity vary with stellar mass at @xmath66 for resolved ( @xmath90 star particles ) galaxies from our 16 , 32 , and 64 @xmath91 simulation volumes . in both cases ,
the trend and the scatter are consistent between the different volumes , giving us further confidence that our simulated sfhs are in fact resolved .
a related way to look for resolution issues is to ask whether spocwill yield similar answers when it is applied to same - mass galaxies at different resolutions .
we now demonstrate that , indeed , spocdoes recover similar parameters for galaxies at different resolutions , and so combining different resolution simulations into a larger set is justified . of course , one does not need to do so in order to use spoc , it is merely a convenient avenue to increase the dynamic range spanned by our model galaxies . for convenience , we study a370 hcm6a as it can readily be fit by our models .
we apply spoc using three sets of models derived from the jvzw simulations : once using models from the @xmath92 volume ( j16 " ) , once using models from the @xmath93 volume ( j32 " ) , and once using both sets .
a lack of numerical convergence would result in systematic offsets between the probability densities from the first two fits , while the combined result shows how the models from the two volumes combine to yield our full probability density .
figure [ fig : spoc_resolution ] shows the derived probability density functions for the various physical parameters that we consider .
the ranges agree well , and the best estimates from the j16 and j32 volumes ( defined as the means of the probability density functions ) are consistent at the @xmath78 level . in detail , the j16 models return fits with somewhat lower stellar mass , sfr , and metallicity than the j32 models while the j32 models yield ages that are younger by about 40 myr .
the small offsets in mass , sfr , and metallicity do not indicate resolution problems as they are expected even in the absence of any convergence issues .
briefly , the j16 volume contributes lower mass models owing to the slope of the mass function ( at the massive end ) and our 64 star particle mass resolution cut ( at the low - mass end ) . on the other hand , the age offset results from a well - known numerical resolution limitation whereby galaxies in lower - resolution simulations take longer to condense beyond a given critical density in order to begin forming stars , yielding younger ages at a given stellar mass and redshift .
fortunately , the offset is comparable to the intrinsic uncertainty on this parameter .
we have repeated this test using object sbm03#1 with the 32 and @xmath94 volumes and found similar results .
hence we do not believe that our results using combined simulation samples are significantly hampered by numerical resolution effects .
-0.1 in 0.3 in the triple arc in abell 2218 , dubbed abell 2218 kesr by @xcite after its discoverers @xcite , is probably the best - studied @xmath34 object at present , and its physical parameters have been constrained through sed fitting by various authors . hence it provides a good test case for exploring the systematics that result from using numerically simulated model galaxies , and comparing to results employing more traditional simple sfhs . the flux from abell 2218 kesr can be measured from two lensed images in the _ hubble _ advanced camera for surveys ( acs ) @xmath81 , wide - field planetary camera 2 ( wfpc2 ) @xmath80 , and near - infrared camera and multi - object spectrograph ( nicmos ) @xmath82 and @xmath95 bands , and only one image in the _
spitzer_/irac 3.6 and 4.5 @xmath69 m bands ( the other image is blended with a nearby submillimeter source at irac s spatial resolution ) .
@xcite note that the fluxes measured by different authors in the optical / near - infrared bands disagree due to the inherent difficulties of measuring photometry from extended arcs , and that the different images of the galaxy do not agree in the _
hubble_/acs @xmath81 band .
following their suggestion , we use the weighted mean of the two images in the optical / near - infrared bands ( their sed1 ) and impose a minimum 0.15 mag uncertainty in all bands in order to account for differential lensing across the images .
band do not significantly affect the derived parameters . in the case of the @xmath96 limit
, we have verified this . ]
-0.2 in figure [ fig : spectra ] shows that the seds of the best - fitting model galaxies from our three galactic outflow recipes and two cosmologies all reproduce the observations with reduced @xmath97 .
moreover , they are remarkably similar .
all four models possess a very blue rest - frame uv continuum owing to young age and low metallicity as well as a pronounced balmer break owing to the presence of older stars .
the best - fit parameters are similar , indicating that the simulation s ability to reproduce the observations is robust to the choice of superwind feedback prescription and detailed cosmological parameters ( to the extent of the variations considered ) .
this result is akin to the findings from studies employing one - parameter model sfhs that good fits can be obtained via a variety of assumed sfhs and metallicities . to quantify this point , figure [ fig : comparewind ] and table [ table : comparewind ]
show how the derived probability densities for the physical parameters of abell 2218 kesr depend on the choice of wind model .
the entries in table [ table : comparewind ] list the mean and variance of the histograms shown in figure [ fig : comparewind ] .
each curve in figure [ fig : comparewind ] has been normalized to unit area and then scaled so that all four curves fit on the same plot .
all of the derived physical parameters except @xmath12 are remarkably robust to our choice of superwind feedback prescription .
generally , the data seem consistent with negligible dust reddening , a mean stellar age of 100200 myr , sfrs of 13 @xmath62 , a stellar mass of 38@xmath98 , and a redshift @xmath58 , in good agreement with other determinations @xcite% of the stellar mass of stars formed in galaxies that are more massive than @xmath99 at @xmath100 remains in stellar form at @xmath100 . ] . the tightness of the constraints on the various intrinsic physical parameters results from the relatively narrow range of sfhs experienced by the galaxies in the simulations .
by contrast , the uncertainty in the inferred redshift is determined by our treatment of igm extinction since the inferred redshift is dominated by the position of the lyman-@xmath7 break . comparing the results from the models in detail ,
both the cw and nw models are less efficient at suppressing star formation via outflows .
for this reason , at fixed number density ( or equivalently , dark matter halo mass ) the cw and nw model galaxies have formed more stars , retained more of their metals , have a higher sfr , and are older @xcite .
similarly , at a given stellar mass , the vzw galaxies are younger , have expelled a larger fraction of their heavy elements , and for these reasons require more dust reddening in order to match a given observed colour owing to the age metallicity dust degeneracy .
the jvzw models give results that are the similar to the vzw results owing to the similar feedback treatment .
however , the lower values used in the jvzw simulation for the cosmological parameters @xmath101 and @xmath102 delay the growth of structure at early times , yielding fewer model galaxies for us to match against observations for a given simulated volume . as a result ,
parameter space is less well - sampled and the probability density curves ( notably @xmath14 ) exhibit more stochasticity .
next to stellar mass and redshift , the mean stellar metallicity is the most tightly constrained parameter , followed by the mean age and sfr .
metallicity is the only parameter for which the different wind models disagree at the @xmath103 level .
that the models could agree on all of the derived parameters except for the metallicity reflects the fact that the effect of metallicity on the photometry is small compared to the effect of stellar age @xcite . on the other hand , the tightness of the metallicity constraint follows from the tight mass - metallicity relation that the simulation predicts ( [ sec : performance ] ) .
thus , the apparent disagreement between the metallicity constraints is simply a reflection of the tight priors imposed by the different simulations combined with the robust constraints on stellar mass .
note that because the vzw model agrees the best with the distribution of metals in the igm ( od06 ) and the mass - metallicity relationship of star - forming galaxies @xcite , it probably makes the most believable prediction of abell 2218 kesr s mean stellar metallicity .
we investigated the effect that varying the dust prescription has on the derived physical parameters and found , in agreement with @xcite , that this has no significant effect on the derived physical parameters other than @xmath14 when the simulated models were used .
the derived @xmath14 were roughly 0.1 mag lower for the @xcite and @xcite laws ; these differences are expected given the extra extinction imposed at rest - frame uv wavelengths in the former case and ( similarly ) for younger stars in the latter case .
the total amount of light removed by dust extinction , which can be regarded as a prediction of the total infrared luminosity , is roughly @xmath104 , independent of the assumed dust prescription .
two groups have previously published constraints on this object s properties . @xcite
employed a uniform sampling of single - parameter sfhs ; their results are given in table [ table : comparewind ] and by the solid green vertical lines in figure [ fig : comparewind ] , where we have converted their derived sfr and @xmath10 to values appropriate for a chabrier imf .
it is clear that their constraints are entirely consistent with our own , although the tight intrinsic correlations between physical parameters in the simulated models allow us to impose tighter constraints on all of the derived parameters . in the second work
, @xcite exhaustively examined the systematics of assumptions regarding sfh and template spectra with the goal of bracketing the most likely parameter space .
their constraints , given by the dotted magenta vertical lines and listed in [ table : comparewind ] , are also consistent with ours . hence our simulations , despite a different generic form for their galaxies sfhs than has been assumed in previous investigations , can reproduce the properties of abell 2218 kesr equivalently well .
-0.0 in efforts to constrain the physical properties of high - redshift galaxies via sed - fitting invariably encounter a host of degeneracies between the best - fit parameters , the most difficult of which is certainly the age - extinction - sfr degeneracy @xcite .
generally , young stellar population age , high sfr , and low dust extinction all yield bluer photometric colours while old age , low sfr , and high dust extinction all yield redder photometric colours .
these degeneracies also contribute to the uncertainty in the inferred stellar mass via the stellar age - stellar mass degeneracy , whereby older populations have a higher mass - to - light ratio , yielding a higher stellar mass at a given measured flux density . compounding the problem
, there are a number of ways in which the best - fit physical properties depend on any assumptions that are made regarding the shape of a galaxy s star formation history @xcite .
numerical simulations , in contrast , provide tight relationships between galaxy star formation rate , stellar mass , and metallicity ( from which extinction may be inferred ) .
these predicted relationships translate into tighter constraints on physical parameters that depend strongly on the shapes of the trial sfhs . in order to quantify how effectively the simulations reduce well - known degeneracies in best - fit parameters
, we generated 1000 monte carlo re - observations of abell 2218 kesr by adding scatter to the photometric measurements in a way that was consistent with the reported observational uncertainties .
for each data set , we then determined the best - fit parameters from the vzw models as well as from the decaying , rising , and constant sfh models .
figure [ fig : degeneracy ] shows the locus of best - fit parameters in several famously degenerate projections of parameter space .
the solid black , dotted blue , short - dashed red , and long - dashed cyan contours enclose 68% , 95% , and 99% of the best - fit solutions for the vzw , decaying , rising , and constant model sets , respectively . the expected degeneracies that result from one - parameter sfhs are easy to see in each panel of figure [ fig : degeneracy ] . at the young end ,
each of the one - parameter models includes a parameter space corresponding to a galaxy that has formed all of its stars in a rapid burst lasting less than 100 myr . our use of 100-myr average sfrs guarantees these models a high sfr ( @xmath105 ) .
because these models are very young , they are intrinsically blue and can require high dust reddening ( @xmath106 mag ) to match the observations .
additionally , their young ages guarantee that their observed optical flux is dominated by short - lived o and b stars with low mass - to - light ratios , leading to low inferred stellar masses .
these dramatic burst - dominated models have no analogue in the simulations .
the old end is dominated by the constant - sfr models , equivalent to the @xmath107 limit of the decaying models .
these models are characterized by intrinsically red colours and high mass - to - light ratios , leading to low dust extinctions and high stellar masses .
the oldest of these fits requires the galaxy to have been forming stars at @xmath108 when the universe was less than 10 myr old ; our simulations can not produce this because gas densities have not grown high enough to support such sfrs at such early times . at the @xmath78 confidence level ,
the one - parameter model that most closely approximates the simulated galaxies is the rising sfh model , albeit with a preference for somewhat older ages . returning to figure [ fig : sfh ] , we see that this is expected because the simulated galaxies sfhs are generically characterized by a slowly rising sfr at these redshifts .
conversely , at the @xmath109 level , the constantly rising sfh model allows for a wider range of models that have no analogues in the simulations .
the relatively small range of simulated galaxy sfhs typified by the examples in figure [ fig : sfh ] leads directly to the relatively tight range of inferred ages for the solid contours in figure [ fig : degeneracy ] . assuming that abell 2218 kesr is located at @xmath100 , the sfhs in figure [ fig : sfh ] suggest that it may have formed its first stars before @xmath110 , with ( 10% , 50% ) of its stars in place by @xmath111(11,8 ) ; in other words , roughly half of the best - fitting models stars at the epoch of observation are over 150 myr old .
this relatively old population readily accounts for the pronounced balmer break that is visible in figure [ fig : spectra ] , and is typical in our simulations . in summary , a key point of this paper
is demonstrated by the fact that , in each panel of figure [ fig : degeneracy ] , the confidence intervals obtained from simulated galaxies fall well within the range described by the complete set of one - parameter model sfhs , while yielding the tightest constraints .
the tighter constraints owe to the relatively small range of sfhs and the tight correlations between the various parameters that generically occur in hierarchical simulations of galaxy formation .
this illustrates that , if the simulations are broadly correct , the physical properties of high-@xmath6 galaxies can be more precisely constrained using spoc .
however , since at present the simulations are largely untested at these epochs ( modulo the broad successes in * ? ? ?
* ) , it may be more appropriate to regard the tight simulation constraints as predictions to be tested against future observations .
-0.0 in in this work we focus on reionization - epoch objects for which rest - frame optical measurements are available because the rest - frame optical flux is dominated by relatively low - mass stars whose mass - to - light ratio evolves slowly relative to the stars that dominate the rest - frame uv . for reasonable sfhs , this allows tighter constraints to be placed not only on the galaxy s stellar mass , but also on its sfh . to quantify this point , figure [ fig : ffba ] plots the light - weighted median age of a2218 kesr versus rest - frame wavelength for the sfhs in figure [ fig : sfh ] .
for example , a point at 2000 and 15 myr indicates that , for that model , 50% of the photons with rest - frame wavelength of 2000 are emitted by stars that are 15 myr old or younger . for all of the models that we consider ,
photons from bluewards of the balmer break are generated by stars that are under 100 myr old .
this is expected since b stars live roughly 100 myr @xcite and justifies the use of rest - frame uv flux as a constraint on the current sfr .
conversely , it explains why rest - frame uv data do not constrain a galaxy s sfh prior to @xmath112 myr before the epoch of observation .
by contrast , data from longer wavelengths sample the sfh at earlier epochs because the stars that dominate these wavelengths live much longer . in the case of a2218 kesr
, the constant and decaying models in figure [ fig : ffba ] suggest that rest - frame optical data constrain the galaxy s sfh roughly 80 and 200 myr before the epoch of observation , while the rising and vzw models fall in between these limits ; the differences probably owe to the detailed interplay between the slope of the sfh and the rate at which low - mass stars fade .
it is possible that the rough agreement in the optical portion of figure [ fig : ffba ] explains the tendency of all of the best - fit models in figure [ fig : sfh ] to be in rough agreement with each other during the interval @xmath113 even though they diverge prior to @xmath114 .
note that this agreement is nontrivial given that all of the one - parameter models can in principle match the observations with burst - like solutions ( figure [ fig : degeneracy ] ) .
most importantly , it highlights the potential of rest - frame optical measurements redwards of the balmer break to constrain the sfh for reionization - epoch galaxies .
-0.0 in -0.0 in while the example of abell 2218 kesr is illustrative , the ultimate purpose of spoc is to constrain physical parameters for a large sample of galaxies , in order to characterize the observed galaxy population and to constrain the underlying galaxy formation model . to illustrate the sort of insights gained using spoc
, we apply it to a sample of observed @xmath2 objects that have published broadband photometry in the rest - frame uv and optical bands . for each object , we used spoc to determine its best - fit physical parameters with the jvzw , constant , rising , and exponentially - decaying models . for objects
whose redshift has been measured spectroscopically , we run spoc twice : once with the redshift constrained to the measured value and once with the redshift left as a free parameter .
the first run allows more accurate derivations of the physical parameters , while the second enables us to study the accuracy of spoc s photometric redshifts .
table [ table : many_objs ] compares the means and 95% confidence intervals for the various physical parameters of each galaxy that result from the various models , while figure [ fig : degall ] shows the 68 and 99% contours in the @xmath10-sfr space for the same models .
while fitting with one - parameter models , we confirmed that the probability density varies only weakly with @xmath12 @xcite ; for this reason we have omitted @xmath12 from table [ table : many_objs ] .
our simulations , combined with spoc , suggest that all of the objects have @xmath115 @xcite .
figure [ fig : many_objs ] compares the synthetic spectra of the best - fit jvzw models with the published rest - frame uv through optical data for our sample .
table [ table : many_objs ] shows that we obtain very good fits ( @xmath116 ) with the simulated library in all cases except for the unusual object glare#3001 .
note that these spectra and @xmath117 values are purely representative ; in practice we obtain physical constraints through the bayesian analysis described in [ sec : spoc ] rather than by simply examining the best - fitting model .
it is encouraging that the fits obtained from our simulated galaxy library are generally equally as compelling , in terms of goodness - of - fit measure @xmath117 , as the fits that result from the one - parameter sfh model libraries .
taken together , these findings strongly support the idea that most observed reionization - epoch galaxies possess theoretical analogues within our simulations .
however , some of these objects show minor discrepancies that may be hinting at some failing in the models .
given the currently large observational uncertainties it is difficult to place robust constraints on the models , but it is nevertheless worthwhile to discuss each of these systems in detail in order to illustrate how spoc can be used to both reveal physical characteristics as well as test the underlying model . _
_ we have taken the photometric data and errors in the observed r , z , j , h , k , and _
spitzer_/irac bands from ( * ? ? ?
* hereafter cse05 ) , and we have adopted their lensing magnification factor of 4.5 .
the measured @xmath118 m flux is anomalously large ; cse05 interpret the excess over the inferred stellar continuum as h@xmath7 emission line flux and derive an sfr of @xmath119 . in our simulations galaxies at @xmath65 that possess this object
s 3.6@xmath69 m flux are forming stars at 212@xmath62 . since
such systems are not expected to produce h@xmath7 equivalent widths that would contribute significantly to the broad - band flux , we adopt the uncorrected 4.5@xmath69 m measurement .
we verified that , if we use the h@xmath7-corrected 4.5@xmath69 m flux obtained by cse05 , the inferred sfr drops from @xmath120 to @xmath121 ; this value is in conflict with the assumed line strength as expected .
the @xmath122 m limits quoted in the text of cse05 are different from the limit plotted in their figure 2 ; we adopt the less restrictive limit from the figure .
when we do not enforce the spectroscopic redshift , the photometric redshift @xmath123 is dominated by the @xmath124 limit and is roughly @xmath125 below the spectroscopic redshift ; the formal redshift uncertainty is dominated by the size of the @xmath124 baseline .
although the one - parameter models yield more accurate photometric redshifts in this case , the difference is comparable to the formal uncertainty and we regard it more as a coincidence than as a clue to the sfh of this object .
a more accurate photometric redshift would require us to include the unpublished measurement in @xmath32 and an accurate @xmath126 transmission curve ; we have used the sdss @xmath127 profile because a transmission curve for the @xmath126 measurement used by @xcite is not available . the fits that we obtain are very good , although we confirm that the 4.5@xmath69 m measurement is difficult to interpret as purely stellar continuum emission .
in fact , even without the anomalous @xmath118 m flux the uv continuum is difficult to understand owing to the dip at @xmath128 @xcite , and the measured ly@xmath7 flux is difficult to reconcile with the observed ly@xmath7 luminosity function at @xmath36 @xcite .
thus , until more precise measurements of this object are available , any inferences regarding its physical properties should probably be regarded as suggestive . with these caveats in mind , we find that using the simulated models yields inferred stellar mass , sfr , and @xmath14 that are fully consistent with previous derivations as well as with the one - parameter model sfhs . the range of stellar masses allowed by the jvzw models nicely brackets the range between the emission line - dominated and stellar continuum - dominated solutions suggested by cse05 , and the range of sfrs is consistent with the derivation of @xcite based on the ly@xmath7 line and well below the inferred h@xmath7-based sfr of cse05 .
turning to the one - parameter models , the widest uncertainties for all of the parameters are provided by the decaying model , with the results from the constant and rising models spanning a somewhat smaller subset of this space .
the results from the rising model are generally the closest match to the jvzw models although they permit a somewhat larger range of sfr ; these are burst - like models that have no analogue in our simulations . _
_ we adopted the measured broadband fluxes and spectroscopic redshifts for these objects from tables 1 and 2 in @xcite .
object sbm03#1 is the same as object # 1ab in @xcite . comparing the derived fluxes in these two papers , we find disagreement at the @xmath129 level in the acs @xmath130 , @xmath81 , and nicmos @xmath95 bands . to be conservative
, we therefore impose a minimum uncertainty of 0.15 mag in all bands for these objects . when fitting for the photometric redshifts we include the @xmath130 fluxes ; otherwise we exclude this datum following @xcite .
figure [ fig : many_objs ] shows that we obtain an excellent fit for sbm03#1 and satisfactory fits for sbm03#3 and glare#3001 , with the chief difficulty in the latter objects being the anomalous fluxes in @xmath131 as noted by @xcite . for objects
sbm03#1 @xcite and sbm03#3 @xcite spoc deduces stellar masses of @xmath132 and mean ages that include the range 100300 myr irrespective of the assumed sfh , consistent with the findings of @xcite .
the minimum age for each model is older than the ages for the other objects in our sample because the small measured uncertainties on the 3.6@xmath69 m fluxes for sbm03#1 and # 3 create the strongest case for a pronounced balmer break . in our simulations ,
such objects are roughly 130200 myr old and are forming stars at a healthy 3060 @xmath62 ( both @xmath78 ) .
the resulting intrinsically blue colours cause spoc to select moderate dust extinctions @xmath14 of 0.40.9 and 00.1 for # 1 and # 3 , respectively , in order to match their uv continuum slopes . while these sfrs are within the full range inferred via the one - parameter models , they are generally more active and younger than inferred via the decaying and constant models .
the discrepancy owes primarily to the fact that the constant and decaying models permit higher sfrs at early times ( @xmath133 ) than occur in our simulations . because the rising model excludes such early episodes , it yields constraints that are more similar to the results from the jvzw models .
the photometric redshifts for sbm03#1 and # 3 are 13@xmath134 below the spectroscopic values for both simulation and one - parameter models , suggesting that unknown systematic uncertainties such as uncertainty owing to inaccurate filter profiles should be folded into the formal uncertainties ; in future work an enforced minimum redshift uncertainty of @xmath135 seems reasonable .
object glare#3001 is a particularly interesting case because it represents the worst fit for all our models .
this object shows a relatively flat sed with little evidence for a balmer break ( figure [ fig : many_objs ] ) .
the photometric redshifts are systematically high although they are accurate at the @xmath78 level .
spoc finds that it is perhaps an order of magnitude less massive than sbm03#1 and # 3 . in our simulations , the analogues to glare#3001
are of roughly the same age as the sbm objects while their sfrs are roughly one tenth as large , leading to similarly strong balmer breaks and dust extinctions .
comparing with the one - parameter model sfhs , table [ table : many_objs ] shows that , for this object , constant and rising models yield masses , sfrs , and ages that are marginally consistent with the results from our simulations ( although they are skewed to higher sfr ) while decaying models yield extremely young , burst - like fits whose parameters are entirely inconsistent with the jvzw results ( this is especially clear in figure [ fig : degall ] ) .
the chief difficulty in fitting glare#3001 is the anomalously high flux measured in @xmath131 , as noted by @xcite .
if real , this would suggest that this is a low - mass objects undergoing a burst .
table [ table : many_objs ] shows that the @xmath117 per degree of freedom is lower for the one - parameter models than for the simulated models , because the one - parameter models have the freedom to produce a higher @xmath131 flux through a higher sfr and @xmath14 together with a lower stellar mass ; galaxies with these combinations of high sfr ( @xmath136 ) and low stellar mass ( @xmath137 ) simply do not occur in our simulations
. it would be preliminary to stake the final interpretation of this galaxy on one broadband measurement ( particularly one in @xmath131 ) , especially since even the one - parameter model fits are not particularly good ( @xmath138 ) .
but this object does illustrate how spoc can pick out galaxies that may provide the most stringent tests of galaxy formation models .
_ _ we adopted the _ hubble_/acs+nicmos and _ spitzer_/irac fluxes for objects # 5abc and # 7ab from the @xmath139 sample in @xcite and imposed a minimum uncertainty of 0.15 mag in all bands as before .
spoc failed to find an acceptable fit for object # 5abc , for either simulated or one - parameter model sfhs .
this object is evidently a blend of at least three components located at different redshifts and should therefore be fit with multiple components @xcite ; however , this is beyond the scope of the present work .
we therefore do not show the results from # 5abc .
by contrast , we obtain excellent fits for object # 7ab .
this is somewhat surprising given that # 7ab is clearly a mixture of two components ( figure 1 in @xcite ) , and probably owes largely to the fact that it does not show anomalous single - band fluxes in the way that hcm 6a , sbm03#3 , and glare#3001 do .
the photometric redshift is constrained to @xmath140 , @xmath125 higher than the photometric redshift determined by @xcite ( although they do not quote an uncertainty ) .
stellar mass is relatively poorly constrained owing to the relatively low signal - to - noise in the observed 3.6@xmath69 m band .
as figure [ fig : degall ] shows , when we apply spoc with our simulated models , the poorly - constrained stellar mass leads directly to a poorly - constrained sfr . comparing with the one - parameter models , we find that these span an even larger ( overlapping ) space .
this object illustrates the importance of high - quality near - infrared data in order to understand the physical properties of high - redshift galaxies .
@xcite applied exponentially - decaying models to this object and found a burst - like best - fit solution with a stellar mass of @xmath141 ( when converted to a chabrier imf ) , little current star formation ( @xmath142 ) , no dust extinction , and an age of 50100 myr .
the stellar mass is fully consistent with our jvzw results .
however , we find disagreement in the inferred dust extinction , sfr , and age because in the simulations galaxies of this stellar mass and redshift are invariably older and are still forming stars .
in particular , we expect @xmath143 , @xmath144 , and age = @xmath145 myr ( 1 @xmath134 uncertainty ) .
once again , the absence of burst - like models in our simulations constitutes an effective prior that excludes such solutions .
when we apply our own exponentially - decaying models to this object we find that the probability density possesses a sharp peak at low ages ( @xmath146 myr ) and a broad but slightly lower plateau at older ages .
so while we formally obtain the lowest @xmath5 for burst - like models , folding in our simulation prior leads to an older age being preferred .
this object is therefore a classic case where simply taking the lowest @xmath5 among all sfhs results in a substantially different interpretation than that obtained using physically - motivated priors
. it will be interesting to see with improved obervations whether or not the simulation prediction ends up being correct .
the sixth column of table [ table : many_objs ] combines @xmath10 and @xmath11 with the age of the universe @xmath147 at each galaxy s redshift into an estimate of the ratio of the present- to past - averaged sfr , @xmath148 .
a value of @xmath149 less than unity indicates a declining sfh whereas a value greater than unity indicates a rising sfh .
does not involve the inferred age of a galaxy s stars ,
this exercise is not circular ; e.g. , results from fitting exponentially decaying models do not necessarily require that @xmath150 . ]
@xcite have previously used this quantity to argue that galaxies that exhibit balmer breaks at @xmath33 must be experiencing a declining sfh .
in contrast , table [ table : many_objs ] indicates that , while this possibility is permitted by the decaying models at the @xmath79 level for all but one of the galaxies that we consider , it is by no means required while other models do not because decaying models can yield blue colors along with a lower specific sfr than the other models . ] . for our physically - motivated models
it is not permitted owing to the generally rising sfhs in our simulations , but in fact even for the decaying models it is not the favored solution when the full range of possibilities is taken into account .
hence we do not find that the presence of a balmer break requires a declining sfh .
the disagreement between our results and those of @xcite owes partly to the slightly different model sets used in the two analyses , and partly to our employing the full probability densities rather than restricting our attention to the model with the lowest @xmath5 .
most importantly , however , this disagreement emphasizes the fact that the form of the correct sfh is not constrained by the current measurements and hence the conclusions are highly model - dependent . in these fits
, we have not considered the possibility of any agn contribution to the seds . from rest
- uv photometry alone it is difficult to rule out this possibility .
fitting power laws of the form @xmath151 to their rest - frame uv continua , we obtain slopes in the range @xmath152 $ ] with a weighted mean of -0.61 .
these are not significantly different from the slopes inferred by @xcite from 39 quasars at @xmath153 , with a mean and standard deviation of -0.79 and 0.34 , respectively . however , with _ spitzer_/irac photometry the balmer breaks in a2218 kesr , sbm03#1 , sbm03#3 , and y05#7ab are clearly evident .
furthermore , @xcite and @xcite argue that agn can be ruled out for three of the objects ( sbm03#1 , sbm03#3 , and glare#3001 ) based on upper limits to the flux of the n@xmath154v doublet at 1240 , the lack of any x - ray detection , and the relatively small velocity width of the observed ly@xmath7 lines ( @xmath155 ) .
these arguments also apply to a370 hcm 6a .
additionally , cse05 note that the latter object is similar to high - redshift ly@xmath7 emitters , for which the agn fraction is constrained to be less than 5% ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this suggests that agn do not dominate their rest - frame optical spectra , and emphasizes the usefulness of _ spitzer_/irac measurements for constraining the properties of reionization - epoch objects .
hence it seems reasonable to model all of these objects with stellar population synthesis models alone , as we have done . in summary , we are usually but not always able to find galaxies within our simulations whose photometry suggests that they could be the theoretical counterpart to observed reionization - epoch galaxies . using these models as inputs to our sed - fitting engine , spoc , we obtain physical constraints on the properties of the galaxies that are , in five out of six cases , consistent with the results from one - parameter model sfhs . in the remaining case ,
the observed sed leads the one - parameter models to prefer burst - like scenarios that do not occur in our simulations . in all cases
the tightness of constraints from the simulated models is equal to or better than that from one - parameter model sfhs .
albeit small , this sample of systems illustrates how using spoc to compare among sfhs can provide insights into the physics of early galaxy formation , and identify unusual galaxies that provide the tightest constraints on such models .
in this paper we present a bayesian sed - fitting engine called spoc , which provides constraints on the physical properties of galaxies from photometric data .
spoc takes as input a galaxy s photometry and a set of model galaxy spectra , obtained either by assuming an analytic form for the galaxy s star formation history or from numerical simulations of galaxy formation , and outputs probability distributions for the physical properties of individual galaxies based on those model priors .
here we compare and explore implications for different models , with an eye towards better constraining the physical properties of high - redshift galaxies . because spoc is intended to test whether predicted sfhs match those inferred from data , it provides constraints on models of galaxy formation that complement comparisons to bulk properties such as galaxy luminosity functions or clustering .
a successful model must not only reproduce observed bulk statistics , but must also reproduce the colors of individual galaxies across all observed bands . galaxies with photometry that can not be well fit in a given model provide insights into model failings , and ultimately into the physics that drives galaxy formation .
using spocit is possible to quantitatively determine how well a particular model matches an individual galaxy .
such a methodology is particularly useful at epochs when only a small number of galaxies are detectable , such as for the earliest galaxies known .
we show that spoc accurately recovers the input physical parameters of model galaxies when fitted with sfh derived from the simulations themselves , with typical systematic errors less than the formal @xmath78 fitting errors .
when a simulated galaxy is fit with one - parameter star formation histories or the incorrect extinction model , then larger deviations can occur .
since the true sfh and extinction law are unknown , these deviations can be regarded as systematic errors on the determination of physical parameters . in most cases ,
such systematic errors are less than 50% in stellar mass , star formation rate , and @xmath31 , and @xmath156 in redshift .
we then apply spoc to six galaxies at @xmath0 that have published photometry spanning the 4000 break ( which is necessary in order to obtain meaningful constraints on physical properties ) .
we begin with a more in - depth study of abell 2218 kesr ( @xmath157 ) , since at present it is probably the best - studied reionization - epoch system , and then apply spoc to five more galaxies in order to investigate a wider variety of galaxy properties .
our main conclusions are summarized as follows : * the physical parameters derived for abell 2218 kesr using our library of simulated galaxies are in line with previous determinations , and those employing various one - parameter sfhs .
the formal uncertainties are smaller when using simulated galaxies , because the simulations intrinsically produce a relatively narrow range of sfhs for galaxies at these epochs .
the existence of an object with the properties of abell 2218 kesr at these epochs does not pose a significant challenge to current models of galaxy formation . *
all inferred physical parameters for abell 2218 kesr except metallicity are remarkably insensitive to the explored choices for superwind feedback model , dust extinction law , and cosmology .
the fundamental reason for this is that regardless of such choices , simulations produce a similar _ form _ for the sfh of early galaxies .
disappointingly , this means that photometry alone can not constrain such modeling uncertainties . on the other hand , this means that simulation predictions of physical parameters using spoc are quite robust .
discrimination between , say , superwind feedback models can be obtained through comparisons with bulk properties such as luminosity functions ( as shown in * ? ? ? * ) . * exploring a set of six @xmath0 galaxies , we find that for five of the systems the simulated galaxies and one - parameter models produce overlapping probability contours and best - fit physical parameters that are formally consistent , and the minimum reduced @xmath5 is similar between all models .
this shows that simulations are usually but not always able to produce galaxies with properties similar to observed @xmath158 systems . *
the spectra of simulated galaxies always show a significant balmer break , despite the fact that their sfhs are best characterized as constantly - rising ( not decaying ) .
the median stellar age is @xmath159 myr ( @xmath79 ) for all six objects , spanning a range in stellar masses from @xmath160 .
hence the existence of somewhat older stellar populations in these early systems is consistent with simulation predictions . *
the object glare#3001 is an outlier that is poorly fit by our simulations , though in fairness the fits with one - parameter models are not optimal either .
the best fits tend to favor models with younger stellar ages , lower masses , and higher sfrs than any simulated galaxies , perhaps indicating that this is a galaxy undergoing a burst .
however , as discussed in the text , these fits are mainly driven by a high @xmath161-band flux .
if that data point were confirmed it would suggest that the simulations may be missing some physical process(es ) that governs a small fraction of galaxies at this epoch .
our hydrodynamic simulations of galaxy formation make some clear predictions for the star formation histories of galaxies at these early epochs .
in particular , they predict that the stellar mass , star formation rate , and metallicity are all tightly correlated @xcite .
indeed such trends are generically seen in most hydrodynamic simulations of galaxy formation , though not necessarily in semi - analytic models .
our preliminary comparisons to @xmath158 galaxies show that , while the simulations are not necessarily statistically favored over other classes of sfhs , the available data at least do not rule out these simulation predictions . together with the fact that such simulations can reproduce key bulk properties of early galaxies @xcite , this suggests that models of early galaxy formation are able to reproduce a range of observed properties of galaxies at these epochs . in this work
we have performed the simplest possible spectral synthesis calculations , primarily because our focus has been a comparison between different sfh models . in the future
we plan to investigate different population synthesis models ( e.g. , * ? ? ?
* ) , imfs ( e.g. * ? ? ?
* ) , and nonstellar contributions to the observed fluxes such as emission lines from hii regions .
eventually we will apply spoc to a larger sample of galaxies , which will enable a number of interesting investigations .
first , the bulk statistics ( e.g. stellar mass or star formation rate functions ) derived using spoc can be compared against that directly produced in simulations . while this is partly a circular comparison
because simulated galaxies are being used to infer the physical properties , in practice there is no guarantee that consistency will be achieved because the simulated sfhs are quite generic ; hence this should provide a stringent test of models of galaxy formation .
second , spoc can be used to identify populations of galaxies that deviate dramatically from simulation predictions ( such as glare#3001 ) , in order to isolate which physical processes may be missing in models . in principle this could quantify the contributions from nonstellar emission - dominated sources such as agn and extremely dusty objects .
finally , the redshift evolution of the galaxy population provides a strong test of models , particularly investigating the dramatic change in the nature of massive galaxies that appears to occur at @xmath162 @xcite .
spoc is a general - purpose code , in principle able to utilize any type of model that produces detailed galaxy sfhs , be it a hydrodynamic , semi - analytic , or analytic model . in principle
there is nothing that restricts spoc to high redshift use , but in practice other physical phenomena such as agn contamination may become more important at lower redshifts .
spoc could also provide a useful tool to identify galaxy classes that are not readily reproduced in current models .
we are currently planning to make spoc and our latest library of simulated galaxies publicly available upon publication of this paper .
we hope that spoc will be a useful and flexible tool for conducting detailed comparisons between simulations and observations , as are critical for advancing our understanding of galaxy formation .
we thank c. papovich for a great deal of encouragement and an inexhaustible stream of useful suggestions .
we also thank d.zaritsky , e. egami , r. thompson , e. bell , and a. bunker for helpful conversations , and v. springel and l. hernquist for providing us with gadget-2 .
finally , we thank the referee , f. governato , for comments that improved the paper .
the simulations were run in part on the xeon linux supercluster at the national center for supercomputing applications , along with a 100-processor cluster here at steward observatory .
kmf acknowledges support from a national science foundation graduate research fellowship .
support for this work was also provided by nasa through grant number hst - ar-10647 from the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under nasa contract nas-26555 .
support for this work , part of the spitzer space telescope theoretical research program , was further provided by nasa through a contract issued by the jet propulsion laboratory , california institute of technology under a contract with nasa .
ccccccc nw & @xmath163 & @xmath164 & @xmath165 & @xmath166 & @xmath167 & @xmath168 + cw & @xmath169 & @xmath170 & @xmath171 & @xmath172 & @xmath173 & @xmath168 + vzw & @xmath174 & @xmath175 & @xmath176 & @xmath166 & @xmath177 & @xmath168 + jvzw & @xmath178 & @xmath179 & @xmath180 & @xmath181 & @xmath182 & @xmath168 + @xcite & & & 50450 & 0.13.3 & 8.58.8 & 6.66.8 + @xcite & @xmath183 & & 3400 & 1.51.9 & 7.79.0 & 6.07.2 + [ table : comparewind ] lccccccc a2218 kesr + jvzw & 0.23 ( 00.53 ) & 160 ( 132181 ) & 2.4 ( 1.23.7 ) & 8.7 ( 8.59.0 ) & 3.5 ( 2.54.2 ) & @xmath184 & 0.57 + constant & 0.42 ( 01.0 ) & 107 ( 7350 ) & 6.7 ( 225 ) & 8.7 ( 8.19.1 ) & 17.2 ( 1.586.1 ) & @xmath185 & 0.62 + decaying & 0.34 ( 01.00 ) & 104 ( 8370 ) & 6.9 ( 1.025.2 ) & 8.7 ( 8.19.1 ) & 19.2 ( 0.586.0 ) & @xmath186 & 0.55 + rising & 0.33 ( 00.87 ) & 129 ( 5268 ) & 3.5 ( 1.512.5 ) & 8.7 ( 8.29.1 ) & 6.7 ( 2.256.3 ) & @xmath187 & 0.60 + a370 hcm#6a + jvzw & 0.49 ( 01.08 ) & 191 ( 122250 ) & 7.8 ( 2.718.2 ) & 9.3 ( 8.99.6 ) & 2.9 ( 2.34.4 ) & @xmath188 & 1.7 + constant & 0.76 ( 0.131.24 ) & 182 ( 24402 ) & 14.0 ( 4.128.1 ) & 9.6 ( 9.09.9 ) & 3.9 ( 1.117.3 ) & @xmath189 & 1.5 + decaying & 0.55 ( 01.37 ) & 217 ( 9502 ) & 16.4 ( 0.683.5 ) & 9.5 ( 8.79.9 ) & 9.2 ( 0.486.1 ) & @xmath190 & 1.5 + rising & 0.82 ( 0.131.37 ) & 138 ( 8282 ) & 16.1 ( 4.240.1 ) & 9.4 ( 8.89.8 ) & 6.1 ( 2.243.4 ) & @xmath190 & 1.6 + sbm03#1 + jvzw & 0.67 ( 0.360.93 ) & 175 ( 123231 ) & 42.1 ( 30.962.1 ) & 10.1 ( 9.910.2 ) & 3.7 ( 2.44.7 ) & @xmath191 & 1.7 + constant & 0.45 ( 0.250.62 ) & 388 ( 235498 ) & 24.5 ( 18.029.3 ) & 10.2 ( 10.110.4 ) & 1.4 ( 1.02.1 ) & @xmath192 & 1.2 + decaying & 0.29 ( 00.75 ) & 357 ( 89601 ) & 20.7 ( 0.6399.6 ) & 10.3 ( 10.010.5 ) & 1.3 ( 0.110.1 ) & @xmath192 & 0.9 + rising & 0.78 ( 0.501.12 ) & 212 ( 51332 ) & 51.7 ( 29.5102.0 ) & 10.2 ( 10.010.4 ) & 3.4 ( 2.09.0 ) & @xmath193 & 1.6 + sbm03#3 + jvzw & 0.44 ( 0.090.69 ) & 170 ( 124231 ) & 46.1 ( 24.975.3 ) & 10.1 ( 9.910.3 ) & 3.9 ( 2.44.8 ) & @xmath194 & 2.2 + constant & 0.18 ( 00.37 ) & 385 ( 225498 ) & 25.0 ( 18.029.3 ) & 10.3 ( 10.010.4 ) & 1.4 ( 1.02.3 ) & @xmath195 & 1.7 + decaying & 0.25 ( 00.87 ) & 301 ( 30601 ) & 41.9 ( 8.3251.0 ) & 10.3 ( 9.910.6 ) & 3.0 ( 0.220.4 ) & @xmath196 & 1.5 + rising & 0.58 ( 0.251.00 ) & 188 ( 25332 ) & 64.9 ( 28.4159.2 ) & 10.2 ( 9.910.4 ) & 4.4 ( 2.113.7 ) & @xmath197 & 2.0 + glare#3001 + jvzw & 0.19 ( 00.46 ) & 183 ( 123253 ) & 5.2 ( 3.48.7 ) & 9.2 ( 9.09.3 ) & 3.7 ( 2.45.9 ) & @xmath198 & 8.7 + constant & 0.47 ( 0.120.62 ) & 31 ( 6126 ) & 18.9 ( 6.229.3 ) & 8.8 ( 8.49.2 ) & 37.3 ( 4.5101.2 ) & @xmath199 & 3.5 + decaying & 0.59 ( 0.130.87 ) & 15 ( 685 ) & 30.3 ( 7.463.3 ) & 8.7 ( 8.49.2 ) & 63.5 ( 5.6101.2 ) & @xmath200 & 3.3 + rising & 0.53 ( 00.87 ) & 42 ( 5200 ) & 17.6 ( 5.140.1 ) & 8.8 ( 8.49.3 ) & 38.1 ( 3.4101.2 ) & @xmath201 & 3.0 + y05#7ab + jvzw & 0.34 ( 00.94 ) & 175 ( 126227 ) & 9.2 ( 3.726.7 ) & 9.4 ( 9.09.8 ) & 3.4 ( 2.64.8 ) & @xmath202 & 0.56 + constant & 0.56 ( 01.00 ) & 210 ( 27448 ) & 16.2 ( 4.728.7 ) & 9.7 ( 8.810.3 ) & 4.3 ( 1.219.0 ) & @xmath203 & 0.18 + decaying & 0.48 ( 01.24 ) & 212 ( 10532 ) & 22.7 ( 0.8105.3 ) & 9.7 ( 8.510.4 ) & 12.4 ( 0.592.8 ) & @xmath204 & 0.14 + rising & 0.63 ( 01.37 ) & 140 ( 5299 ) & 20.3 ( 4.663.8 ) & 9.5 ( 8.710.2 ) & 7.5 ( 2.261.1 ) & @xmath205 & 0.35 + [ table : many_objs ] ajiki , m. , mobasher , b. , taniguchi , y. , shioya , y. , nagao , t. , murayama , t. , & sasaki , s. s. 2006 , apj , 638 , 596 balogh , m. l. , pearce , f. r. , bower , r. g. , & kay , s. t. 2001 , mnras , 326 , 1228 bentez , n. 2000 , apj , 536 , 571 berger , e. et al . ,
apj , submitted , astro - ph/0603689 bouwens , r. j. , thompson , r. i. , illingworth , g. d. , franx , m. , van dokkum , p. g. , fan , x. , dickinson , m. e. , eisenstein , d. j. , rieke , m. j. 2004 , apj , 616 , l79 bouwens , r. j. , et al .
2004 , apjl , 606 , l25 bouwens , r. j. , illingworth , g. d. , blakeslee , j. p. , & franx , m. 2006 , apj , in press , astro - ph/0509641 bruzual , g. & charlot , s. 2003 , mnras , 344 , 1000 bunker , a. j. , stanway , e. r. , ellis , r. s. , mcmahon , r. g. , & mccarthy , p. j. 2003 , mnras , 342 , l47 calzetti , d. , armus , l. , bohlin , r. c. , kinney , a. koornneef , j. , & storchi - bergmann , t. 2000 , apj , 533 , 682 cardelli , j. a. , clayton , g. c. , & mathis , j. s. 1989 , apj , 345 , 245 charlot , s. , & fall , s. m. 2000 , apj , 539 , 718 chary , r .-
r . , stern , d. , & eisenhardt , p. 2005 , apjl , 635 , l5 ( cse05 ) cuby , j .-
, le fvre , o. , mccracken , h. , cuillandre , j .- c . , magnier , e. , & meneux , b. 2003 , a&a , 405 , l19 dav , r. , finlator , k. , & oppenheimer , b. 2006 , mnras , accepted , astro - ph/0511532 dav , r. , finlator , k. , & oppenheimer , b. 2006 , to appear in proc .
`` chemodynamics 2006 : from first stars to local galaxies '' , astro - ph/0608537 dawson , s. , et al .
2004 , apj , 617 , 707 dickinson , m. , et al .
2004 , apjl , 600 , l99 drory , n. , salvato , m. , gabasch , a. , bender , r. , hopp , u. , feulner , g. , & pannella , m. 2005 , apjl , 619 , l131 dunlop , j. s. , cirasuolo , m. , & mclure , r. j. , mnras , submitted , astro - ph/0606192 erb , d. et al .
2006 , apj , accepted , astro - ph/0604041 egami , e. et al .
2005 , apj , 618 , l5 eyles , l. p. , bunker , a. j. , stanway , e. r. , lacy , m. , ellis , r. s. , & doherty , m. 2005 , mnras , 364 , 443 eyles , l. p. et al .
2006 , mnras , submitted , astro - ph/0607306 fan , x. , et al .
2001 , aj , 121 , 31 fardal , m. , katz , n. , weinberg , d. , & dav , r. 2006 , mnras , submitted , astro - ph/0604534 finlator , k. , dav , r. , papovich , c. , & hernquist , l. 2006 , apj , 639 , 672 gawiser , e. , et al .
2006 , apjl , 642 , l13 gonzlez - serrano , j. i. , cepa , j. , prez - garca , a. m. , gallego , j. , alfaro , e. j. , snchez - portal , m. , & gonzlez , j. j. 2005 , revista mexicana de astronomia y astrofisica conference series , 24 , 247 gordon , k. d. , clayton , g. c. , misselt , k. a. , landolt , a. u. , & wolff , m. j. 2003 , apj , 594 , 279 gwyn , s. d. j. , & hartwick , f. d. a. 2005 , aj , 130 , 1337 haardt , f. & madau , p. 2001 , in proc .
xxxvith rencontres de moriond , eds .
neumann & j.t.t .
hu , e. m. , cowie , l. l. , mcmahon , r. g. , capak , p. , iwamuro , f. , kneib , j .- p . , maihara , t. , & motohara , k. 2002 , apjl , 576 , l99 hu , e. m. , cowie , l. l. , capak , p. , mcmahon , r. g. , hayashino , t. , & komiyama , y. 2004 , aj , 127 , 563 iben , i. j. 1967 , ara&a , 5 , 571 idzi , r. , somerville , r. , papovich , c. , ferguson , h. c. , giavalisco , m. , kretchmer , c. , & lotz , j. 2004 , apjl , 600 , l115 kashikawa n. , et al . 2006 , astro - ph/0604149 kauffmann , g. , et al .
2003 , mnras , 341 , 33 kawai , n. , et al .
2006 , nature , 440 , 184 kennicutt , r. c. 1998 , apj , 498 , 541 keres , d. , katz , n. , weinberg , d. h. , & dav , r. 2005 , mnras , in press kneib , j .-
ellis , r. s. , santos , m. r. , & richard , j. 2004 , apjl , 568 , l75 kodaira , k. , et al .
2003 , pasj , 55 , l17 kolatt , t. s. et al .
1999 , apjl , 523 , l109 kurk , j. d. , cimatti , a. , di serego alighieri , s. , vernet , j. , daddi , e. , ferrara , a. , & ciardi , b. 2004 , a&a , 422 , l13 labb , i. , et al .
2006 , astro - ph/0608444 madau , p. apj , 441 , 18 malhotra , s. & rhoads , j. 2004 , apj , 617 , l5 malhotra , s. , et al .
2005 , apj , 626 , 666 maraston , c. , et al .
2006 , apj , submitted , astro - ph/0604530 martin , c. l. 2005 , apj , 621 , 227 mckee , c. f. & ostriker , j. p. 1977 , apj , 218 , 148 murray , n. , quatert , e. , & thompson , t. a. 2005 , apj , 618 , 569 mobasher , b. et al .
2005 , apj , in press nagao , t. , et al .
2004 , apjl , 613 , l9 nagao , t. , et al .
2005 , apj , 634 , 142 night , c. , nagamine , k. , springel , v. , & hernquist , l. 2005 , mnras , accepted , astro - ph/0503631 oppenheimer , b. d. & dav , r. 2006 , mnras , in press , astro - ph/0605651 ( od06 ) ouchi , m. , et al .
2005 , apjl , 620 , l1 papovich , c. , dickinson , m. , & ferguson , h. c. 2001 , apj , 559 , 620 papovich , m. et al .
2005 , apj , submitted rhoads , j. e. , & malhotra , s. 2001 , apjl , 563 , l5 rhoads , j. e. , et al .
2003 , aj , 125 , 1006 rhoads , j. e. , et al .
2004 , apj , 611 , 59 rupke , d. s. , veillaux , s. , & sanders , d. b. 2005 , apjs , accepted , astro - ph/0506611 santos , m. r. , ellis , r. s. , kneib , j .-
p . , richard , j. , kuijken , k. 2004 , apj , 606 , 683 schaerer , d. , & pell , r. 2005 , mnras , 362 , 1054 shapley , a. e. , steidel , c. c. , adelberger , k. l. , dickinson , m. , giavalisco , m. , & pettini , m. 2001 , apj , 562 , 95 shapley , a. e. , steidel , c. c. , erb , d. k. , reddy , n. a. , adelberger , k. l. , pettini , m. , barmby , p. , & huang , j. 2005 , apj , 626 , 698 shimasaku , k. , et al .
astro - ph/0602614 smail , i. , chapman , s. c. , blain , a. w. , & ivison , r. j. 2004 , apj , 616 , 71 somerville , r. s. , primack , j. r. , & faber , s. m. 2001 , mras , 320 , 504 d. n. spergel _ et al .
_ , apjs , 148 , 175 d. n. spergel _ et al .
_ , astro - ph/0603449 springel , v. & hernquist , l. 2002 , mnras , 333 , 649 springel , v. & hernquist , l. 2003a , mnras , 339 , 289 springel , v. & hernquist , l. 2003b , mnras , 339 , 312 springel , v. , frenk , c. s. , & white , s. d. m. 2006 , nature , 440 , 1137 stanway , e. r. , bunker , a. j. , & mcmahon , r. g. 2003 , mnras , 342 , 439 stanway , e. r. , et al .
2004 , apjl , 604 , l13 stanway , e. r. , bunker , a. j. , mcmahon , r. g. , ellis , r. s. , treu , t. , & mccarthy , p. j. 2004 , apj , 607 , 704 stern , d. , yost , s. a. , eckart , m. e. , harrison , f. a. , helfand , d. j. , djorgovski , s. g. , malhotra , s. , & rhoads , j. e. 2005 , apj , 619 , 12 stiavelli , m. , et al .
2005 , apjl , 622 , l1 wang , j. x. , et al .
2004 , apjl , 608 , l21 wang , j. x. , malhotra , s. , & rhoads , j. e. 2005 , apjl , 622 , l77 westra , e. , et al .
2005 , a&a , 430 , l21 yan , h. , et al .
2005 , apj , 634 , 109 | we conduct a detailed comparison of broad - band spectral energy distributions of six @xmath0 galaxies against galaxies drawn from cosmological hydrodynamic simulations .
we employ a new tool called spoc , which constrains the physical properties of observed galaxies through a bayesian likelihood comparison with model galaxies .
we first show that spoc self - consistently recovers the physical properties of a test sample of high - redshift galaxies drawn from our simulations , although dust extinction can yield systematic uncertainties at the @xmath1 level .
we then use spoc to test whether our simulations can reproduce the observed photometry of six @xmath2 galaxies drawn from the literature .
we compare physical properties derived from simulated star formation histories ( sfhs ) versus assuming simple models such as constant , exponentially - decaying , and constantly rising . for five objects ,
our simulated galaxies match the observations at least as well as simple sfh models , with similar favored values obtained for the intrinsic physical parameters such as stellar mass and star formation rate , but with substantially smaller uncertainties .
our results are broadly insensitive to simulation choices for galactic outflows and dust reddening .
hence the existence of early galaxies as observed is broadly consistent with current hierarchical structure formation models .
however , one of the six objects has photometry that is best fit by a bursty sfh unlike anything produced in our simulations , driven primarily by a high @xmath3-band flux .
these findings illustrate how spoc provides a robust tool for optimally utilizing hydrodynamic simulations ( or any model that predicts galaxy sfhs ) to constrain the physical properties of individual galaxies having only photometric data , as well as identify objects that challenge current models .
galaxies : formation , galaxies : evolution , galaxies : high - redshift , cosmology : theory , methods : numerical |
we consider the problem of determining the coefficients and exponents of an unknown sparse multivariate polynomial , given a `` black box '' procedure that allows for its evaluation at any chosen point .
our new technique is a variant on the classical kronecker substitution .
say @xmath0 is an @xmath1-variate polynomial with max degree less than @xmath2 and coefficients in a ring @xmath3 .
the kronecker substitution produces a univariate polynomial @xmath4 $ ] by substituting powers of @xmath5 in the evaluation of @xmath0 : @xmath6 this map is invertible because there is a one - to - one correspondence between terms in @xmath0 and in @xmath7 , but the price of such convenience is an exponential increase in the degree . for example , consider the following bivariate polynomial : @xmath8 the standard kronecker substitution would be @xmath9 every term in @xmath7 comes from a single term in @xmath0 , and the original exponents can be determined by the base-@xmath2 expansion of exponents in @xmath7 .
our randomized kronecker substitution is also a map from multivariate to univariate polynomials obtained by evaluating at powers of a single indeterminate .
we choose @xmath1 integers @xmath10 at random and perform the substitution @xmath11 .
when these integers are not too large ( in particular , if each @xmath12 ) , the degree of @xmath7 will be less than in the usual kronecker substitution .
the price of a decreased degree is that the map is no longer invertible , for two reasons .
first , we may have two or more distinct terms in @xmath0 converge to a single term in @xmath7 .
this is called a _
collision_. the second difficulty is that the original multivariate exponents can not be determined directly from the terms in a single substitution @xmath7 .
we will show how performing @xmath13 such random substitutions can overcome both difficulties . in the example
above , choose @xmath14 and @xmath15 , so that @xmath16 in this case @xmath17 , so there were no term collisions , even though their order has changed .
the advantage is that the degree is significantly less than that from the usual kronecker substitution .
choosing instead @xmath18 and @xmath19 , the result is @xmath20 which again has a reduced degree , but in this case we have a _ collision _ : the two terms @xmath21 in @xmath0 collided to produce a single term @xmath22 under this substitution .
nonetheless , for the two terms not involved in a collision , both images can be used to recover the original terms in @xmath0 .
the two terms with coefficient @xmath23 in the substitutions are @xmath24 and @xmath25 .
this produces the linear system @xmath26 which is solved to reveal exponents @xmath27 and @xmath28 of the original term @xmath29 .
crucial to our success is determining a suitable set of integers from which to choose the @xmath30 .
section [ sec : subs ] provides bounds to construct such sets with provably many `` good '' choices which will not produce many collisions . recovering the original terms
requires a way to correlate like terms in the images @xmath7 , and section [ sec : diverse ] shows how a separate randomization allows the coefficients to be used to identify like terms in separate images @xmath7 .
section [ sec : interp ] describes a multivariate interpolation algorithm that makes use of these randomizations .
polynomial interpolation dates from the 18th century , where the goal was to discover a polynomial approximation to an unknown function by collecting sufficiently many observations .
it is often beneficial to search for a _ sparse _ model , where only a bounded number of terms in the polynomial are nonzero ( see , e.g. , @xcite ) .
our setting is more restrictive in two senses .
first , we require the ability to choose the points at which the unknown polynomial is evaluated . in this _ black - box _ interpolation setting , the polynomial a procedure which can be probed with any desired input , at some cost ( see , e.g. , @xcite ) .
we also require that a sparse polynomial truly exists `` inside the box '' , and all evaluations even those with numerical noise come from the same @xmath31-sparse polynomial .
this is different from the general setting above where the true function need not be @xmath31-sparse or indeed a polynomial at all .
( allowing for outliers as in @xcite provides another option . ) algorithmic progress in sparse interpolation can be divided into two categories depending on the complexity .
algorithms with polynomial dependence on the partial degree bounds , sparsity bound , and number of variables are called _ sparse _ algorithms @xcite .
those with only logarithmic dependence on the degree are _ supersparse _ and can be useful even in the case of univariate polynomials @xcite . as shown in the introduction , kronecker subtitution @xcite can turn a univariate interpolation into a multivariate one , by evaluating the multivariate polynomial at high powers of the univariate interpolation points .
this is especially effective in conjunction with supersparse algorithms @xcite .
zippel s sparse interpolation method proceeds one variable at a time and uses randomization to identify which portions of the unknown sparse polynomial vanish . this is another way to turn any univariate algorithm into a multivariate one @xcite .
there is recent work on implementations of sparse interpolation @xcite and applications to gcd and factorization @xcite .
the kronecker substitution in particular has been applied to integer and polynomial multiplication @xcite .
sparse interpolation has a strong connection to the important theoretical problem of polynomial identity testing ( see @xcite for a survey ) .
of particular interest is @xcite , where a similar technique to ours is used for identity testing and interpolation .
a significant advantage of their algorithms is the very small number of random bits required for the computation .
our randomized kronecker substitution provides yet another way to transform any univariate interpolation algorithm into a multivariate one .
table [ table : compare ] summarizes our contribution as compared to @xcite , zippel s method , and the normal kronecker substitution .
the table shows , for each method , how many times a univariate interpolation algorithm must be called , and the degree of the univariate polynomials that must be interpolated , relative to a given bound @xmath2 on the max degree of the unknown multivariate polynomial , and the number of variables @xmath1 . ( the sparsity bound for the univariate algorithms will always be simply @xmath31 . )
.uni - to - multivariate interpolation methods[table : compare ] [ cols="^,^,^",options="header " , ]
for an unknown polynomial @xmath32 $ ] , our main technical contribution is a way of choosing integers @xmath33 such that the substitution @xmath34 results in a lower degree than the usual kronecker substitution , while probably not introducing too many term collisions .
we begin with the case of @xmath35 variables , where our result is stronger than the general case and we always choose the random substitution exponents @xmath36 to be prime numbers .
bivariate polynomials naturally constitute a large portion of multivariate polynomials of interest , and they correspond to the important case of converting between polynomials in @xmath37 $ ] and multiple - precision integers @xcite . throughout this subsection , we assume @xmath38 $ ] is an unknown bivariate polynomial , written as @xmath39 we further assume upper bounds @xmath40 on the @xmath41 and @xmath42 , respectively , and @xmath43 on the number of nonzero terms @xmath44 .
the general idea here is to perform the substitution @xmath45 for random chosen prime numbers @xmath46 and @xmath47 .
we want to choose @xmath46 and @xmath47 as small as possible , so as to minimize @xmath48 , but large enough so that there are not too many collisions .
our approach to choosing primes is based on the following technical lemma , which shows how to guarantee a high probability success while minimizing the degree of @xmath7 .
[ lem : lambda - pq ] let @xmath38 $ ] with partial degrees less than @xmath40 and at most than @xmath31 nonzero terms , @xmath49 be a chosen bound on the probability of failure , and @xmath50 be the index of a nonzero term in @xmath0 .
define @xmath51 @xmath52 by choosing primes @xmath53 uniformly at random from the ranges @xmath54 $ ] and @xmath55 $ ] respectively , the probability that the @xmath56th term of @xmath0 collides in @xmath57 is less than @xmath58 . the primes @xmath46 and @xmath47 constitute the random choices in this discussion .
we say a pair @xmath59 is _ bad _ if the term @xmath60 collides with any other term in @xmath57 .
we will show that the number of bad pairs @xmath59 is at most @xmath58 times the total number of prime pairs that could be chosen .
we begin with a simple lower bound on the latter quantity . applying equation ( 3.8 ) in @xcite
twice shows that the total number of ordered pairs @xmath59 is at least @xmath61 next we obtain an upper bound on the number of bad pairs by counting the total number of times the @xmath56th term collides in every possible @xmath62 .
observe that for any set @xmath63 of nonzero integers , and any bound @xmath64 , the number of times any prime @xmath65 divides any of the integers in @xmath63 is at most @xmath66 a collision with the @xmath56th term occurs whenever @xmath67 for some @xmath68 , @xmath69 .
this happens only when @xmath70 and @xmath71 . since @xmath72 , these exponent differences @xmath73 and @xmath74 can not both be zero . furthermore ,
if one exponent difference is zero , then the collision can never occur .
therefore all collisions occur at indices in the set @xmath75 the total number of times the @xmath56th term collides in any @xmath62 is equal to the sum over all @xmath76 of the number of pairs @xmath59 such that @xmath70 and @xmath71 . now define , for each @xmath46 , the subset of _ possible _ collision indices as @xmath77 as each @xmath78 , we have from that @xmath79 for each prime @xmath46 , the total number of times the @xmath56th term collides is the number of indices @xmath80 such that @xmath81 . as each of these differences is less than @xmath82 , using again this sum is at most @xmath83
. therefore the total number of times the @xmath56th term collides is at most @xmath84 using the definition of @xmath85 and the observation that @xmath86 , we can rewrite this bound as @xmath87 which is exactly @xmath58 times .
hence the probability of choosing a bad pair @xmath59 is at most @xmath58 .
choosing primes @xmath53 from such sets provides a good bound on the degree of the resulting polynomial @xmath7 .
[ cor : deg - bivar ] let @xmath38 $ ] with partial degrees less than @xmath40 and at most @xmath31 nonzero terms .
then for any constant error probability @xmath88 , and primes @xmath53 chosen randomly as in lemma [ lem : lambda - pq ] , the substitution polynomial @xmath89 has degree at most @xmath90 the degree of a standard kronecker substitution is @xmath91 . because @xmath31 is always less than this
, the randomized substitution will never have degree more than a logarithmic factor larger than @xmath91 .
the benefit comes when the polynomial is sparse , i.e. , @xmath92 , in which case the randomized substitution has much smaller degree , albeit at the expense of a few collisions
. when @xmath0 has at least 3 variables , the preceding analysis no longer applies .
the new difficulty is that potentially - colliding terms could have exponents that differ in two _ or more _
variables , meaning that the simple divisibility conditions are no longer sufficient to identify every possible collision .
consequently , our randomly - chosen exponents in this case will be somewhat larger , and not necessarily prime .
for this subsection , @xmath32 $ ] is an unknown @xmath1-variate polynomial , written as @xmath93 where @xmath94 and each @xmath95 . @xmath2 and @xmath31 are upper bounds on the max degree and number of terms in @xmath0 . our general approach here is to choose a random vector @xmath96 of integers below a certain bound , and then perform the substitution @xmath97 . the following lemma , similar in purpose to lemma [ lem : lambda - pq ] , shows how large the integers in @xmath98 must be in order to guarantee a small likelihood of collisions .
[ lem : lambda ] let @xmath99 $ ] with max degree less than @xmath2 and at most @xmath31 nonzero terms , @xmath49 be a chosen bound on the failure probability , and @xmath50 be the index of a nonzero term in @xmath0 . define @xmath64 to be the least prime number satisfying @xmath100 .
if integers @xmath101 are chosen uniformly at random from @xmath102 $ ] , then the probability that the @xmath56th term of @xmath0 collides in @xmath103 is less than @xmath58 . adopt the notation @xmath104 for the finite field with @xmath64 elements , which we will represent as @xmath105 .
write @xmath106 for the randomly - chosen vector in @xmath107 .
now let @xmath69 , @xmath68 , and consider the @xmath108th term of @xmath0 .
writing @xmath109 for the exponent vectors of these terms as in , define @xmath110 , which can not be the zero vector as @xmath56 and @xmath108 are distinct terms .
we see that these two terms collide in the substitution @xmath103 if and only if @xmath111 .
define @xmath112 such that @xmath113 is the largest power of @xmath64 that divides every entry in @xmath114 , and write @xmath115 .
this means that @xmath116 and @xmath117 .
furthermore , @xmath118 if and only if @xmath119 .
now , if @xmath119 , then this also holds modulo @xmath64 , so @xmath98 must lie in the @xmath120-dimension null space of @xmath121 , call it @xmath122 , where @xmath123 .
the probability that @xmath124 is @xmath125 , and therefore the probability that terms @xmath56 and @xmath108 collide is at most @xmath125 as well . since there are @xmath126 terms that the @xmath56th term could collide with , i.e. , @xmath126 choices for @xmath108 , the probability that the @xmath56th term collides with any other term is at most @xmath127 .
from the definition of @xmath64 , this is less than @xmath58 . due to bertrand s postulate
, we will have @xmath128 .
the following corollary shows how this bound on the size of entries in the randomly - chosen @xmath98 affects the degree of the reduced univariate polynomial .
compared to @xmath129 , the degree of the univariate polynomial from the usual kronecker substitution , we see a significant reduction when @xmath130 .
let @xmath99 $ ] with max degree less than @xmath2 and at most @xmath31 nonzero terms . for any constant error probability @xmath88 , and integers
@xmath131 chosen randomly as in lemma [ lem : lambda ] , the polynomial @xmath132 has degree at most @xmath133 .
consider @xmath99 $ ] as in : @xmath134 each choice of @xmath106 for a randomized kronecker substitution maps @xmath0 to a univariate polynomial @xmath135 . in order to recover the exponent tuples of the original polynomial @xmath0
, it will be necessary in the next section to perform multiple such substitutions and correlate terms in each @xmath7 that correspond to the same unknown term in @xmath0 .
the notion of _ diversification _ , introduced by @xcite , will be used to correlate terms in the substituted polynomials @xmath7 .
the basic idea is that distinct terms in @xmath0 will be made , through a randomization , to have distinct coefficients .
in fact , as there will be some small number of collisions in each substituted polynomial @xmath7 , we require the notion of _ generalized diversity _ from @xcite .
the idea is that not only must the terms in @xmath0 have distinct coefficients , but some small number of _ sums _ of terms in @xmath0 must additionally be distinct .
this problem of diversification is to choose @xmath136 from a suitable set so that , with high probability , @xmath136 is not a root of any in a set @xmath137 of polynomials . in the original notion of diversity ,
@xmath137 simply consists of the set of @xmath138 pairwise term differences from @xmath0 . to achieve generalized diversity
, we must also consider polynomials @xmath139 of the form @xmath140 where @xmath63 and @xmath141 each either comprise a single term or a set of terms appearing in a collision .
[ def : divset ] let @xmath142 and bounds @xmath2 , @xmath143 , and @xmath58 be given , and let @xmath144 $ ] be an unknown subset of nonzero polynomials satisfying @xmath145 and with each @xmath146 having max degree less than @xmath2 .
we say @xmath147 is a @xmath148-_diversifying set _ if the probability is less than @xmath58 that any evaluation point @xmath149 , with entries chosen at random from @xmath150 , is a root of _ any _ of the @xmath146 .
that is , @xmath151 \ge 1-\mu.\ ] ] from the discussion above , we see that the set @xmath137 the differences between any of the @xmath152 single terms and any of the @xmath153 sets of collisions , which is at most @xmath154 .
let @xmath99 $ ] with degree less than @xmath2 and at most @xmath31 nonzero terms , and @xmath88
. if there are at most @xmath153 collisions in some set of evaluations of @xmath0 , and @xmath155 is chosen at random from a set @xmath147 that is a set , then , with probability at least @xmath156 , every coefficient of @xmath157 is distinct from every other coefficient and from the coefficients of all @xmath153 collisions
. follows from the discussion above and the definition of a diversifying set . from the definition of diversifying set ,
a simple application of the schwartz - zippel lemma could be used to generate diversifying sets as long as the field @xmath3 is sufficiently large .
theorems 3.1 and 4.6 in @xcite define diversifying sets for large finite fields and fixed - precision complex numbers , respectively , in the univariate case @xmath158 .
our more recent work in @xcite constructs smaller diversifying sets by choosing _ vectors _ of substitutions , again for the univariate case .
we restate these results in our current notation and refer the reader to the aforementioned results for further details .
let bounds @xmath2 , @xmath143 , and @xmath58 be given . if @xmath47 is a prime power satisfying @xmath159 , then the set @xmath160 of all nonzero elements in the finite field of size @xmath47 is a @xmath161-diversifying set .
@xcite also considers the case when @xmath162 $ ] , @xmath0 is given by a numerical black box .
their proof does not apply here as the polynomials in @xmath137 for us are not always binomials .
we hope that a similar result would hold for multivariate diversification , but do not consider the question here .
let bounds @xmath2 , @xmath143 , and @xmath58 be given as above , @xmath47 a prime power , and set @xmath163 then the set @xmath164 of @xmath165-tuples from a size-@xmath166 extension of the finite field with @xmath47 elements is a @xmath161-diversifying set is actually a @xmath165-tuple of evaluations , and the coefficients in @xmath167 are actually @xmath165-tuples in @xmath168 . ] .
we show , more generally , that vectorization may be applied to any diversifying set .
let @xmath169 be given and @xmath137 an unknown set of polynomials as in definition [ def : divset ] .
if @xmath170 is a @xmath148-diversifying set , then @xmath171 is a @xmath172-diversifying set , where addition and multiplication in @xmath173 are component - wise . as @xmath150 is a @xmath161-diversifying set , then by definition , a randomly selected row vector @xmath174 satisfies @xmath175 for all @xmath176 with probability at least @xmath156 .
suppose @xmath177 is chosen randomly from @xmath178 , and note that @xmath179 .
thus the probability that @xmath180 is the probability that @xmath181 for every @xmath182 , which is at most @xmath183 . rather than rehash the univariate diversification procedures , we refer the reader to the aforementioned results and provide the following connection which shows that univariate diversifying sets , with success probability scaled by a factor of @xmath1 , become multivariate diversifying sets .
[ thm : mult - diverse ] let @xmath3 be an integral domain and @xmath169 be given and @xmath137 an unknown set of polynomials as in definition [ def : divset ] . if @xmath170 is a @xmath184-diversifying set , then @xmath150 is also a @xmath148-diversifying set .
the proof is by induction on @xmath1 .
when @xmath158 , the statement holds trivially .
so assume @xmath185 and also that any @xmath186-diversifying set is also a @xmath187-diversifying set .
we know that @xmath137 is a set of @xmath1-variate polynomials , each with max degree less than @xmath2 .
rewrite each @xmath139 as a polynomial in @xmath188 with coefficients in @xmath189 $ ] , and define @xmath190 to be the leading coefficient of @xmath139 in terms of @xmath188 .
each @xmath190 is an @xmath120-variate polynomial with max degree less than @xmath2 .
furthermore , every @xmath190 must be nonzero since @xmath191 .
now define @xmath192 .
this is a set of @xmath120-variate polynomials with degrees less than @xmath2 .
therefore , by the induction hypothesis , a random evaluation point @xmath193 with elements chosen from @xmath150 is a zero for any polynomial in @xmath194 with probability less than @xmath195 .
now consider the set @xmath196 , whose leading coefficients are all nonzero with probability at least @xmath197 .
@xmath194 is a set of at most @xmath143 univariate polynomials with degrees less than @xmath2 . from the original definition of @xmath150 , choosing @xmath198 at random from @xmath150
makes every @xmath199 nonzero with probability at least @xmath200 .
therefore a randomly - chosen point @xmath201 is a root of any @xmath146 only if @xmath202 is a root of some polynomial in @xmath194 , or @xmath198 is a root of some polynomial in @xmath203 . as the probability of each of these is less than @xmath195 and @xmath204 , respectively , the probability
either occurs must be less than @xmath58 , as required .
in this section we show how one can interpolate @xmath0 using randomized substitutions and a univariate interpolation algorithm . as in ,
write @xmath205.\ ] ] with known bound @xmath206 . if @xmath207 , then we simply perform @xmath1 substitutions @xmath208 , , @xmath209 , each of which reveals the single term and its exponent in one of the variables .
no randomization is necessary in this case and the solution is trivial .
therefore for the remainder of this section we assume that @xmath210 .
the first step in interpolating @xmath0 is to select @xmath211 randomized kronecker substitutions , @xmath212 , where each @xmath213 .
we require the @xmath214 to be chosen in such a way that , with high probability , every term of @xmath0 avoids collision for at least half of the substitutions @xmath214 . to achieve this
we first randomly select kronecker substitutions such that any fixed term of @xmath0 avoids collision for the substitution @xmath98 with probability exceeding @xmath215 . for the bivariate case
, we would choose primes @xmath216 $ ] , and @xmath217 $ ] , for @xmath218 , where @xmath219 and @xmath220 are determined by setting @xmath221 in lemma [ lem : lambda ] .
applying lemma [ lem : lambda ] for the general multivariate case , we would select each integer @xmath222 $ ] , where @xmath64 is the least prime greater than @xmath223 .
given such choices of @xmath214 , the following lemma shows how many substitutions @xmath211 are required so that every term of @xmath0 appears without collisions in at least half of them .
[ lem : nu ] let @xmath99 $ ] with max degree less than @xmath2 and at most @xmath31 nonzero terms
. set @xmath224 and choose @xmath211 vectors @xmath225 such that , for any single @xmath98 and any particular term in @xmath0 , the probability that the term collides with another is less than @xmath226 .
then , with probability at least @xmath227 , every term of @xmath0 collides with no others for at least @xmath228 of the substitutions . by hoeffding s inequality @xcite the probability that any fixed term of @xmath0 collides in a proportion of at least @xmath229 of the substitutions is at most @xmath230 .
thus the probability is at most @xmath231 that any term of @xmath0 collides in more than @xmath232 of the substitutions @xmath98 .
the next step is to find an appropriate diversifying set for the interpolation problem .
note the images @xmath233 contain at most @xmath234 nonzero terms in total .
every term @xmath235 from one of the images @xmath236 , @xmath237 , is the image of the ( possibly empty ) sum of terms of @xmath0 of degree @xmath238 satisfying @xmath239 .
we take @xmath240 to be the set of nonzero differences of all such sums .
thus , in order to obtain the appropriate diversity with some desired probability @xmath241 , we require a @xmath242-diversifying set . per theorem
[ thm : mult - diverse ] , it suffices that we find a @xmath243-diversifying set @xmath150 .
we randomly select @xmath244 from @xmath245 and then use a univariate interpolation algorithm of our choosing in order to construct the set of images @xmath246 having the property that , with high probability , every pair of terms @xmath235 of @xmath247 and @xmath248 of @xmath249 ( @xmath250 ) , sharing a coefficient @xmath251 , are images of the same sum of terms of @xmath0 .
[ lem : m ] let @xmath99 $ ] with max degree less than @xmath2 and at most @xmath31 nonzero terms .
assume @xmath211 substitution vectors are chosen according to lemma [ lem : nu ] .
set @xmath252 and choose a @xmath253-diversifying set @xmath254 . then , with probability at least @xmath255 , any nonzero coefficient @xmath251 that appears in at least @xmath256 of the substitution polynomials @xmath247 is the image of a single term in @xmath0 . from the proof of lemma [ lem : nu ] ,
the probability is at least @xmath227 that every term in @xmath0 appears without collision in at least @xmath256 of the substitutions . assuming this is the case
, there can be at most @xmath257 sums of terms that collide in any image @xmath247 , since each collision involves at least two terms , there are at least @xmath258 terms that do not collide , and the total number of terms in all images is @xmath234 .
hence the set of term differences @xmath137 will consist of the differences of any pair in a set of @xmath259 polynomials .
the number of such pairs is less than @xmath143 given in the statement of the lemma . from the definition of @xmath150 , the probability that any of these polynomials in @xmath137 vanish on @xmath174 is less than @xmath231 , so the total probability that each term in @xmath0 is uninvolved in collisions in at least @xmath256 of the images , and all distinct terms and collisions in the image polynomials @xmath247 have distinct coefficients , is at least @xmath260 .
a direct consequence is that any fixed subset of terms of @xmath0 must collide in fewer than half of the @xmath247 . for every nonzero coefficient that occurs in at least @xmath256 of the images
@xmath247 , we know those terms with coefficient @xmath251 are probably images of the same fixed term of @xmath0
. an alternate method might be to allow the @xmath261 sums of terms that appear in collisions to sometimes share the same coefficient , as long as these coefficients are not the same as any of the @xmath31 coefficients of actual terms in @xmath262 .
this would reduce the @xmath143 in determining the diversifying set to @xmath263 , a factor of @xmath1 improvement from the bound above .
the cost of such weakened diversifying sets would be that some number @xmath264 of terms in the final recovered polynomial @xmath139 are not actually terms in @xmath0 . by iterating @xmath265 times , such `` garbage terms '' could be eradicated . for each coefficient
@xmath251 that appears in at least @xmath256 of the images @xmath247 , we attempt to find @xmath1 linearly independent substitution vectors , call them @xmath266 , such that every substitution polynomial @xmath249 with substitution vector @xmath267 , for @xmath268 , contains the coefficient @xmath251 in a nonzero term . in the bivariate case
this is straightforward .
any @xmath269 linear system formed by two substitution vectors @xmath270 must have nonzero determinant since in the bivariate case the entries are all distinct prime numbers .
the general multivariate case is more involved , as @xmath1 substitution vectors may not always be linearly independent . for this case we will randomly select @xmath228 vectors @xmath271^n$ ] ,
from which we will search for @xmath1 linearly independent vectors . to that end
we require a bound on the probability that such @xmath1 independent vectors do not exist .
[ lem : linsys ] let @xmath64 be a prime number and @xmath99 $ ] , and suppose that a term of @xmath0 avoids collision in an image @xmath272 for a randomly chosen @xmath273^n$ ] with probability at least @xmath215 .
let @xmath274 be row vectors , chosen uniformly from @xmath102^n$ ] .
given that a term of @xmath0 avoids collision in the images @xmath275 for @xmath276 , then @xmath277 has rank less than @xmath1 with probability at most @xmath278 .
since all entries in each @xmath98 , and thereby everything in each @xmath279 and every element in @xmath280 , is less than @xmath64 , we can consider all these objects over the finite field @xmath104 .
if @xmath280 has rank less than @xmath1 , then @xmath281 all lie in some dimension-@xmath120 subspace @xmath123 .
the number of distinct substitution vectors that could lie in the same subspace @xmath122 is @xmath282 .
each dimension-@xmath120 subspace @xmath123 may be specified by a nonzero vector spanning its orthogonal space , unique up to a scalar multiple .
thus the number of such subspaces is less than @xmath283 , and so the number of possible @xmath228-tuples comprised of substitution vectors that do not span @xmath107 is at most @xmath284 meanwhile , there are @xmath285 possible @xmath228-tuples of substitution vectors , and so the probability that such a tuple does span @xmath286 is at most @xmath287 .
furthermore , by the hypothesis , the probability that a term of @xmath0 avoids collision for each substitution @xmath288 is @xmath289 , and thus the conditional probability that @xmath280 is not full rank given that a fixed term of @xmath0 avoids collision for each @xmath288 is at most @xmath290 .
given such a high probability of each term producing a rank-@xmath1 system of substitution vectors , it is a simple matter to show that with high probability _ every _ term of @xmath0 admits some such rank-@xmath1 linear system of substitutions without collisions
. let @xmath99 $ ] as above , and set @xmath211 according to lemma [ lem : nu ] , @xmath149 according to lemma [ lem : m ] , and @xmath291 according to lemma [ lem : lambda ] .
with probability at least @xmath292 , for every term in @xmath0 , there exists a rank-@xmath1 set of substitution vectors @xmath293 , such that the given term of @xmath0 does not collide in any of the substitutions @xmath294 , for @xmath295 . as we have discussed , the case @xmath207 is trivial and when @xmath35 we choose primes for the vectors @xmath98 and the @xmath296 linear systems always have full rank .
so assume @xmath210 and @xmath297 .
we know from lemma [ lem : linsys ] that the probability of a single term _ not _ admitting a rank-@xmath1 system of substitution vectors is less than @xmath278 , so the probability that any term does not have a rank-@xmath1 system of non - colliding substitution vectors is less than @xmath298 .
since @xmath64 is chosen as the least odd prime greater than @xmath223 , we see that @xmath299 . and because @xmath300 and @xmath297 , @xmath301 .
combining this with the probability bound from lemma [ lem : m ] , the overall success probability is at least @xmath302 .
from the set of @xmath228 vectors @xmath303 we can find @xmath1 linearly independent vectors by inspection of the @xmath304 factorization of the matrix whose @xmath228 rows are the @xmath305 . by @xcite , we can do this in @xmath306 operations in @xmath104 , for a bit cost of @xmath307 .
we suppose by reordering of the @xmath305 that @xmath308 are our @xmath1 linearly independent vectors .
then , if @xmath309 is the exponent of the term with coefficient @xmath251 appearing in @xmath310 , we may find the degree @xmath238 of the term with coefficient @xmath251 in the diversified multivariate polynomial @xmath311 by way of the linear system @xmath312 we construct and solve such a linear system for every term of @xmath0 , giving us the polynomial @xmath313 from which we easily obtain @xmath0 as @xmath314 .
procedure [ proc : interp ] describes the approach laid out in sections [ sec : fam - subs]-[sec : lin - sys ] .
[ proc : findprimes ] @xmath315 @xmath316 @xmath317 a @xmath253-diversifying subset of @xmath3 @xmath318 an element of @xmath245 chosen uniformly at random @xmath319 we can now state the tangible benefit of our new kronecker substitution technique . the diversifying sets are included in these theorems even though we do not actually count the cost of possibly extending the ring @xmath3 to include such sets . for given bounds @xmath320 and an unknown polynomial
@xmath38 $ ] with @xmath321 , @xmath322 , and @xmath323 , procedure [ proc : interp ] succeeds in finding @xmath0 with probability at least @xmath292 and requires @xmath265 calls to univariate interpolation with @xmath31 nonzero terms and degree @xmath324 , a @xmath325-diversifying set in @xmath3 , and @xmath326 additional bit operations , where @xmath327 . since @xmath35
, we have @xmath328 .
this is the number of calls to the univariate interpolation algorithm , and the degree bound comes from corollary [ cor : deg - bivar ] .
the size of @xmath137 in the diversifying set comes from the fact that @xmath329 .
two steps dominate the bit complexity .
first , we must choose @xmath330 primes in @xmath331 $ ] .
this can be accomplished via @xmath332 applications of the miller - rabin primality test , performing @xmath333 trials each time to ensure a negligible probability of error .
the cost of these tests is @xmath334 , which is @xmath335 .
the other dominating step in bit complexity is simply the cost of computing with the @xmath234 exponents in images @xmath247 and @xmath31 exponent vectors in the final result .
there are @xmath336 such exponents , each with @xmath337 bits , for a total bit cost of @xmath338 . for given bounds @xmath339 and an unknown polynomial
@xmath99 $ ] with @xmath340 and @xmath323 , procedure [ proc : interp ] succeeds in finding @xmath0 with probability at least @xmath292 and requires @xmath341 calls to univariate interpolation with @xmath31 nonzero terms and degree @xmath133 , a @xmath342-diversifying set in @xmath3 , and @xmath343 additional bit operations , where @xmath344 is the exponent of matrix multiplication .
the analysis of the first two parts is the same as in the bivariate case . for the bit complexity
, we do not have to worry about primality testing here .
however , the size of all exponents in the polynomials becomes @xmath345 , and the cost of performing each lu factorization on a @xmath346 matrix is @xmath347 operations on integers with @xmath265 bits .
as @xmath31 such lu factorizations are required , the total bit cost of the linear algebra is @xmath348 .
interpolating @xmath0 entails the probabilistic steps of ( 1 ) selecting a set of randomized substitutions that produce few collisions and ( 2 ) selecting @xmath149 from a diversifying set @xmath150 such that all term sums in all images have distinct coefficients in those images .
the @xmath297 case has in addition the probabilistic step ( 3 ) of guaranteeing that we can construct a full - rank linear system in order to solve for every exponent of @xmath0 .
the probability of failure in each of these steps has been controlled above so that the overall success probability is at least @xmath292 .
if a higher success probability , say @xmath241 , is desired , we simply run the interpolation algorithm described in sections [ sec : fam - subs][sec : lin - sys ] with some @xmath349 times . again using hoeffding s inequality , the probability that the algorithm fails at least @xmath350 times is at most @xmath351 .
thus , if we wish to discover @xmath0 with probability @xmath241 , we merely run the algorithm as suggested some @xmath352 times , and select the the polynomial @xmath0 that is returned a majority of the time . with probability at least @xmath241 ,
such an @xmath0 exists and is in fact the correct answer .
we have presented a new randomization that maps a multivariate polynomial to a univariate polynomial with ( mostly ) the same terms .
this improves on the usual kronecker map by reducing the degree of the univariate image when the polynomial is known to be sparse .
we have also shown how a small number of such images can be combined to recover the original terms of the unknown multivariate polynomial .
there are numerous questions raised by this result .
perhaps foremost is whether there is any practical gain in any particular application by using this approach .
we know that the randomized kronecker substitution will result in smaller degrees than the usual kronecker substitution whenever the polynomial is sufficiently large and sufficiently sparse , so in principle the applications should include any of the numerous results on sparse polynomials that use a kronecker substitution to accommodate multivariate polynomials . unfortunately , in practice , the situation is not so clear .
many of the aforementioned results that rely on a kronecker substitution either do not have a widely - available implementation , or do not usually involve sparse polynomials .
however , for the particular applications of sparse gcd and sparse multivariate multiplication , there is considerable promise particularly in the case of bivariate polynomials with degree greater than 1000 or so and sparsity between @xmath2 and @xmath353 .
an efficient implementation comparison in these situations would be useful and interesting , and we are working in that direction .
there are also questions of theoretical interest . for one ,
we would like to know how far off the bounds on the size of primes from lemmata [ lem : lambda - pq ] and [ lem : lambda ] are compared to what is really necessary to avoid collisions .
an important question is whether our current results are optimal in any sense . in the bivariate case , when @xmath354 our result gives @xmath355 , which is optimal in terms of @xmath31 .
that is because @xmath31 could be as large as @xmath356 , and therefore any monomial substitution exponent less than @xmath357 would by necessity have more than a constant fraction of collisions .
however our result for @xmath185 gives each @xmath358 , which in terms of @xmath31 is off by a factor of @xmath359 from the optimal .
it may be possible to improve these bounds simply with a better analysis , or with a different kind of randomized monomial substitution . in either case , it is clear that , at least for @xmath297 and in particular for @xmath360 , it should be possible to improve on the results here and achieve univariate reduced polynomials with even lower degree . another interesting question would be whether some of this randomization can be avoided .
here we have two randomizations , the diversification and the ( multiple ) randomized kronecker substitutions . and
this is besides any randomization that might occur in the underlying univariate algorithm !
it seems plausible that , for example in the application of multivariate multiplication , the known aspects of the monomial structure might be used to make some choices less random and more `` intelligent '' .
however , we do not yet know any reasonable way to accomplish this .
we wish to thank zeev dvir for pointing out the previous work of klivans and spielman , and the reviewers for their helpful comments .
we also thank the organizers of the siam ag13 meeting for the opportunity to discuss preliminary work on this topic .
the second author is supported by the national science foundation , award # 1319994 .
michael ben - or and prasoon tiwari .
a deterministic algorithm for sparse multivariate polynomial interpolation . in _ proceedings of the twentieth annual acm symposium on theory of computing _ , stoc 88 , pages 301309 , new york , ny , usa , 1988 .
doi : 10.1145/62212.62241 .
jrmy berthomieu and grgoire lecerf .
reduction of bivariate polynomials from convex - dense to dense , with application to factorizations . _ math .
_ , 81:0 17991821 , 2012 .
doi : 10.1090/s0025 - 5718 - 2011 - 02562 - 7
. james r. bunch and john e. hopcroft . triangular factorization and inversion by fast matrix multiplication .
_ mathematics of computation _ , 280 ( 125):0 231236 , 1974 .
url http://www.jstor.org/stable/2005828 .
emmanuel j. cands , justin k. romberg , and terence tao .
stable signal recovery from incomplete and inaccurate measurements .
_ communications on pure and applied mathematics _ , 590 ( 8):0 12071223 , 2006 .
doi : 10.1002/cpa.20124 .
matthew t. comer , erich l. kaltofen , and clment pernet .
sparse polynomial interpolation and berlekamp / massey algorithms that correct outlier errors in input values . in _ proceedings of the 37th international symposium on symbolic and algebraic computation _
, issac 12 , pages 138145 , new york , ny , usa , 2012 .
doi : 10.1145/2442829.2442852 .
mark giesbrecht and daniel s. roche .
diversification improves interpolation . in _ proceedings of the 36th international symposium on symbolic and algebraic computation _
, issac 11 , pages 123130 , new york , ny , usa , 2011 .
doi : 10.1145/1993886.1993909 .
mark giesbrecht , george labahn , and wen - shin lee .
symbolic - numeric sparse interpolation of multivariate polynomials .
_ journal of symbolic computation _ , 440 ( 8):0 943 959 , 2009 .
doi : 10.1016/j.jsc.2008.11.003 .
dima yu .
grigoriev , marek karpinski , and michael f. singer .
fast parallel algorithms for sparse multivariate polynomial interpolation over finite fields .
_ siam journal on computing _ , 190 ( 6):0 10591063 , 1990 .
doi : 10.1137/0219073 .
ming - deh a huang and ashwin j rao .
interpolation of sparse multivariate polynomials over large finite fields with applications .
_ journal of algorithms _ , 330 ( 2):0 204228 , 1999 .
doi : 10.1006/jagm.1999.1045 .
seyed mohammad mahdi javadi and michael monagan .
on factorization of multivariate polynomials over algebraic number and function fields . in _ proceedings of the 2009 international symposium on symbolic and algebraic computation _ , issac 09 , pages 199206 , new york , ny , usa , 2009 .
doi : 10.1145/1576702.1576731 .
seyed mohammad mahdi javadi and michael monagan .
parallel sparse polynomial interpolation over finite fields . in _ proceedings of the 4th international workshop on parallel and symbolic computation _ , pasco 10 , pages 160168 , new york , ny , usa , 2010 .
doi : 10.1145/1837210.1837233 .
erich l. kaltofen .
fifteen years after dsc and wlss2 : what parallel computations i do today [ invited lecture at pasco 2010 ] . in _ proceedings of the 4th international workshop on parallel and symbolic computation _ ,
pasco 10 , pages 1017 , new york , ny , usa , 2010 .
doi : 10.1145/1837210.1837213 .
adam r. klivans and daniel spielman .
randomness efficient identity testing of multivariate polynomials . in _ proceedings of the thirty - third annual acm symposium on theory of computing _
, stoc 01 , pages 216223 , new york , ny , usa , 2001 .
doi : 10.1145/380752.380801 .
arnold schnhage .
asymptotically fast algorithms for the numerical multiplication and division of polynomials with complex coefficients . in jacques calmet , editor , _ computer algebra _ ,
volume 144 of _ lecture notes in computer science _ , pages 315 .
springer berlin heidelberg , 1982 .
doi : 10.1007/3 - 540 - 11607 - 9_1 .
amir shpilka and amir yehudayoff .
arithmetic circuits : a survey of recent results and open questions . _ foundations and trends in theoretical computer science _ , 50 ( 3 - 4):0 207388 , 2010 .
doi : 10.1561/0400000039 .
richard zippel .
probabilistic algorithms for sparse polynomials . in edward
ng , editor , _ symbolic and algebraic computation _ , volume 72 of _ lecture notes in computer science _ , pages 216226 .
springer berlin / heidelberg , 1979 .
doi : 10.1007/3 - 540 - 09519 - 5_73 .
richard zippel .
interpolating polynomials from their values .
_ journal of symbolic computation _ , 90 ( 3):0 375403 , 1990 .
doi : 10.1016/s0747 - 7171(08)80018 - 1 .
computational algebraic complexity editorial . | we present new techniques for reducing a multivariate sparse polynomial to a univariate polynomial .
the reduction works similarly to the classical and widely - used kronecker substitution , except that we choose the degrees randomly based on the number of nonzero terms in the multivariate polynomial .
the resulting univariate polynomial often has a significantly lower degree than the kronecker substitution polynomial , at the expense of a small number of term collisions . as an application
, we give a new algorithm for multivariate interpolation which uses these new techniques along with any existing univariate interpolation algorithm . |
despite , or perhaps because of the uncertainties in representing the strong interaction by a potential , it has been a longstanding quest in nuclear physics to find simple algebraic relationships between the excitation energies of a nucleus wherever possible .
attempts to achieve this have often been inspired by approximate symmetries such as wigner s supermultiplets @xcite or elliot s su(3 ) scheme @xcite .
a textbook example leading to a simple energy formula is provided by the rotational - vibrational model @xcite .
later , the interacting boson model @xcite used linear combinations of casimir operators of various chains of subgroups of u(6 ) to generate simple expressions for nuclear spectra .
indeed , investigations into algebraic cluster models of energy levels continue to provide fruitful fields of research to the present day @xcite . however , the symmetries invoked by these models are never perfectly realised in real nuclei , and although suggestively close agreement can often be obtained in a number of favourable cases , they generally struggle to achieve precision at the level of a few 10 s of kev .
we suggest that it might be profitable to analyse the experimental data more closely without any preconceived ideas about underlying models , to see if this alone suggests any simple relations between excitation energies that could be used for predictive purposes
. there have already been such suggestions from our earlier work @xcite and encouraged by these findings we examine here the deviations of our previous expressions from measured values with a view to improving their precision without sacrificing their simplicity .
it will turn out that the experimental excitation energies of five consecutive states of the ground state band must be known for our recurrence relations to get started .
whenever this applies , they will provide a useful tool for predicting and/or confirming the energies of the succeeding states of that band , and thus will facilitate the task of nuclear data evaluation .
we believe they should be superior to the common practice of seeking to fit the energies with a rotationally inspired formula such as @xmath0 ^ 2 $ ] , since they are based purely on empirical values and are completely independent of any underlying model . in particular , they will be helpful in predicting the energy of the next state in a band beyond the highest identified to date .
we have previously analysed recursion relations involving the angular momenta @xmath1 and corresponding excitation energies @xmath2 of states belonging to the ground state bands of even - even nuclei @xcite . given a pair of angular momenta @xmath3 with @xmath4 we found empirically that @xmath5 where @xmath6 increases in steps of two from an initial value @xmath7 .
we note that as the spin values @xmath8 and @xmath9 change with @xmath6 their sum remains equal to @xmath10 . in the earlier analysis @xcite we found that eq.(1 ) was satisfied to within a few percent with the deviations from that equation strongly correlated .
these deviations are given by @xmath11 and here we investigate the nature and extent of the correlations between the various @xmath12 , and how they can be used to improve the agreement between theory and data .
in the database shown in table 1 @xcite we have included all 87 even - even nuclei between ba and cm which have more than four protons ( or proton - holes ) and four neutrons ( or neutron - holes ) outside closed shells , and for which the ground state bands are accurately known ( i.e. with firm @xmath13 assignments ) up to at least @xmath14 .
we concentrate on the typical case defined by @xmath15 and @xmath16 leaving a more complete analysis to be presented elsewhere @xcite .
figure 1 shows that the deviations are surprisingly well described by the simple relation @xmath17 indeed , a least squares straight line fit to all 87 data points yields a gradient of @xmath18 and an intercept of @xmath19 .
the figure also indicates that the six points with the highest values of @xmath20 deviate from the straight line in a statistically significant manner .
these points correspond to the nuclei @xmath21ce , @xmath22ce , @xmath22nd , @xmath23sm , @xmath24gd and @xmath25os .
the results for @xmath22ce are probably problematic due to backbending .
there are two @xmath26 states in close proximity at 2.81111 and 3.0176 mev ( we take the first of them as the band member ) and similarly three @xmath27 states at 3.2086 , 3.7193 and 3.81765 mev ( we take the second of these as the band member ) .
this band crossing feature is generally expected for nuclei with low deformation at the beginning and end of the rare - earth region @xcite . it therefore comes as no surprise that similar deviations , albeit to a lesser extent , are observed for four other nuclei in the neighbourhood of @xmath22ce .
table 2 lists the states of the ground state band , and an excited band which can be expected to mix with it , for all six of these outlying nuclei .
we regard these few deviations in a positive light as pinpointing unusual nuclear spectra that are worthy of more detailed scrutiny , and here we would include also @xmath28sm , @xmath29os , @xmath30th and in particular @xmath31pu . the framework introduced above can be used to locate missing members of a band , or to check experimental assignments to the band .
we illustrate the procedure for the case where all the band members up to @xmath32 are known and an estimate of the excitation energy of the @xmath33 state is required . by eq.(2 ) @xmath34 which together with eq.(3 ) result in @xmath35 table 1 lists the individual ratios @xmath36 as well as the differences @xmath37 between the theoretical values of @xmath38 given by eq.(5 ) and their experimental counterparts . omitting the clearly anomalous case of @xmath22ce we find an r.m.s .
deviation of @xmath39 kev , an order of magnitude better than the results obtained when setting @xmath40 in either of eqs.(4 ) . of interest
also is that the further analysis undertaken below suggests that eq.(5 ) is a particular case of a more general recurrence relation given by @xmath41 with @xmath42 .
this energy recurrence relation can also be obtained by assuming that the excitation energy @xmath43 may be expanded as a taylor series in angular momentum @xmath44 which converges sufficiently rapidly that fifth order and higher terms may be neglected .
expanding @xmath45 , @xmath46 and @xmath47 about @xmath48 we obtain @xmath49 @xmath50 @xmath51 by successively subtracting the @xmath52 and @xmath53 terms from their @xmath54 and @xmath55 counterparts in the three equations above we completely eliminate all even powers from the taylor series expansions obtaining respectively , @xmath56 = 2e^{\prime}(n_0 ) + { e^{\prime\prime\prime}(n_0 ) \over 3 } + \ldots\ ] ] @xmath57 = 6e^{\prime}(n_0 ) + 9e^{\prime\prime\prime}(n_0 ) + \ldots\ ] ] @xmath58 = 10e^{\prime}(n_0 ) + { 250e^{\prime\prime\prime}(n_0 ) \over 6 } + \ldots\ ] ] subtracting twice eq.(10 ) from eq.(11 ) yields @xmath59 and the right hand side of eq.(12 ) is exactly five times that of eq.(13 ) .
this allows us to write @xmath60 so that taking @xmath61 we obtain eq.(5 ) which , taken together with eq.(4 ) , yields eq.(3 ) . also by putting @xmath62 in eq.(14 )
we recover eq.(6 ) .
we thus find that eq.(6 ) , which with @xmath63 generates the excellent agreement with the data shown in table 1 , can be derived from a taylor series expansion if terms of order five and higher are ignored .
of course , we have not rigorously proved convergence of the taylor series , which depends on the ( unknown ) behaviour of the higher derivates of @xmath64 as a function of @xmath65 .
what we have done , starting from a taylor series , is to completely eliminate all even powers of @xmath65 , and show that eqs.(3 ) , ( 5 ) and ( 6 ) follow if the remaining odd terms of order 5 and higher are ignored .
this suggests that the band excitation energies @xmath2 considered here take the form of a polynomial of maximum degree four in the angular momentum @xmath65 , and we note that only cubic forms were considered in earlier work @xcite . since eq.(6 )
is satisfied by any e - dependence on @xmath65 that is of quartic or lower degree , it encompasses a number of models .
it is satisfied , for example , by the idealised quadratic form @xmath66 of a perfect rotor with a constant moment of inertia , or by the anharmonic vibrator ( ahv ) model for which @xmath67 , where @xmath68 and @xmath69 and @xmath70 are parameters fitted to each nucleus in turn @xcite .
we note , however , that although eq.(6 ) is obeyed by these models it is satisfied to the extent that inserting the model quantities on the right hand side of eq.(6 ) generates the model quantity on the left hand side .
as stated in our introductory paragraph , our emphasis is to extend previous work @xcite by finding more accurate _ direct _ relations between _ experimental _ quantities .
in conclusion we have found strong correlations between the deviations @xmath12 from our original relations of eq.(1 ) above .
this has enabled us to write down a highly successful recursion relation , not involving any free parameters , for locating members of ground state bands from a knowledge of the excitation energies of the lower members of the band .
for the specific case of @xmath15 and @xmath16 we have shown how a knowledge of the excitation energies of the @xmath71 band members allows us to predict the excitation energy of the @xmath33 state with an r.m.s .
error of @xmath72 kev .
we note that this is likely to be a particular case of a more general relation .
although we have restricted attention here to values of @xmath65 no higher than 10 , we have observed that eq.(6 ) can be used to predict energies for states with much greater values of @xmath65 ( up to @xmath73 in some cases ) and we intend to investigate this further elsewhere .
1truecm * acknowledgements : * one of us ( s.m.p . )
acknowledges partial support for this work from the national research foundation of south africa under grant uid 85382 . the opinions , findings and conclusions expressed in any publication generated by nrf supported research are those of the authors , and the nrf accepts no liability whatsoever in this regard . 99 e.p.wigner , phys
51 * 105 ( 1937 ) j.p.elliott , proc .
* a245 * 128 ( 1958 ) and * a245 * 562 ( 1958 ) a.bohr and b.r.mottelson , nuclear structure ( benjamin , new york , 1969 ) vol.2 p.10 e.eichler , rev .
* 36 * 809 ( 1964 ) a.arima and f.iachello , advances in nuclear physics , * 13 * 139 ( 1984 ) j.cseh and g.levai , ann . phys . * 230 * 165 ( 1994 ) b.buck , a.c.merchant and s.m.perez , phys . rev .
* c81 * 034322 ( 2010 ) brookhaven national nuclear data center at http://www.nndc.bnl.gov/ensdf/ b.buck , a.c.merchant and s.m.perez in preparation k.hera and y.sum , nucl .
a529 * 445 ( 1991 ) n.v.zamfir and r.f.casten , phys .
lett . * 75 * 1280 ( 1995 ) against the deviations @xmath74 for the 87 nuclei listed in table 1 .
a least squares straight line fit to the data indicates a slope of @xmath18 and an intercept of @xmath19 .
the solid line corresponds to @xmath75 . ]
.database of 87 rare - earth and actinide nuclei .
we highlight the ratios @xmath76 from eq.(4 ) , which are all close to 2.0 , and @xmath37 the difference between the values of the @xmath27 excitation energy derived using eq.(5 ) and that obtained from direct measurement .
[ cols="<,^,^,^,^ , > , > , > , > , > " , ] | the excitation energies of states belonging to the ground state bands of heavy even - even nuclei are analysed using recurrence relations .
excellent agreement wih experimental data at the 10 kev level is obtained by taking into account strong correlations which emerge in the analysis .
this implies that the excitation energies can be written as a polynomial of maximum degree four in the angular momentum .
25.5pt 21.10.-k , 21.10.hw , 23.20.lv and 27.20.+n |
relativistic jets are the most extreme expression of the power that can be generated by a super - massive black hole ( smbh ) in the center of an agn .
these objects have a total bolometric luminosity of up to 10@xmath5 erg s@xmath4 ( e.g. * ? ? ?
* ; * ? ? ?
* ) , with a large fraction of the power emitted at high energies .
before the launch of the _ fermi _ satellite only two classes of agn were known to generate relativistic jets and to emit in the @xmath0-ray energy range : blazars and radio galaxies , both hosted in giant elliptical galaxies @xcite .
the first two years of observations by _ fermi_-lat confirmed that these two populations represent the majority of the identified sources in the extragalactic @xmath0-ray sky @xcite .
however the discovery of variable @xmath0-ray emission from four radio - loud nlsy1 galaxies revealed the presence of a possible third class of agn with relativistic jets @xcite .
the nlsy1s are a class of agn identified by @xcite and characterized by the following optical properties : narrow permitted lines ( fwhm ( h@xmath6 ) @xmath7 2000 km s@xmath4 ) , [ oiii]/h@xmath6 @xmath7 3 , and a fe ii bump ( for a review see * ? ? ?
they have smaller central black hole masses ( i.e. 10@xmath8 - 10@xmath9 m@xmath10 ) than those in blazars and radio galaxies , and higher accretion rates close to or above the eddington limit @xcite .
these sources are generally radio - quiet , with only a small fraction ( @xmath77% ) classified as radio - loud @xcite , and objects with high values of radio - loudness ( @xmath11 ) are even more sparse ( @xmath22.5% ) .
the strong radio emission and the flat radio spectrum , together with variability studies , suggest the presence of a relativistic jet in some radio - loud nlsy1s . this has been confirmed by the _ fermi_-lat detection of @xmath0-ray emission from some of them . by considering that nlsy1s are usually hosted in spiral galaxies ( e.g. * ? ? ?
* ) the presence of a relativistic jet is in contrast to the paradigm that the formation of relativistic jets could happen only in elliptical galaxies ( see e.g. * ? ? ?
* ) , giving an indication that relativistic jets can form and develop independently of their host galaxies .
sbs 0846@xmath1513 was not in the first and second fermi - lat ( 1fgl and 2fgl ) catalogues , indicating that the source was not detected with test statistic ( ts ) @xmath12 25 in either one year or two years of _ fermi _ observations @xcite . integrating over the first two years of _ fermi _ operation
the likelihood fit yielded a ts = 14 , with a 2-@xmath13 upper limit of 8.5@xmath1410@xmath15 photons @xmath16 s@xmath4 in the 0.1300 gev energy range , assuming a photon index @xmath17 . on the contrary ,
the likelihood fit with a power - law model to the data integrated over the third year of _ fermi _ operation ( 2010 august 4 2011 august 4 ; mjd 5541255777 ) in the 0.1300 gev energy range results in a ts = 653 , with an integrated average flux of ( 6.7 @xmath18 0.5 ) @xmath1410@xmath19 ph @xmath16 s@xmath4 and a photon index of @xmath20 = 2.23 @xmath18 0.05 .
figure [ fig1 ] ( left panel ) shows the @xmath0-ray light curve of the third year of _ fermi _ observations built using 1-month time bins .
if ts @xmath7 10 the value of the flux was replaced by the 2-@xmath13 upper limits .
a clear increase of the flux was observed in the period 2011 june 4july 4 , for which 5-day time bins were used in fig .
[ fig1 ] . considering the high activity of the source we extracted a spectrum over that period , obtaining a photon index of @xmath20 = 1.98 @xmath18 0.05 and a flux of ( 24.4 @xmath18 2.1 ) @xmath1410@xmath19 ph @xmath16 s@xmath4 .
the spectral evolution during the flaring activity in 2011 june observed in @xmath0 rays from sbs 0846@xmath1513 is a common behaviour in bright fsrqs and low - synchrotron - peaked bl lacs detected by _
@xcite , with a change in photon index @xmath7 0.20.3 .
the peak of the emission was observed on 2011 july 1 , with a flux of ( 98 @xmath18 19 ) @xmath1410@xmath19 ph @xmath16 s@xmath4 in the 0.1300 gev energy range , corresponding to an isotropic @xmath0-ray luminosity of @xmath21.1@xmath1410@xmath3 erg s@xmath4 .
further details about the lat data analysis and results are given in @xcite .
100 mev ) light curve of sbs0846@xmath1513 obtained from 2010 august 4 to 2011 august 4 .
arrows refer to 2-@xmath13 upper limits on the source flux . _
right panel : _ vlba image at 15.3 ghz of sbs 0846 + 513 . on the image we provide the peak flux density , in mjy / beam , and the first contour intensity ( f.c .
, in mjy / beam ) that corresponds to three times the noise measured on the image plane .
contour levels increase by a factor of 2 .
the beam is plotted on the bottom left corner of the image .
the vectors superimposed on the total intensity contours show the position angle of the e vector , where 1 mas length corresponds to 2.9 mjy / beam.,width=226 ] 100 mev ) light curve of sbs0846@xmath1513 obtained from 2010 august 4 to 2011 august 4 .
arrows refer to 2-@xmath13 upper limits on the source flux . _
right panel : _ vlba image at 15.3 ghz of sbs 0846 + 513 . on the image we provide the peak flux density , in mjy / beam , and the first contour intensity ( f.c .
, in mjy / beam ) that corresponds to three times the noise measured on the image plane .
contour levels increase by a factor of 2 .
the beam is plotted on the bottom left corner of the image .
the vectors superimposed on the total intensity contours show the position angle of the e vector , where 1 mas length corresponds to 2.9 mjy / beam.,width=170 ]
when imaged with the high spatial resolution of the vlba sbs 0846@xmath1513 is resolved in two components with a core - jet structure , as also pointed out in a previous work by @xcite . at 15 ghz the jet structure ( component e in fig . [ fig1 ] , right panel ) shows an extended low - surface brightness structure with a steep spectrum . on the other hand the core ( component w ) is resolved into two compact components ( labelled w1 and w2 in fig .
[ fig1 ] , right panel ) . to investigate a possible proper motion of the jet
, we compared the separation between w1 , considered the core region , and w2 , assumed to be a knot in the jet , at the four mojave observing epochs available up to now . we model - fitted the visibility data using gaussian components of the four - epoch data by means of the model - fitting option in difmap . from this comparison we found that w1 and w2 are separating with an apparent velocity of ( 0.32 @xmath18 0.04 ) mas / yr , which corresponds to ( 10.9 @xmath18 1.4)@xmath21 .
this apparent superluminal velocity suggests the presence of boosting effect . before the @xmath0-ray flaring episode
, the simultaneous multifrequency observations performed by effelsberg showed a flat radio spectrum up to 32 ghz .
after the flare , the spectral shape observed changed , becoming convex .
the spectral variability was also accompanied by variations in the radio flux density , as observed by ovro and medicina telescopes , indicating a blazar - like behaviour .
further details are given in @xcite .
after pmn j0948@xmath10022 @xcite , sbs 0846@xmath1513 is the second nlsy1 observed to generate such a high power in @xmath0 rays .
this could be an indication that all the radio - loud nlsy1s are able to host relativistic jets as powerful as those in blazars , despite the lower bh mass ; alternatively some nlsy1s could have peculiar characteristics allowing the development of these relativistic jets .
the mechanism at work for producing a relativistic jet is not clear , and the physical parameters that drive the jet formation is still under debate .
one fundamental parameter could be the black hole mass , with only large masses allowing for the efficient formation of a relativistic jet .
the large radio loudness of sbs 0846@xmath1513 could challenge this idea if the black hole mass ( between 8.2@xmath22 and 5.2@xmath23 ) estimated by @xcite is confirmed . according to the `` modified spin paradigm '' discussed in @xcite , another fundamental parameter for
the efficiency of a relativistic jet production should be the bh spin , with smbhs in elliptical galaxies having on average much larger spins than smbhs in spiral galaxies . in this context the discovery of relativistic jets in a class of agn usually hosted by spiral galaxies , such as the nlsy1s , was a great surprise .
we note that bh masses of radio - loud nlsy1s are generally larger with respect to the entire sample of nlsy1s @xcite , even if still small when compared to radio - loud quasars .
the larger bh masses of radio - loud nlsy1s with respect to radio - quiet nlsy1s could be related to prolonged accretion episodes that can spin - up the bhs .
the small fraction of radio - loud nlsy1s with respect to radio - loud quasars could be an indication that in the former the high accretion usually does not last sufficiently long to significantly spin - up the bh @xcite .
to conclude , sbs 0846@xmath1513 shows all the characteristics of the blazar phenomenon and could be a low mass ( and possible younger ) version of blazar .
the _ fermi _ 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 .
this research has made use of data from the mojave database that is maintained by the mojave team ( lister et al .
2009 , aj , 137 , 3718 ) . | the presence of a relativistic jet in some radio - loud narrow - line seyfert 1s ( nlsy1 ) galaxies , first suggested by their variable radio emission and the flat radio spectra , is now confirmed by the _ fermi_-lat detection of five nlsy1s in @xmath0 rays . in particular , a strong @xmath0-ray flare from sbs 0846@xmath1513 was observed in 2011 june by _
fermi_-lat reaching a @xmath0-ray luminosity ( 0.1300 gev ) of @xmath210@xmath3 erg s@xmath4 , comparable to that of bright flat spectrum radio quasars . apparent superluminal velocity in the jet was inferred from 20112012 vlba images , suggesting the presence of a highly relativistic jet . both the power released by this object during the flaring activity and the apparent superluminal velocity are strong indicators of the presence of a relativistic jet as powerful as those in blazars .
in addition , variability and spectral properties in radio and @xmath0-ray bands indicate a blazar - like behaviour , suggesting that , except for some distinct optical characteristics , sbs 0846@xmath1513 could be considered as a young blazar at the low end of the blazar s black hole mass distribution .
address = dip .
di fisica , universit degli studi di perugia and infn , via a. pascoli , i-06123 perugia , italy address = dip . di astronomia , universit di bologna , via ranzani 1 , i-40127 bologna , italy + inaf - istituto di radioastronomia , via gobetti 101 , i-40129 bologna , italy address = u.s . naval research laboratory , code 7653 ,
4555 overlook ave .
sw , washington , dc 20375 - 5352 , usa |
let @xmath7 be an oriented matroid , of rank @xmath8 , on the ground set @xmath9 @xmath10 , with set of topes @xmath11 ; throughout we will suppose that it is _ simple _ , that is , it contains no loops , parallel or _
antiparallel _ elements . in this paper
the sign components @xmath12 and @xmath13 of maximal covectors are replaced by the real numbers @xmath1 and @xmath2 , respectively .
the topes @xmath14 are interpreted as elements of the real euclidean space @xmath15 of row vectors ; if @xmath16 then @xmath17 , and @xmath18 .
wedenoteby@xmath19thestandardbasisof@xmath15,thatis,@xmath20 + @xmath21 , @xmath22 .
the _ positive tope _
@xmath23 is the vector @xmath24 @xmath25 .
if @xmath26 then the corresponding _ positive halfspace _
@xmath27 is the tope subset @xmath28 . if @xmath29 and @xmath30 then @xmath31 denotes the vector obtained from @xmath32 by _ sign reversal _ or _ reorientation _ on the set @xmath33 : @xmath34 when @xmath35 , and @xmath36 when @xmath37 .
the oriented matroid whose topes are obtained from the topes of @xmath0 by reorientation on the set @xmath33 is denoted by @xmath38 .
we denote by @xmath39 the _ negative part _ @xmath40 of the tope @xmath32 ; the _ positive part
_ @xmath41 of @xmath32 is the set @xmath42 .
the vertices of the _ tope graph _
@xmath43 of the oriented matroid @xmath0 are its topes ; a pair of topes @xmath44 is an edge of the graph @xmath43 if the topes @xmath45 and @xmath46 are _ adjacent _ , that is , they cover some _ subtope _ in the _ big face lattice _ of @xmath0 .
separation set _
@xmath47 of topes @xmath45 and @xmath46 is defined by @xmath48 .
the _ graph distance _
@xmath49 between the topes @xmath45 and @xmath46 is the cardinality of the separation set @xmath47 , see ( * ? ? ?
4.2.3 ) , that is , @xmath50
if @xmath51 then we denote by @xmath52 the _ tope poset _ of @xmath0 based at the tope @xmath5 ; by convention , we have @xmath53 in @xmath52 iff @xmath54 @xmath55 .
if @xmath56 then @xmath57 stands for the set of minimal elements of the subposet @xmath58 .
let @xmath59 be a maximal chain in the tope poset @xmath52 , and @xmath60 .
the union @xmath3@xmath61@xmath62isthevertexsetofthe@xmath4 + @xmath63 in the tope graph @xmath43 ; by convention , we have @xmath64 , @xmath65 . the subset of topes from @xmath3 with inclusion - maximal positive parts is denoted by @xmath66 . if @xmath30 then @xmath67 .
let @xmath68 be the sequence defined by @xmath69 ; note that @xmath70 .
the chain @xmath71 is a basis of the space @xmath15 ; indeed , the square _ sign matrix _
@xmath72 is similar to the nonsingular matrix @xmath73 the absolute value of its determinant is @xmath74 .
the @xmath75th row @xmath76 , @xmath22 , of the inverse matrix @xmath77 of @xmath78 is @xmath79 thus , if @xmath29 then we have @xmath80 for some row vector @xmath81 such that @xmath82 a subset @xmath83 is called a _ tope committee _ for @xmath0 if @xmath84 see @xcite .
the committee @xmath85 is called _ minimal _ if any its proper subset is not a committee for @xmath0 .
if the sum of the members of the minimal tope committee @xmath85 is the positive tope , @xmath86 then we say that the committee @xmath85 is _
critical_. recall that if @xmath4 is a symmetric cycle in the tope graph @xmath43 of the oriented matroid @xmath0 , then for any tope @xmath29 there exists a unique inclusion - minimal subset @xmath87 such that @xmath88 this set , of odd cardinality , of linearly independent elements of @xmath15 is @xmath89 where the vector @xmath90 is defined by ( [ eq:4 ] ) . as a consequence , a tope subset @xmath83 is a committee for @xmath0 iff @xmath91 in section [ section:1 ] we discuss relation ( [ eq:2 ] ) and describe the structure of the sets @xmath92 in more detail .
for a tope @xmath32 of the oriented matroid @xmath0 and for a symmetric cycle @xmath4 in its tope graph , we describe the corresponding inclusion - minimal subset @xmath87 from expression ( [ eq:2 ] ) in terms of the tope poset of @xmath0 .
dual assertions could be made because the tope poset of @xmath0 is self - dual in the sense of ( * ? ? ?
4.2.15(ii ) ) .
let @xmath93 be a symmetric cycle in the tope graph @xmath43 of the oriented matroid @xmath0 .
pick a tope @xmath51 and consider the vertex set @xmath3 of the cycle @xmath4 as a subposet of the tope poset @xmath52 based at @xmath5 . if @xmath97 , there is nothing to prove .
suppose that @xmath98 , and reorient the items of the poset @xmath52 on the negative part @xmath99 of the tope @xmath5 ; in other words , consider the tope poset @xmath100 of the acyclic oriented matroid @xmath101 based at the positive tope @xmath23 .
if@xmath102isatopeoftheorientedmatroid@xmath101,thentheposet + rank @xmath103 of @xmath102 in the tope poset @xmath100 of @xmath101 is equal to the cardinality @xmath104 of the negative part of @xmath102 .
atope@xmath102oftheorientedmatroid@xmath101belongstothe + set@xmath105iffforthe@xmath106-path@xmath107,where@xmath108,inthe + symmetriccycle@xmath109@xmath110@xmath111@xmath112@xmath113we + have @xmath114 . by (
5.6 ) , the set @xmath115 of minimal elements of the subposet @xmath116 is the inclusion - minimal subset of topes with the property @xmath117 @xmath118 , that is , @xmath119 this means that @xmath120 is a critical tope committee for the acyclic oriented matroid @xmath101 .
relation ( [ eq:1 ] ) is equivalent to ( [ eq:3 ] ) .
consider the hasse diagram , which is depicted in figure [ figure:1](a ) , of the tope poset @xmath52 of a simple oriented matroid @xmath0 of rank @xmath123 , where @xmath124 .
fix the symmetric cycle @xmath4 @xmath61 @xmath125 @xmath126@xmath127@xmath128@xmath129 + @xmath130@xmath131@xmath132@xmath133 + @xmath134 @xmath135 in the tope graph of @xmath0 .
thehassediagram , borrowedfrom ( * ? ? ?
4.2.2),ofthetope + poset @xmath136 of the acyclic oriented matroid @xmath137 is shown in figure [ figure:1](b ) .
we have @xmath138 @xmath139 and @xmath140 similarly , @xmath141 and @xmath142 based at the tope @xmath124 , and the @xmath3 which is the vertex set of a symmetric cycle @xmath4 in the tope graph .
( b ) : the hasse diagram of the tope poset @xmath136 based at the positive tope @xmath143 . ]
the subposet of the poset @xmath52 , depicted in figure [ figure:2 ] , is the vertex set of another symmetric cycle @xmath144 @xmath126@xmath127@xmath128@xmath145 @xmath130@xmath131@xmath132@xmath133 + @[email protected]@xmath6 @xmath147 and @xmath148
let @xmath149 @xmath127@xmath150@xmath145@xmath130 + @xmath151 be one more symmetric cycleinthetopegraphoftheorientedmatroid@xmath0,seefigure.pick + thetope@[email protected]@xmath121 + @xmath153 , we have @xmath154 | let the sign components of the maximal covectors of a simple oriented matroid @xmath0 be represented by the real numbers @xmath1 and @xmath2 . consider the vertex set @xmath3 of a symmetric cycle @xmath4 of adjacent topes in the tope graph of @xmath0 as a subposet of the tope poset of @xmath0 .
if @xmath5 is the bottom element of the tope poset then @xmath5 is equal to the unweighted sum of the members of the set @xmath6 of minimal elements of the subposet @xmath3 ; if @xmath5 is the positive tope then the set @xmath6 is a critical tope committee for the acyclic oriented matroid @xmath0 . |
for a long time , we believed that landau symmetry - breaking theory@xcite describes all possible orders in materials , and all possible ( continuous ) phase transitions .
however , in last twenty years , it has become more and more clear that landau symmetry - breaking theory does not describe all possible orders . for example , different fractional quantum hall ( fqh ) states@xcite all have the same symmetry .
thus it is impossible to use symmetry breaking to characterize different fqh states .
if landau symmetry breaking theory is not enough , then what should we use to describe those new states of matter ?
it turns out that we need to develop a totally new theory , to described the new types of orders
topological / quantum order@xcite that appear in the fqh states and the spin liquid states . in , a systematic understanding of a large class of topological orders in strongly correlated bosonic systems without symmetry has been developed based on string - net condensations . in , the string - net classification of topological orders was generalized , based on local unitary ( lu ) transformations . in , topological orders with symmetry are studied using projective symmetry group and tensor network renormalization .
but so far , we still do not have a complete classification of topological orders for interacting systems .
recently , for non - interacting gapped fermion systems with certain symmetries , a complete classification of topological phases has been developed based on the k - theory.@xcite a generalization of the free fermion result to interacting cases has been obtained for 1d systems.@xcite for 1d bosonic systems , studied quantized berry phases for spin rotation symmetric systems . studied the entanglement spectrum and the symmetry properties of matrix product states . using those tools
, they obtained some interesting results which are special cases of the situation considered here .
in this paper , we will apply the approach used in to 1d strongly correlated systems .
we will combine the local unitary transformation with the symmetry properties of matrix product states , and try to obtain a complete classification of all gapped phases in 1d quantum spin systems with certain simple symmetries .
we find that 1 .
_ if there is no symmetry ( including translation symmetry ) , all gapped 1d spin systems belong to the same phase . _
_ for 1d non - translation invariant ( nti ) spin systems with only an on - site symmetry described by a group @xmath0 , all the phases of gapped systems that do not break the symmetry are classified by the equivalence classes in the second cohomology group @xmath1 of the group @xmath0 , provided that the physical states on each site form a linear representation of the group @xmath0 . _ ( note that the equivalence classes in @xmath1 classify the types of projective representations of @xmath0 over the field @xmath2 of complex numbers .
appendix [ prorep ] gives a brief introduction on projective representation and the second cohomology group . ) in certain cases where @xmath0 has infinitely many 1d representations , for example when @xmath3 , further classification according to different 1d representations exist .
but quantum states are defined only up to global change of phases , therefore the symmetry operations of group @xmath0 only needs to be represented by operators @xmath4 which satisfy @xmath5 for any group element @xmath6 and @xmath7 .
such operators form a projective representation of group @xmath0 .
when we consider this general case , we find that the classification result remains the same as with linear representations , that is , 1 . _ for 1d non - translation invariant ( nti ) spin systems with only an on - site symmetry described by a group @xmath0 , all the phases of gapped systems that do not break the symmetry are classified by the equivalence classes in the second cohomology group @xmath1 of the group @xmath0 , provided that the physical states on each site form a projective representation of the group @xmath0 .
_ in certain cases where @xmath0 has infinitely many 1d representations , further classification according to different 1d representations exist .
applying these general results to specific cases allows us to reach the following conclusions : first , result ( a ) means that there is no non - trivial topological order in 1d systems without any symmetry ; using result ( b ) , we find that nti integer - spin chain with only on - site @xmath8 spin rotation symmetry can have two and only two different phases that do not break the @xmath8 symmetry .
result ( c ) implies that nti half - integer - spin chain with only on - site @xmath8 spin rotation symmetry ( which is represented projectively ) also have two and only two gapped phase that does not break the @xmath8 symmetry ; we note the cyclic @xmath9 group has no non - trivial projective representations , thus nti spin chain with only on - site @xmath9 symmetry can have one and only one gapped phase that does not break the @xmath9 symmetry ; the @xmath10 symmetry group has no non - trivial projective representation either , however , due to the special structure of the group of 1d representations of @xmath10 , nti spin chain with only on - site @xmath10 symmetry can have three and only three gapped phase that does not break the @xmath10 symmetry .
we also considered systems with translation invariance ( ti ) and correspondingly many results have been obtained . 1 . _
if there is no other symmetry , all gapped ti systems belong to the same phase . _
( this has been discussed as the generic case in ) .
2 . _ for 1d spin systems with only translation symmetry and an on - site symmetry described by a group @xmath0 , all the phases of gapped systems that do not break the two symmetries are labeled by the equivalence classes in the second cohomology group @xmath1 of the group @xmath0 and different 1d representations @xmath11 of @xmath0 , provided that the physical states on each site form a linear representation of the group @xmath0 .
_ similar to the nti case , we should consider projective representations of @xmath0 at each site .
however , we find that , 1 . _ there is no translation invariant gapped ground state symmetric under on - site symmetry of group @xmath0 which
is represented projectively on the state space at each site .
_ in particular , we can show that the @xmath8 spin rotation symmetric integer spin chain has two different gapped ti phases:@xcite the spin-0 trivial phase and the haldane phase@xcite , while translation invariant @xmath8 symmetric half - integer spin chain must either be gapless or have degeneracy in ground space due to broken discrete symmetries@xcite . on the other hand , the @xmath12 symmetric spin chains where on - site degrees of freedom contain both integer - spin and half - integer - spin representations have only one gapped ti phase .
we also show that the spin chain with only translation and parity symmetry ( defined as exchange of sites together with an on - site @xmath13 operation ) has four different gapped ti phases.@xcite for systems with time reversal symmetry , we find that nti time reversal symmetric systems belong to two phases while ti time reversal symmetric phases on integer spin systems have two phases and those on half - integer spin systems are either gapless or have degeneracy in ground space . the paper is organized as follows : section [ qphase ] gives a detailed definition of gapped quantum phases and explains how that gives rise to an equivalence relation between gapped ground states within the same phase ; section [ 1dgapmps ] shows that short - range correlated matrix product states represents faithfully 1d gapped ground states and hence will be our object of study ; section [ notop ] discusses the situation where no symmetry is required and found no topological order in 1d ; section [ symtop ] gives the classification of phases for 1d systems with certain symmetries , for example on - site symmetry and time reversal symmetry .
it classifies phases in translational invariant systems , and furthermore in systems where translational invariance is present together with other symmetries such as on - site symmetry , parity symmetry and time reversal symmetry ; section [ highd ] generalizes some simple 1d results to higher dimensions ; finally in section [ conc ] we summarize our results and conclude this paper .
to obtain the above stated results , we need to first briefly discuss the definition of quantum phases .
a more detailed discussion can be found in .
quantum phase describes an equivalence relation between quantum systems .
systems we consider live on a @xmath14-dimensional lattice and interactions are local ( with finite range ) . a gapped quantum phase is usually defined as a class of gapped hamiltonians which can smoothly deform into each other without closing gap and hence without any singularity in the local properties of ground state .
such an equivalence relation between hamiltonians can be reinterpreted as an equivalence relation between ground states , as discussed in : _ two gapped ground states belong to the same phase if and only if they are related by a local unitary ( lu ) transformation_. to @xmath15 might not map the corresponding hamiltonian @xmath16 to @xmath17 , but instead to some @xmath18 . however , as @xmath17 and @xmath18 are both gapped and have the same ground states , their equivalence is obvious . ] as lu transformations can change local entanglement structure but not the global one , states in the same phase have the same long range entanglement ( lre ) and hence the same topological order.@xcite states equivalent to product states have only short range entanglement ( sre ) and hence trivial topological order .
all the states with short range entanglement belong to the same phase while states with long range entanglement can belong to the different phases .
these considerations lead to the phase diagram as illustrated in fig .
[ fig : phase](a ) , for the class of systems without any symmetry requirement .
a lu transformation @xmath19 can either take the form of finite time evolution with a local hamiltonian @xmath20 \label{lu_h}\end{aligned}\ ] ] where @xmath21 denotes time ordered integral and @xmath22 is a sum of local hermitian terms . or @xmath19 can take the form of a constant depth quantum circuit @xmath23 which is composed of @xmath24 layers of unitaries and the @xmath25 s within each layer @xmath26 are local and commute with each other.@xcite these two forms of lu transformations are equivalent to each other and we will mainly take the quantum circuit form for the discussion in this paper ( the time evolution form is used for discussion of translation invariant systems ) .
more generally , we will consider equivalence relation between states defined on different hilbert spaces and hence we allow a broader notion of unitarity ( which is called generalized lu transformation in and corresponds to the disentanglers and isometries in mera@xcite ) .
each lu operation @xmath25 we consider in the quantum circuit will act unitarily on the support space of the reduced density matrix of the region with @xmath27 of the original hilbert space and @xmath28 of the support space of reduced density matrix .
while the total hilbert space might change under such an operation , the entanglement structure of the state remains intact and hence the state remains in the same phase with the same topological order .
if the class of systems under consideration have further symmetry constraints , two hamiltonians are in the same phase if they can be connected by a smooth path that stays within this symmetric region .
correspondingly , the equivalence relation between gapped ground states of the same phase needs to be modified : @xcite _ if the class of systems have certain symmetry , two gapped ground states belong to the same phase if and only if they are related by a lu transformation which does not break the symmetry .
_ such an restricted equivalence relation leads to a phase diagram with more structure , as shown in fig .
[ fig : phase](b ) .
first not all short range entangled states belong to the same phase .
short range entangled states with different symmetry breaking will belong to different phases .
those symmetry breaking phases with short range entanglement ( sb - sre ) are well described by landau s symmetry breaking theory @xcite .
can all phases with short range entanglement be described by symmetry breaking ?
landau s symmetry breaking theory suggest that states with the same symmetry always belong to the same phase , which implies that all phases with short range entanglement are described by symmetry breaking . however , this result turns out to be not quite correct .
states which do not break any symmetry can still belong to different phases as well.@xcite we will refer to the order in symmetric short range entanglement ( sy - sre ) states as _ _ symmetry protected topological order__@xcite . for example , in the presence of parity symmetry , the haldane phase and the @xmath29 phase of spin-1 chain belong to two different phases even though both phases have short range entanglement and do not break the parity symmetry . also , in the presence of time - reversal symmetry , the topological insulators and the band insulators belong to two different phases .
again both phases have short range entanglement . for systems with long range entanglement , the phase diagram similarly divides into symmetry breaking ( sb - lre ) and symmetric ( sy - lre ) phases .
the charge @xmath30 superconducting states@xcite and the symmetric @xmath31 states@xciteare examples of the sb - lre phases and the sy - lre phases respectively . without any symmetry restriction .
( b ) the possible phases for class of hamiltonians @xmath32 with some symmetries . the shaded regions in ( a ) and ( b ) represent the phases with short range entanglement . ]
having defined the universality classes of phases as the equivalence classes of states under ( symmetric ) local unitary transformations , we would then like to know which phases exist , or in other words , to classify all possible phases in strongly correlated systems .
some partial classifications have been discussed for strongly correlated systems through string - net states,@xcite and for free fermion systems with certain symmetries through k - theory.@xcite in this paper , we would like to consider 1d gapped strongly correlated spin systems both with and without symmetry , and try to classify all such systems whose ground state does not break any symmetry .
( in other words , the ground state has the same symmetry as the hamiltonian . )
completely classifying strongly correlated spin systems seems to be a hard task as in general strongly interacting quantum many - body systems are very hard to solve .
however , the recent insight about describing 1d gapped ground states of spin systems with matrix product state formalism@xcite provides us with a handle to deal with this problem .
a matrix product states ( mps ) is expressed as @xmath33}_{i_1}a^{[2]}_{i_2} ...
a^{[n]}_{i_n})|i_1i_2 ... i_n\rangle\ ] ] where @xmath34 with @xmath35 being the physical dimension of a spin at each site , @xmath36}_{i_k}$ ] s are @xmath37 matrices on site @xmath26 with @xmath38 being the inner dimension of the mps .
it has been shown that matrix product states capture the essential features of 1d gapped ground state , for example an entanglement area law@xcite and finite correlation length@xcite , and provide an efficient description of such states@xcite .
on the other hand , generic matrix product states satisfying a condition called ` injectivity ' are all gapped ground states of local hamiltonians@xcite .
therefore , studying this class of mps will enable us to give a full classification of 1d gapped spin systems .
now the question of what phases exist in 1d gapped spin systems can be restated as what equivalence classes of matrix product states exist under lu transformations
. gave a specific way to apply such lu transformations , which realizes a renormalization group transformation on mps that removes local entanglement and takes the states to a simple fixed point form .
a partial classification of mps is also given in . in the following
we will use this procedure to classify gapped phases of 1d spin system , in particular the 1d systems with various symmetries .
we see that the possible phases in 1d strongly correlated systems depend on the symmetry of the class of systems .
first we will briefly review how the renormalization group transformation @xcite is done . for the identification and optimal removal of local entanglement ,
a particularly useful mathematical construction is the double tensor @xmath39}_{\alpha\gamma,\beta\chi}=\sum_{i } a^{[k]}_{i,\alpha\beta } \times ( a^{[k]}_{i,\gamma\chi})^*$ ] of the mps .
@xmath39}$ ] uniquely determines the state up to a local change of basis on each site@xcite , that is , if @xmath40}_{\alpha\gamma,\beta\chi}=\sum_{i } a^{[k]}_{i,\alpha\beta } \times ( a^{[k]}_{i,\gamma\chi})^ * = \sum_{i } b^{[k]}_{i,\alpha\beta } \times ( b^{[k]}_{i,\gamma\chi})^*\end{aligned}\ ] ] then @xmath36}_{i,\alpha\beta}$ ] and @xmath41}_{i,\alpha\beta}$ ] are related by a unitary transformation @xmath42}$ ] : @xmath43}_{i,\alpha\beta}= \sum_j u^{[k]}_{ij } a^{[k]}_{j,\alpha\beta } .\end{aligned}\ ] ] therefore , states described by @xmath36}_i$ ] and @xmath41}_i$ ] have exactly the same entanglement structure , which is faithfully captured in @xmath39}$ ] .
a proof of this fact can be found in . for clarity , we present the proof in appendix [ dbt ] following the notation of this paper .
local unitary operations on mps can be applied through manipulation of @xmath39}$ ] . treat @xmath39}$ ] as a @xmath44 matrix with row index @xmath45 and column index @xmath46 . to apply a unitary operation on @xmath14 consecutive sites ,
we combine the double tensor of the @xmath14 sites together into @xmath47 } \mathbb{e}^{[2 ] } ...
\mathbb{e}^{[n]}\end{aligned}\ ] ] and then decompose @xmath48 into a set of matrices @xmath49
s @xmath50 note @xmath51 is determined up to a unitary transformation on @xmath52 .
the index @xmath52 of @xmath53 , up to an unitary transformation , can be viewed as the combination of @xmath54 , the indices of @xmath55}_{i_1,\alpha\beta}$ ] , @xmath56}_{i_2,\alpha\beta}$ ] , ...
@xmath57}_{i_n,\alpha\beta}$ ] .
going from original indices @xmath54 to the effective index @xmath52 corresponds to applying a unitary operation on the @xmath14-block and @xmath49 describes the new state after operation .
the unitary operation can be chosen so that local entanglement is maximally removed .
@xmath48 contains all the information about the entanglement of the block with the rest of the system but not any detail of entanglement structure within the block .
hence we can determine from @xmath48 the optimal way of decomposition into @xmath58 which corresponds to the unitary operation that maximally removes local entanglement while preserving the global structure .
to do so , think of @xmath59 as a matrix with row index @xmath60 and column index @xmath61 .
it is easy to see that with such a recombination , @xmath48 is a positive matrix and can be diagonalized @xmath62 where we have kept only the non - zero eigenvalues @xmath63 and the corresponding eigenvectors @xmath64 .
@xmath58 is then given by @xmath65 which are the matrices representing the new state .
in retaining only the non - zero eigenvalues , we have reduced the physical dimension within the block to only those necessary for describing the entanglement between this block and the rest of the system .
local entanglement within the block has been optimally removed .
each renormalization step in the renormalization procedure hence works by grouping every @xmath14 consecutive sites together and then applying the above transformation to map @xmath55}$ ] , @xmath56}$ ] , ... , @xmath57}$ ] to @xmath66 .
so one renormalization step maps the original matrices @xmath67}_{i_k})^{(0)}$ ] on each site to renormalized matrices @xmath67}_{i_k})^{(1)}$ ] on each block .
repeating this procedure for a finite number of times corresponds to applying a finite depth quantum circuit to the original state .
if the matrices reaches a simple fixed point form @xmath67}_{i_k})^{(\infty)}$ ] ( up to local unitaries ) , we can determine from it the universal properties of the phase which the original state belongs to . such a renormalization procedure hence provides a way to classify matrix product states under lu transformations by studying the fixed point @xmath67}_{i_k})^{(\infty)}$ ] that a state flows to .
two states are within the same phase if and only their corresponding fixed points states can be transformed into each other by ( symmetric ) lu transformations . in the following
we will apply this method to study short - range correlated matrix product states which faithfully represent the class of 1d gapped ground states .
the short - range correlation is an extra constraint on the set of matrix product states that we will consider .
not all matrix product states describe gapped ground states of 1d spin systems .
in particular , 1d gapped ground states all have finite correlation length@xcite for equal time correlators of any local operator , while matrix product states can be long - range correlated .
the finite correlation length puts an extra constraint on mps that the eigen - spectrum of @xmath68 should have a non degenerate largest eigenvalue ( set to be 1 ) ( see appendix [ ndg]).@xcite therefore , we will assume this property of @xmath68 in our following discussion and corresponding mps will be called short - range correlated ( src ) mps .
this renormalization method is well suited for the study of systems without translational invariance , which we will discuss in detail .
we will also make an attempt to study translational invariant systems with this method .
while the full translational symmetry is reduced to block translational symmetry in the rg process , by studying the resulting equivalence classes for different values of block size @xmath14 , we expect to obtain a more complete classification of translational invariant 1d systems .
indeed , the classification result is further confirmed by using a translational invariant lu transformation in the time evolution form to study equivalence between ti systems . in the following sections , we will present our analysis and results for different cases .
first we will consider the situation where no specific symmetry is required for the system .
when no symmetry is required for the class of system , we want to know what kind of long range entanglement exists and hence classify topological orders in 1d gapped spin systems .
we will show that @xmath69 in other words , there is no topological order in 1d .
this is similar to the generic case discussed in . to obtain such a result ,
we use the fact that gapped 1d spin states such that the state is the non - degenerate gapped ground state of @xmath70 . ]
are described by short - range correlated ( src ) matrix product states .
then one can show that all src matrix product states can be mapped to product states with lu transformations and hence there is no topological order in 1d .
consider a generic system without any symmetry ( including translation symmetry ) whose gapped ground state is described as an mps with matrices @xmath36}_{i}$ ] that vary from site to site .
gives a ` canonical form ' for the matrices so that the double tensor @xmath39}_{\alpha\gamma,\beta\chi}$ ] , when treated as a matrix with row index @xmath45 and column index @xmath46 , has a left eigenvector @xmath71}_{\alpha\gamma}=\lambda^{[k]}_{\alpha}\delta_{\alpha\gamma}$ ] and corresponding right eigenvector @xmath72}_{\beta\chi}=\lambda^{[k+1]}_{\beta}\delta_{\beta\chi}$ ] . here
s are positive numbers and @xmath74 .
@xmath75 when @xmath76 and @xmath77 otherwise.}_{i}$ ] .
] this eigenspace has the largest eigenvalue in @xmath39}$]@xcite and is usually set to be @xmath78 .
note that the right eigenvector on site @xmath26 is the same as the left eigenvector on site @xmath79 and has norm @xmath78 , therefore when multiplying the double tensors together , this one dimensional eigenspace will always be of eigenvalue @xmath78
. there could be other eigenvectors of eigenvalue @xmath78 in @xmath39}$ ] .
however , this will lead to an infinite correlation length@xcite and hence not possible in 1d gapped state .
therefore , for short - range correlated mps , @xmath39}$ ] must have a non - degenerate largest eigenvalue @xmath78 .
when multiplying the double tensors together , the remaining block of @xmath39}$ ] will decay exponentially with the number of sites .
this consideration is essential for determining the fixed point of the renormalization procedure when applied to the mps , as shown below .
now we apply the renormalization procedure as discussed in the previous section to remove local entanglement from a general src mps .
take block size @xmath14 .
the double tensor on the renormalized sites are given by @xmath80})^{(1)}=\prod_{k\in k } ( \mathbb{e}^{[k]})^{(0)}$ ] , where @xmath26 s are the @xmath14 sites in block @xmath81 .
( @xmath68 , again , is treated as a @xmath44 matrix with row index @xmath45 and column index @xmath46 . ) after repeating the renormalization process a finite number of times , @xmath82})^{(r)}$ ] will be arbitrarily close to a fixed point form @xmath82})^{(\infty)}$ ] with non - degenerate eigenvalue @xmath78 and @xmath82})^{(\infty)}_{\alpha\gamma,\beta\chi}=\t \lambda^{[k]}_{\alpha\gamma}\t \lambda^{[k+1]}_{\beta\chi}$ ] , where @xmath83}_{\alpha\gamma}=\t \lambda^{[k]}_{\alpha}\delta_{\alpha\gamma}$ ] and @xmath84}_{\beta\chi}=\t \lambda^{[k+1]}_{\beta}\delta_{\beta\chi}$ ] .
now we can decompose @xmath82})^{(\infty)}$ ] into matrices to find the fixed point state .
one set of matrices giving rise to this double tensor is given by @xmath85}_{i^li^r,\alpha\beta})^{(\infty)}=\sqrt{\t \lambda^{[k]}_{i^l } } \delta_{i^l \alpha } \cdot \sqrt{\t \lambda^{[k+1]}_{i^r } } \delta_{i^r \beta}\end{aligned}\ ] ] @xmath86 .
here we use a pair of indices @xmath87 to label the effective physical degrees of freedom on the renormalized site @xmath26 , and @xmath67}_{i^l , i^r})^{(\infty)}$ ] is a set of matrices that defines the fixed - point mps .
it is clear from the form of the matrices that at fixed point every site is composed of two virtual spins of dimension @xmath38 .
every virtual spin is in an entangled pair with another virtual spin on the neighboring site @xmath88}_{i}|i , i\rangle$ ] and the full many - body state is a product of these pairs .
an illustration of this state is given in fig .
[ fig : fp](upper layer ) .
obviously we can further disentangle these pairs by applying one layer of local unitary transformations between every neighboring sites and map the state to a product state ( fig .
[ fig : fp ] , lower layer ) .
therefore , through these steps we have shown that all src matrix product states can be mapped to product states with lu transformations and hence there is no topological order in 1d nti system .
if the class of systems under consideration has certain symmetry , the equivalence classes of states are defined in terms of lu transformations that do not break the symmetry .
therefore , when applying the renormalization procedure , we should carefully keep track of the symmetry and make sure that the resulting state has the same symmetry at each step .
due to such a constrain on local unitary equivalence , we will see that gapped ground states which do not break the symmetry of the system divide into different universality classes corresponding to different symmetry protected topological orders .
we will first discuss the case of on - site symmetries in detail for non - translational invariant ( nti ) systems , i.e. , the system has only an on - site symmetry and no translation symmetry .
then we shall make an attempt to study translational invariant ( ti ) systems , with the possibility of having on - site symmetry or parity symmetry .
lastly , we shall consider the case of time reversal symmetry .
a large class of systems are invariant under on - site symmetry transformations .
for example , the ising model is symmetric under the @xmath13 spin flip transformation and the heisenberg model is symmetric under @xmath8 spin rotation transformations . in this section
, we will consider the general case where the system is symmetric under @xmath89 with @xmath90 being a unitary representation of a symmetry group @xmath0 on each site .
the representation can be linear or projective .
that is , for any @xmath91 , @xmath5 where @xmath92 in a linear representation and @xmath93 could take non - trivial value in a projective representation .
a projective representation of a symmetry group is generally allowed in a quantum description of system because the factor @xmath94 only changes the global phase of a quantum state but not any physically measurable quantity .
therefore , in our classification , we will consider not only the case of linear representation , but also in general projective representations .
the on - site symmetry is the only symmetry required for the class of system . in particular
, we do not require translational symmetry for the systems .
however , for a simple definition of phase , we will assume certain uniformness in the state , which we will define explicitly in the following .
we will classify possible phases for different @xmath0 when the ground state is invariant ( up to a total phase ) under such on - site symmetry operations and is gapped ( short - range correlated ) . specifically , the ground state @xmath95 on @xmath96 sites satisfies @xmath97 and @xmath96 dependent phase factors .
first , let us consider the simpler case where @xmath90 form a linear representation of @xmath0 .
@xmath98 is then a one - dimensional linear representation of @xmath0 .
now we will try to classify these symmetric ground states using symmetric lu transformations and we find that @xmath99 we will also discuss the case of @xmath10 group which has an infinite number of 1d representations .
we will again assume that all gapped states can be represented as short range correlated matrix product states .
we will use the renormalization flow used before@xcite to simplify the matrix product states and use the fixed - point matrix product states to characterize different equivalent classes of lu transformations , as two symmetric states belong to the same class if and only if their corresponding fixed - point states can be mapped to each other with symmetric lu transformations . in order to compare different equivalent classes under _ symmetric _ lu transformations ,
it is important to keep track of the symmetry while doing renormalization .
first , in the renormalization procedure we group @xmath14 sites together into a new site .
the on - site symmetry transformation becomes @xmath100 , which is again a linear representation of @xmath0 .
the next step in rg transformation applies a unitary transformation @xmath101}_1 $ ] to the support space of new site @xmath26 .
this is actually itself composed of two steps .
first we project onto the support space of the new site , which is the combination of @xmath14 sites in the original chain .
this is an allowed operation compatible with symmetry @xmath0 as the reduced density matrix @xmath102 is invariant under @xmath103 , so the support space form a linear representation for @xmath0 .
the projection of @xmath103 onto the support space @xmath104 hence remains a linear representation of @xmath0 . in the next step
, we do some unitary transformation @xmath101}_1 $ ] within this support space which relabels different states in the space .
the symmetry property of the state should not change under this relabeling . in order to keep track of the symmetry of the state
, the symmetry operation needs to be redefined as @xmath105})^{(1)}(g)=w^{[k]}_1p_n\tilde{u}^{(0)}(g)p_n(w^{[k]}_1)^{\dagger}$ ] . after this redefinition , the symmetry operations @xmath105})^{(1)}(g)$ ] on each new site
@xmath26 form a new linear representation of @xmath0 . by redefining @xmath105})^{(i)}(g)$ ] at each step of rg transformation
, we keep track of the symmetry of the system . finally at the fixed point
( at a large rg step @xmath106 ) , we obtain a state described by @xmath67}_{i^li^r})^{(r)}$ ] which is again given by the fixed point form . to describe a state that does not break the on - site symmetry , here @xmath67}_{i^li^r})^{(r)}$ ] is invariant ( up to a phase ) under @xmath105})^{(r)}(g)$ ] on each site @xmath26 .
therefore@xcite , @xmath107}_{i^li^r , j^lj^r}(g ) a^{[k]}_{j^lj^r } & = \al_{[k]}^{(r)}(g ) n_{[k]}^{-1}(g)a^{[k]}_{i^li^r } m_{[k]}(g ) \nonumber\\ n_{[k]}(g ) & = m_{[k-1]}(g)\end{aligned}\ ] ] must be satisfied with some invertible matrix @xmath108}(g)$ ] and @xmath109}(g)$ ] . here
@xmath26 labels the coarse grained sites and we have dropped the rg step label @xmath24 ( except in @xmath110}^{(r)}(g)$ ] ) .
each coarse grained site is a combination of @xmath111 original lattice sites and @xmath110}^{(r)}(g)$ ] form a 1d ( linear ) representation of @xmath0 . solving this equation we find the following results ( see appendix [ soluu ] ) : + ( a ) @xmath108}(g)$ ] and @xmath109}(g)$ ] are projective representations of @xmath0 ( see ) .
projective representations of @xmath0 belong to different classes which form the second cohomology group @xmath112 of @xmath0 .
( for a brief introduction on projective representation , see appendix [ prorep ] ) .
@xmath109}(g)$ ] and @xmath108}(g)$ ] corresponds to the same element @xmath113 in @xmath112 .
+ ( b ) the linear symmetry operation @xmath114}(g ) $ ] must be of the form @xmath110}^{(r)}(g)u^{[k],l}(g)\otimes u^{[k],r}(g)$ ] where @xmath114,l}$ ] and @xmath114,r}$ ] are projective representations of @xmath0 and correspond to inverse elements @xmath113 and @xmath115 in @xmath116 respectively .
@xmath110}^{(r)}(g)$ ] is a 1d ( linear ) representation of @xmath0 .
@xmath114,l}$ ] and @xmath114,r}$ ] act on the two virtual spins separately ( see ) .
therefore , the fixed point state is formed by entangled pairs @xmath117 of virtual spins which are invariant , up to a phase ( due to the non - trivial @xmath110}^{(r)}(g)$ ] ) , under linear transformation @xmath114,r}(g)\otimes u^{[k+1],l}(g)$ ] .
now we use the uniformness of the state and simplify our discussions .
specifically , we assume that @xmath110}^{(r)}(g)$ ] does not depend on the site index @xmath26 .
certainly , @xmath110}^{(r)}(g)$ ] does not depend on @xmath26 if the state has the translation symmetry .
if the 1d representations of @xmath0 are discrete , then for weak randomness that slightly break the translation symmetry , @xmath110}^{(r)}(g)$ ] still does not depend on @xmath26 .
so we can drop the @xmath26 index and consider @xmath118 .
does different @xmath118 label different symmetric phases ?
first , the answer is no if the number of 1d representations of @xmath0 is finite ( as is the case for @xmath119 , @xmath8 , etc ) . because we can always choose block size @xmath14 properly so that @xmath120 and the difference between symmetric states due to @xmath121 disappears . in the case of @xmath10 group , there are infinitely many different 1d representations @xmath122 , labeled by integer @xmath123 . for two states with positive @xmath124 , @xmath125 , we can always choose block size @xmath126 , @xmath127 respectively , so that the 1d representations become the same .
this is also true for negative @xmath124 , @xmath125 .
but if @xmath124 , @xmath125 take different sign ( or one of them is @xmath128 ) , the 1d representations will always be different no matter what blocking scheme we use .
therefore , @xmath10 symmetric states divide into three classes due to different 1d representations that can be labeled by @xmath129 . after these considerations
, we will ignore the 1d representations @xmath118 in the following discussion .
we find that the entangled pairs @xmath117 of virtual spins in the fixed point state are exactly invariant under linear transformation @xmath114,r}(g)\otimes u^{[k+1],l}(g)$ ] .
the left virtual spin of each site forms a projective representation of @xmath0 corresponding to element @xmath113 in @xmath116 , while the right virtual spin corresponds to element @xmath115 . in appendix
[ equiv ] , we will show that fixed point states with the same @xmath113 can be related by a symmetric lu transformation , while those with different @xmath113 can not .
therefore , the phases of src mps that are invariant under linear on - site symmetry of group @xmath0 are classified by the second cohomology group @xmath116 .
( when @xmath3 , further division of classes due to different 1d representations of @xmath0 exist .
the equivalence classes are labeled by @xmath130 and @xmath131 . ) due to the basic assumption of quantum mechanics that the global phase of a quantum state will not have any effect on the physical properties of the system , it is necessary to consider not only the linear representation of symmetry operations on the system , but also the projective representations .
for example , on a half - integer spin , rotation by @xmath132 is represented as @xmath133 , minus the identity operator instead of @xmath134 .
hence , the rotation symmetry @xmath8 is represented projectively on half integer spins . in order to cover situations like this , we discuss in this section systems with on - site projective symmetry of group @xmath0 .
again , we consider the case when the ground state does not break the symmetry , i.e. latexmath:[$u^{(0)}(g ) \otimes ... \otimes u^{(0)}(g ) |\phi_l\rangle= \alpha_l(g ) representation of group @xmath0 corresponding to class @xmath113 .
assuming uniformness of the state , we require that @xmath113 does not vary from site to site . but this can be reduced to the previous linear case .
as long as @xmath116 is finite and @xmath113 has a finite order @xmath14 , we can take block size @xmath14 so that after blocking , the symmetry operation on the renormalized site @xmath136 corresponds to @xmath137 in @xmath116 .
therefore , the state after one blocking step is symmetric under an on - site linear representation of group @xmath0 and all the reasoning in the previous section applies .
we find that the classification with projective on - site symmetry is the same as linear on - site symmetry .
that is @xmath138 the @xmath10 group does not have a non - trivial projective representation and will not introduce any complication here .
since @xmath139 has no non - trivial projective representations , we find that @xmath140 for spin systems with only spin rotation symmetry , @xmath141 .
@xmath8 has two types of projective representations described by @xmath142 , corresponding to integer and half - integer spin representations .
we find that @xmath143 such a result has some relation to a well known result@xcite for nti spin-1 heisenberg chain @xmath144 the model undergoes an impurity driven second order phase transition from the haldane phase@xcite to the random singlet phase@xcite as the randomness in @xmath145 increases . for half - integer - spin systems
, @xmath8 is represented projectively on each site , yet the classification is the same as the integer case .
we find that @xmath146 representative states of the two phases are nearest - neighbor dimer states , but with dimer between sites @xmath147 and @xmath148 in the first phase and between sites @xmath149 and @xmath147 in the second phase .
the projective representation of @xmath8 on half - integer - spins form a linear representation of @xmath12 .
if we think of the linear representation of @xmath8 on integer - spins as a ( unfaithful ) linear representation of @xmath12 and allow the mixture of integer and half - integer spins on one site , then the two phases of @xmath8 merge into one . therefore , systems with only on - site @xmath12 symmetry ( which implies the mixture of integer and half - integer spins on each site ) belong to one phase as we can map integer - spin singlets into half - integer - spin singlets without breaking the @xmath12 symmetry ( see appendix [ equiv ] )
. such a procedure breaks down if @xmath8 symmetry is required for each site as the direct sum of a linear representation ( on integer - spin ) and a projective representation ( on half - integer - spin ) is no long a projective representation for @xmath8 . in this way
, we have obtained a full classification of the phases of gapped nti 1d spin systems with various on - site symmetry .
we have discussed the gapped phases of 1d nti systems that have some on - site symmetries . in this section
, we would like to discuss translation symmetric systems .
we will consider those translation symmetric systems whose ground states are gapped and translation invariant ( ti ) .
to discuss the ti phases , we need to discuss the equivalence classes under ti lu transformations . however , it is hard to use quantum circuit to describe ti lu transformations .
thus , the quantum circuit formulation used in this paper is inconvenient to describe ti lu transformations .
however , in this section , we will first try to use the quantum circuit formulation of lu transformations to discuss the phases of ti gapped states .
while we are able to identify many different phases in this way , we can not rigorously prove the equivalence of states within each phase using the non - translational invariant circuit . in order to establish this equivalence
, we later apply the other formulation of local unitary transformation a finite time evolution with a local hamiltonian(eqn.[lu_h ] ) where ti can be preserved exactly and we confirm the classification result obtained with quantum circuits(see appendix [ fullti ] ) . -block ti lu transformation described by a quantum circuit of three layers . here
@xmath151 , @xmath152 , @xmath153 .
the unitary transformations @xmath154 on different blocks in the @xmath155 layer are the same . ]
let us consider the lu transformations represented by quantum circuits which is formed by the unitary operators on blocks of @xmath156 sites in the @xmath155 layer ( see fig .
. we will call such lu transformations @xmath150-block lu transformations .
if the unitary operators on different blocks in the same layer of the quantum circuit are the same , we will call the lu transformations a @xmath150-block ti lu transformations . in this paper
, we will try to use quantum circuit formulation of the lu transformations to discuss gapped translation symmetric phases .
one way to do so is to use the equivalent classes of the @xmath150-block ti lu transformations to classify the gapped translation symmetric phases .
however , since the @xmath150-block ti lu transformations are different from the ti lu transformations , the equivalent classes of @xmath150-block ti lu transformations are different from the equivalent classes of the ti lu transformations .
but since the ti lu transformations are special cases of @xmath150-block ti lu transformations , each equivalent class of @xmath150-block ti lu transformations are formed by one or more equivalent classes of ti lu transformations .
therefore , we can use the equivalent classes of @xmath150-block ti lu transformations to describe the gapped ti phases , since different equivalent classes of @xmath150-block ti lu transformations always represent different ti quantum phases . on the other hand ,
the equivalent classes of @xmath150-block ti lu transformations may not separate all gapped ti phases .
sometimes , a single equivalent class of @xmath150-block ti lu transformations may contain several different gapped ti phases . to increase the resolution of the @xmath150-block ti lu transformations
, we would like to introduce the block equivalent classes : two states belong to the same block equivalent classes , iff , _ for all values of @xmath156 _ , they can be mapped into each other through @xmath150-block ti lu transformations .
clearly , each block equivalent class might still contain several different gapped ti phases . in this paper
, we will call a block equivalent class a block phase , or b - phase .
it is possible that the block equivalent classes are the same as the universal classes that represent the gapped ti phases . in this case
the gapped ti b - phases are the same as the gapped ti phases , and we can study the gapped ti phases through block equivalent classes . in the following , we will first study the b - phases for 1d strongly correlated systems with translation and some other symmetries and then confirm that gapped ti b - phases do coincide with gapped ti phases . to describe ti gapped states
we will use ti mps representation with site independent matrices .
but how do we know a ti gapped state can always be represented by a mps with site independent matrices ?
a non - uniform mps in can represent a ti state if the matrices @xmath36}_i$ ] satisfy , for example , @xmath157}_i= m^{-1 } a^{[k]}_i m\end{aligned}\ ] ] for an invertible matrix @xmath158 .
it is proven in that every ti state does have a ti mps representation .
specifically , in the example considered above , we can transform the matrices so that they become site independent .
let us introduce @xmath159}_i= n_{[k-1]}^{-1 } a^{[k]}_i n_{[k]}.\end{aligned}\ ] ] by construction , @xmath36}_i$ ] and @xmath160}_i$ ] will represent the same mps .
we see that @xmath161}_i & = n_{[k]}^{-1 } a^{[k+1]}_i n_{[k+1 ] } \nonumber\\ & = n_{[k]}^{-1}m^{-1 } a^{[k]}_i m n_{[k+1 ] } \\ & = n_{[k]}^{-1}m^{-1 } n_{[k-1 ] } \t a^{[k]}_i n_{[k]}^{-1 } m n_{[k+1 ] } \nonumber\end{aligned}\ ] ] we see that if we choose @xmath162}=m^{-k},\end{aligned}\ ] ] we will have @xmath161}_i=\t a^{[k]}_i.\end{aligned}\ ] ] hence reducing the original non - uniform representation to a uniform one .
we also like to remark that by `` ti gapped state '' , we mean that there exists a ti gapped hamiltonian @xmath163 on a lattice of @xmath96 sites such that the ti gapped state is the ground state of @xmath163 .
we would like to stress that we do not need the above condition to be true for all values of @xmath96 .
we only require @xmath163 to be gapped for a sequence of lattice sizes @xmath164 that @xmath165 . in this paper ,
when we discuss a system of size @xmath96 , we always assume that @xmath96 belongs to such a sequence @xmath164 .
this consideration is important if we want to include in our discussion , for example , boson systems with 1/2 particles per site which can only be defined for even system size @xmath96 . in the above discussion
, we see that for a large block size , the related matrix [ see ] @xmath166 } \mathbb{e}^{[2 ] } ... \mathbb{e}^{[n]}$ ] is dominated by its largest eigenvalue . for translation invariant states , @xmath167}=\mathbb{e}$ ] and @xmath168 .
so the fixed - point @xmath48 , and hence the gapped ti b - phase , is directly determined by the largest eigenvalue and the corresponding left and right eigenvectors of @xmath68 .
since a gapped ti b - phase is directly determined by @xmath68 on each site , we do not need to do any blocking transformation to understand the gapped ti b - phase .
we can directly extract its fixed - point tensor from @xmath68 .
this suggests that gapped ti b - phases coincide with gapped ti phases .
indeed , we can directly deform @xmath68 to the fixed - point @xmath48 by changing other non - largest eigenvalues to zero .
such a procedure allows us to deform the matrix @xmath169 of the initial mps to the fixed - point matrix @xmath170 of the final mps .
since the largest eigenvalue of @xmath68 is not degenerate , the state remains short range correlated ( and gapped ) during the deformation.@xcite ( as we are writing up the above arguments , we learned that a similar method is used by schuch et .
al.@xcite ) also the state does not break the translation symmetry ( unlike the @xmath14-block transformations ) and other symmetries during the deformation .
this allows us to show that the gapped ti phases is characterized by the largest eigenvalue and the corresponding left and right eigenvectors of @xmath68 .
thus gapped ti b - phases coincide with gapped ti phases . for details , see appendix [ fullti ] . in the following
, we will use the @xmath150-block ti lu transformations to discuss various gapped ti phases of 1d systems .
@xmath171 such a result generalizes the earlier result that for 1d systems with no symmetry , there is only one gapped phase . to obtain the new result ,
we basically repeat what we did in section [ notop ] .
the only difference is that the matrices representing the state now are site independent and in section [ notop ] we use @xmath150-block lu transformations to reduce the 1d nti mps , while to derive the new result , here we use @xmath150-block ti lu transformations to reduce the 1d ti mps .
similarly , by repeating the discussions in section [ usymm_l ] and section [ usymm_p ] for @xmath150-block ti lu transformations on the 1d ti mps , we can show that @xmath172 the reason is as follows .
if a 1d state with translation symmetry is short - range correlated , it can be represented by a ti mps .
its fixed - point mps also has an on - site projective unitary symmetry @xmath173 . for a proper choice of block size @xmath14
, we can make @xmath4 and @xmath173 to be the same type of projective representation described by @xmath174 . for ti fixed - point mps , we have @xmath175}=\om_{[k-1]}$ ] since @xmath109}(g)=m_{[k-1]}(g)$](cf .
appendix [ soluu ] ) .
thus @xmath176 , that is , the trivial element in @xmath116 .
so , if @xmath177 , the 1d ti state can not be short - range correlated .
in other words @xmath178 if the ground state of the 1d spin system does not break the on - site symmetry and the translation symmetry , then ground state is not short - range correlated and is gapless .
if the ground state of the 1d spin system breaks the on - site symmetry or the translation symmetry , then the ground state is degenerate . as an application of the above result
, we find that @xmath179 which agrees with the well known result of . to have a gapped ti 1d state with an on - site symmetry
, the symmetry must act linearly ( not projectively ) . in this case
, we can show that the total phase factor of the state @xmath180 breaks up into @xmath96 1d representations @xmath181 ( see appendix [ all ] ) @xmath182^l \end{align } for all values of $ l$ that is large enough . } } \nonumber\ ] ] here @xmath90 is the linear representation of @xmath0 acting on the physical states in each site and @xmath181 is an one - dimensional linear representation of @xmath0 .
let us apply the above result to a boson system with @xmath183 bosons per site . here
the bosons number is conserved and there is an @xmath10 symmetry .
certainly , the system is well defined only when the number of sites @xmath96 has a form @xmath184 ( assuming @xmath185 and @xmath186 have no common factors ) . for such @xmath96 ,
we find that @xmath187 , where @xmath188 is the generating 1d representation of the @xmath10 symmetry group .
so is _ not _
satisfied for some large @xmath96 .
therefore , @xmath189 in higher dimensions , the situation is very different .
a 2d state of conserved bosons with fractional bosons per site can be gapped , and , at same time , does not break the @xmath10 and the translation symmetry .
2d fractional quantum hall states of bosons on lattice provide examples for such kind of states .
also , by repeating the discussions in section [ usymm_l ] for @xmath150-block ti lu transformations on the 1d ti mps , we can show that @xmath190 here @xmath191 is an 1d representation of @xmath0 that appear in .
the symmetric lu transformations can not change 1d representation @xmath181 .
so the different phases are also distinguished by the 1d representations @xmath192 of @xmath0 .
here are a few concrete examples : if we choose the symmetry group to be @xmath139 , we find @xmath193 this is because @xmath9 has no projective representations and has @xmath14 different 1d representations . as an example , consider the following model @xmath194 , \end{aligned}\ ] ] where @xmath195 are the pauli matrices .
the model has a @xmath13 symmetry generated by @xmath196 .
the two different @xmath13 symmetric phases correspond the @xmath197 phase and the @xmath198 phase of the model . if we choose the symmetry group to be @xmath141 , we find @xmath199 this is because @xmath8 has only one 1d representation and @xmath200 .
such a result agrees with the well known result that the aklt state@xcite of spin-1 chain and the direct product state with spin-0 on each site represent two different @xmath8 symmetric ti phases .
the aklt state ( and the related haldane phase@xcite ) has gapless boundary spin-1/2 states@xcite and non - trivial string orders,@xcite which indicate that the aklt state is really different from the spin-0 product state .
actually , the full symmetry of @xmath8 can be relaxed to only the dihedral group @xmath201(@xmath202 ) of rotation by @xmath203 around @xmath204 , @xmath205 and @xmath206 axis . as explained in appendix [ prorep ]
, @xmath201 has one non - trivial projective representation , to which the aklt state corresponds .
aklt is different from the spin-0 product state as long as on - site @xmath201 symmetry is preserved .
this is consistent with the result in . in this section
, we will consider the case of parity symmetry for translational invariant system .
we define the parity operation @xmath207 for a spin chain to be in general composed of two parts : @xmath208 , exchange of sites @xmath14 and @xmath209 ; @xmath210 , on - site unitary operation @xmath211 where @xmath212 .
operation @xmath211 is necessary in the definition of parity if we want to consider for example , fixed point state with @xmath213 be to parity symmetric .
the state is not invariant after exchange of sites , and only maps back to itself if in addition the two virtual spins on each site are also exchanged . ]
similar to the previous discussion , @xmath207 gets redefined as we renormalize the state until at fixed point @xmath208 becomes the exchange of renormalized sites and @xmath210 becomes @xmath214 on every site , @xmath215 .
the fixed point matrices hence satisfies(the @xmath216 label is dropped ) : @xmath217 for some invertible matrix @xmath158 , where we have used that the 1d representation of parity is either @xmath218 or @xmath219 .
we label the two 1d representations with @xmath220 . here
@xmath158 satisfies @xmath221 . but
@xmath222 , therefore , @xmath223 and correspondingly @xmath158 is either symmetric @xmath224 or antisymmetric @xmath225 .
we will label this sign factor as @xmath226 . solving this equation
gives that @xmath227 , where @xmath228 is the exchange operation of two virtual spins @xmath229 and @xmath230 and @xmath231,@xmath232 act on @xmath229,@xmath230 respectively .
@xmath233 and @xmath234 .
it can then be shown that each entangled pair @xmath117 must be symmetric under parity operations and satisfies @xmath235 .
there are hence four different symmetric phases corresponding to @xmath220 and @xmath226 . by enlarging the local hilbert space
, we can show similarly as before that fixed points within each class can be mapped from one to the other with the @xmath150-block ti lu transformation preserving the parity symmetry . on the other hand , fixed points in different classes can not be connected without breaking the symmetries .
therefore , under the @xmath150-block ti lu transformation , there are four block classes with parity symmetry and hence four parity symmetric ti phases : @xmath236 as an example , consider the following model @xmath237 , \end{aligned}\ ] ] where @xmath238 are the spin-1 operators .
the model has a parity symmetry .
the @xmath239 phase and the @xmath240 phase of the model correspond to two of the four phases discussed above .
the @xmath241 state@xcite is in the same phase as the aklt state . in the fixed - point state for such a phase , @xmath242 state
is the @xmath243 state .
its entangled pairs are @xmath244 which do not change sign under the parity transformation .
thus the stability of the haldane / aklt state is also protected by the parity symmetry.@xcite , @xmath245 ( b ) @xmath246 , @xmath245 ( c ) @xmath246 , @xmath247 ( d ) @xmath248 , @xmath247 .
@xmath249 stands for a parity even entangled pair ( e.g. @xmath250 ) , @xmath251 stands for a parity odd entangled pair ( e.g. @xmath252 ) .
each site contains four virtual spins.,width=288 ] to understand why there are four parity symmetric phases instead of two ( parity even / parity odd ) , we give four representative states in fig .
[ fig:4pstates ] , one for each phase .
connected pair of black dots denotes an entangled pair .
@xmath249 stands for a parity even pair , for example @xmath250 , and @xmath251 stands for a parity odd pair , for example @xmath252 .
each rectangle corresponds to one site , with four virtual spins on each site .
the four states are all translational invariant . if the parity operation is defined to be exchange of sites together with exchange of virtual spins @xmath78 and @xmath253 , @xmath254 and @xmath255 on each site , then states ( a ) and ( d ) are parity even while ( b ) and ( c ) are parity odd . but ( a ) and ( d ) ( or ( b ) and ( c ) ) are different parity even ( odd ) states and can not be mapped to each other through local unitary transformations without breaking parity symmetry . written in the matrix product representation , the matrices of the four states will transform with @xmath220 and @xmath226 respectively .
therefore , the parity even / odd phase breaks into two smaller phases and there are in all four phases for parity symmetric systems . time reversal , unlike other symmetries , is represented by antiunitary operator @xmath256 , which is equivalent to the complex conjugate operator @xmath81 followed by a unitary operator @xmath19 .
@xmath256 has two projective representations : one on integer spins with @xmath257 and the other on half - integer spins with @xmath258 .
the classification of gapped 1d time reversal invariant phases follows closely the cases discusses before . in this section
, we will highlight the differences and give our conclusion .
first , a state @xmath259 is called time reversal invariant if @xmath260 , where @xmath261 .
but for anti - unitary @xmath256 , the global phase @xmath262 is arbitrary and in particular we can redefine @xmath263 , such that @xmath264 .
therefore , in the following discussion , we will assume wlog that @xmath265 .
now let us consider system without translational invariance .
@xmath257 or @xmath133 does not make a difference here as we can take block size @xmath254 so that on the renormalized site , @xmath266 is always equal to @xmath134 . using argument similar to the case of on - site unitary symmetry , we can keep track and redefine symmetry operations as we do renormalization .
finally , at the fixed point we have a state described by matrices @xmath67}_{i^li^r})^{(\infty)}$ ] which is invariant under time reversal operation @xmath267})^{(\infty)}=(u^{[k]})^{(\infty)}k$ ] , that is , @xmath268}_{i^li^r , j^lj^r } ( a^{[k]}_{j^lj^r})^*= n_{[k]}^{-1}a^{[k]}_{i^li^r}m_{[k ] } \nonumber\\ & n_{[k]}= m_{[k-1]}\end{aligned}\ ] ] where the fixed - point label @xmath216 has been omitted . solving this equation
we find , + ( a)@xmath109}m^*_{[k]}=e^{i\theta}i$ ] . as @xmath109}$ ] is invertible , @xmath223 .
+ ( b)@xmath114 } = u^{[k],l } \otimes u^{[k],r}$ ] . where @xmath114,l}(u^{[k],l})^*=\pm i$ ] and @xmath114,r}(u^{[k],r})^*=\pm i$ ] .
therefore , each entangled pair is time reversal invariant @xmath269,r}\otimes u^{[k+1],l})k |ep_{k , k+1}\rangle = |ep_{k , k+1}\rangle\ ] ] similar to previous sections , we can show that @xmath270 and @xmath271 correspond to two equivalence classes and two time reversal invariant fixed point states can be mapped into each other if and only if they belong to the same class . therefore , our classification result for time reversal symmetry is @xmath272 if the system has additional translation symmetry , we can similarly classify the ti phases and find that @xmath273 1d integer - spin systems are examples of this case .
the haldane / aklt state and the spin @xmath128 product state are representatives of the two phases respectively@xcite .
we also have @xmath274 1d half - integer - spin systems are examples of such case .
in the last few sections , we classified symmetry protected topological orders in one dimension , using ( symmetric ) lu transformations .
can we use ( symmetric ) lu transformations to classify ( symmetry protected ) topological orders in higher dimensions ? in higher dimensions ,
the situation is much more complicated .
first , infinitely many kinds of non - trivial topological orders exist for class of systems without any symmetries.@xcite a partial classification is given in for such a case in 2d . in the presence of symmetry ,
the phase diagram is even more complicated . studied 2d topological orders protected by time reversal and point group symmetry .
so far , we do not have a detailed understanding of topological phases in the presence of symmetry .
however , using similar arguments as that used for 1d systems , we can obtain some simple partial results for higher dimensions .
for example , we have @xmath275 label four distinct states in integer spin systems with translation and spin rotation symmetries : ( a ) @xmath276 , ( b ) @xmath277 , ( c ) @xmath278 , ( d ) @xmath279 .
the open dots represent 0-spins .
the solid dots represent 1/2-spins .
two solid dots connected by a line represent a spin - singlet formed by two 1/2-spins .
the state in ( a ) has an spin-0 spin on each site .
the state in ( b , c ) has two spin-1/2 spins on each site ( or 4 states per site that form spin 0 + 1 representation of @xmath8 ) .
the state in ( d ) has four spin-1/2 spins on each site ( or 16 states per site that form spin 0 + 0 + 1 + 1 + 1 + 2 representation of @xmath8 ) . ] [ cols="^,^,^",options="header " , ] let us illustrate the above result in 2d by considering a tensor network state ( tns ) on a square lattice , where the physical states living on each vertex @xmath280 are labeled by @xmath281 .
a translation invariant tns is defined by the following expression for the many - body wave function @xmath282 : @xmath283 here @xmath284 is a complex tensor with one physical index @xmath281 and four inner indices @xmath285 .
the physical index runs over the number of physical states @xmath35 on each vertex , and the inner indices runs over @xmath38 values .
the tns can be represented graphically as in fig .
[ tps ] . if the tensor @xmath286 satisfy @xmath287 for some invertible matrices @xmath288 and @xmath289 , then the many - body wave function @xmath290 is symmetric under the on - site symmetry transformation @xmath291 in the on - site symmetry group @xmath0 . here
@xmath181 is a one dimensional representation of @xmath0 , the @xmath38 by @xmath38 matrices @xmath288 form a projective representation represented by @xmath292 , and the @xmath38 by @xmath38 matrices @xmath289 form a projective representation represented by @xmath293 . since symmetric lu transformation can not change @xmath294 , thus @xmath294 label distinct quantum phases .
in fact @xmath294 are all measurable , so they indeed label distinct quantum phases . on a torus of size @xmath295 , the symmetric many - body wave function @xmath290 transforms as the 1d representation @xmath296 under the on - site symmetry transformation @xmath291 .
if @xmath0 has only a finite number of 1d representations , we can always choose @xmath297 and @xmath298 such that @xmath299 . on a cylinder of size @xmath295 with open boundary in the @xmath204-direction , the states on one boundary will form a projective representation which is represented by @xmath300 . similarly ,
if the open boundary in the @xmath205-direction , the states on one boundary will form a projective representation which is represented by @xmath301 . if we choose @xmath297 and @xmath298 properly ( for example to make @xmath302 and @xmath303 ) , we can detect both @xmath304 and @xmath305 .
a system with integer on - site spins gives us an example with @xmath141 , if the system has translation and spin rotation symmetry .
for @xmath141 there is no non - trivial 1d representations .
so we can drop the consideration of @xmath192 .
also @xmath141 has two types of projective representations : @xmath142 , where @xmath306 is the trivial projective representation which correspond to linear representations of @xmath8 ( on integer spins ) and @xmath307 is the non - trivial projective representation which corresponds to half - integer spins .
thus @xmath308 label four distinct states in 2d ( see fig .
[ 4states ] ) .
using the ( symmetric ) local unitary equivalence relation between gapped ground states in the same phase and the matrix product representation of 1d states , we classify possible quantum phases for strongly interacting 1d spin system with certain symmetry when ground state of the system does not break the symmetry .
our result is summarized in table [ table : result ] .
many well known results are rederived using a quite different approach , for example the existence of haldane phase@xcite and the gaplessness of spin @xmath309 heisenberg model@xcite .
those results are also greatly generalized to new situations , for example to the case of time reversal symmetry and @xmath201 symmetry .
we find that the projective representations play a very important role in understanding and formulating those generalized results . in higher dimensions ,
things are more complicated .
nevertheless , similar considerations allow us to obtain some interesting examples of symmetry protected topological orders . a complete classification of higher dimension phases , however , requires at least a full understanding of topological orders , an element that is absent in the 1d phase diagram
. results similar to ours on the classification of integer / half - integer 1d spin chains with so(3 ) symmetry have been obtained independently by alexei kitaev recently.@xcite we would like to thank a. kitaev , j. i. cirac , f. verstraete , g. vidal , g .-
zhang , a. w.w .
ludwig and h. katsura for some very helpful discussions .
xgw is supported by nsf grant no .
zcg is supported in part by the nsf grant no .
nsfphy05 - 51164 .
in this section we present the proof for the theorem that : _ two sets of matrices @xmath310 and @xmath311 , where @xmath312,@xmath313 label different matrices and @xmath60 are the row and column indices of the matrices , give rise to the same double tensor @xmath314 , if and only if they are related by a unitary transformation @xmath315 .
_ the dimension @xmath158 of @xmath312 and the dimension @xmath316 of @xmath313 can be different .
the @xmath317 matrix @xmath19 is in general called unitary if @xmath318 , the identity matrix in @xmath96 dimensional space where @xmath319 and @xmath320 . the complete proof of this theorem can be found in , where this property is discussed in terms of ` the unitary degree of freedom in the operator sum representation of quantum channels ' .
here we re - present this proof following the notation and terminology of the current paper for simplicity of understanding .
first we prove the ` if ' part of the theorem .
suppose that @xmath321 , then @xmath322 therefore the two double tensors are the same .
this proves the first part of the theorem . on the other hand ,
suppose that the two double tensors are the same @xmath323 .
reorder the indices of @xmath68 and treat it as a matrix with row indices @xmath60 and column indices @xmath61 .
we will denote the double tensor after this reordering as @xmath324 .
it is easy to see that @xmath325 is a positive semi - definite matrix , as @xmath326 for any vector @xmath327 diagonalize @xmath325 into @xmath328 with @xmath329 .
define vectors @xmath330 , so that @xmath331 . @xmath332
form a complete orthogonal set and hence we can expand @xmath310 and @xmath333 in terms of them .
@xmath334 then @xmath335 but we know that @xmath331 . therefore , @xmath336 and @xmath207 is a unitary matrix . similarly we can show that @xmath337 is a unitary matrix .
therefore @xmath310 and @xmath311 is related by a unitary transformation @xmath338 where @xmath339 .
we have thus proved both part of the theorem .
in section [ 1dgapmps ] , we cited the property of matrix product state that the finite correlation length of the state is closely related to the non - degeneracy of the largest eigenvalue of the double tensor , as discussed in ref . .
in this section , we give a brief illustration of why this is so .
for simplicity of notation , we focus on the translational invariant case first .
generalization to matrix product states without translational invariance is straightforward and similar conclusions can be reached . for a matrix product state @xmath340 described by matrices @xmath310 with double tensor @xmath341 , define @xmath342_{\alpha\gamma,\beta\chi}=\sum_{ij } o_{ij } a_{i,\alpha\beta } \times a^*_{j,\gamma\chi}$ ] for arbitrary operator @xmath343 .
follow the previous convention and treat @xmath68 and @xmath342 $ ] as matrices with row index @xmath45 and column index @xmath46 .
the norm of the wavefunction is given by @xmath344 , where @xmath316 is total length of the chain .
wlog , we can set the largest eigenvalue of @xmath68 to be @xmath78 and hence the norm goes to a finite value ( dimension of the eigenspace ) as @xmath316 goes to infinity .
the expectation value of any local operator @xmath345 is @xmath346)/\tr ( \mathbb{e}^n)$ ] and the correlation between two operators @xmath347 and @xmath348 becomes @xmath349\mathbb{e}^{l}\mathbb{e}[o_2])/\tr ( \mathbb{e}^n)- \\ \tr(\mathbb{e}^{n-1}\mathbb{e}[o_1])\tr(\mathbb{e}^{n-1}\mathbb{e}[o_2])/\tr^2 ( \mathbb{e}^n ) \end{array}\ ] ] the physical constraints on the expectation value and correlation functions of local operators require that the double tensor @xmath68 has certain properties .
first , put @xmath68 into its jordan normal form and decompose it as @xmath350 , where @xmath351 is the diagonal part and @xmath352 the nilpotent part .
but for the largest eigenvalue @xmath78 , @xmath353 must be @xmath128 as otherwise for large system size @xmath316 @xmath346)/\tr ( \mathbb{e}^n ) \sim \tr((p_1+r_1)^n \mathbb{e}[o])$ ] will be unbounded for any @xmath342 $ ] that satisfies @xmath354)\neq 0 $ ] .
the physical requirement that any local operator has bounded norm requires that @xmath353 must be @xmath128 . next we will show that the dimension of @xmath208 is closely related to the correlation length of the state . at large system size @xmath316 , the correlator @xmath355(\sum_{\lambda}\lambda p_{\lambda}+r_{\lambda})^{l } \mathbb{e}[o_2])/\tr ( p_1 ) - \tr(p_1\mathbb{e}[o_1])\tr(p_1\mathbb{e}[o_2])/\tr^2 ( p_1)$ ] .
when @xmath96 is large , we keep only the first order term in @xmath356 and the correlator goes to @xmath357p_1\mathbb{e}[o_2])/\tr ( p_1)-\tr(p_1\mathbb{e}[o_1])\tr(p_1\mathbb{e}[o_2])/\tr^2 ( p_1)$ ] .
if @xmath208 is one dimensional , the two terms both become @xmath358| v_1\rangle\langle v_1|\mathbb{e}[o_2]|v_1\rangle$ ] and cancel each other for any @xmath359 and the second order term in @xmath356 dominates which decays as @xmath360 . for @xmath361 ,
the correlator goes to zero exponentially and the matrix product state as finite correlation length . on the other hand ,
if @xmath208 is more than one dimensional , the first order term has a finite contribution independent of @xmath96 @xmath362| v_j\rangle\langle v_j|\mathbb{e}[o_2]|v_i\rangle/\tr(p_1 ) - \langle v_i|\mathbb{e}[o_1]|v_i \rangle \langle v_j|\mathbb{e}[o_2]|v_j\rangle/\tr^2(p_1)$ ] , where @xmath363 are eigenbasis for @xmath208 .
therefore , degeneracy of the largest eigenvalue of the double tensor implies non - decaying correlation . to describe quantum states with finite correlation length ,
the double tensor must have a largest eigenvalue which is non - degenerate .
if the system is not translational invariant and @xmath39}$ ] vary from site to site , we can not diagonalize all the double tensors at the same time .
however , as shown in the ` canonical form ' of , there is a largest eigenspace of @xmath39}$](with eigenvalue @xmath78 ) such that the right eigenvector on site @xmath26 is the same as the left eigenvector on site @xmath79 .
therefore , when multiplied together this eigenspace will always have eigenvalue @xmath78
. there could be other eigenspaces with eigenvalue @xmath78 and matching eigenvectors from site to site .
however , then we can show similar to the ti case that this leads to an infinite correlation length . on the other hand
, other eigenspaces could have eigenvalues smaller than @xmath78 or they have mis - matched eigenvectors . if this is the case , all other eigenspaces decay exponentially with the number of sites multiplied together which gives rise to a finite correlation length .
we will say that @xmath39}$ ] has a non - degenerate largest eigenvalue @xmath78 for this case in general .
operators @xmath4 form a projective representation of symmetry group @xmath0 if @xmath364 here @xmath365 , the factor system of the projective representation , satisfies @xmath366 for all @xmath367 . if @xmath368 , this reduces to the usual linear representation of @xmath0 . a different choice of pre - factor for the representation matrices @xmath369 will lead to a different factor system @xmath370 : @xmath371 we regard @xmath372 and @xmath4 that differ only by a pre - factor as equivalent projective representations and the corresponding factor systems @xmath370 and @xmath373 as belonging to the same class @xmath113 .
suppose that we have one projective representation @xmath374 with factor system @xmath375 of class @xmath304 and another @xmath376 with factor system @xmath377 of class @xmath305 , obviously @xmath378 is a projective presentation with factor group @xmath379 .
the corresponding class @xmath113 can be written as a sum @xmath380 .
under such an addition rule , the equivalence classes of factor systems form an abelian group , which is called the second cohomology group of @xmath0 and denoted as @xmath116 .
the identity element @xmath381 of the group is the class that contains the linear representation of the group . here
are some simple examples : + ( a ) cyclic groups @xmath119 do not have non - trivial projective representation . hence for @xmath382 , @xmath116 contains only the identity element .
+ ( b ) a simple group with non - trivial projective representation is the abelian dihedral group @xmath383 .
for the four elements of the group @xmath384 , consider representation with pauli matrices @xmath385 , @xmath386 , @xmath387 , @xmath388 .
it can be check that this gives a non - trivial projective representation of @xmath201 .
+ ( c ) when @xmath141 , @xmath389 .
the two elements correspond to integer and half - integer representations of @xmath8 respectively .
+ ( d ) when @xmath3 , @xmath116 is trivial : @xmath390 .
we note that @xmath391 form a representation of @xmath392 when @xmath123 is an integer . but @xmath391 will form a projective representation of @xmath10 when @xmath123 is not an integer .
but under the equivalence relation , @xmath393 correspond to the trivial projective representation , if we choose @xmath394 .
note that @xmath395 can be a discontinuous function over the group manifold .
in this section , we explicitly solve the symmetry condition .
the goal is to 1 .
classify possible symmetry operations at fixed point and 2 .
find the corresponding symmetric fixed point state . for simplicity
, we drop the site index @xmath396 $ ] and rewrite as @xmath397 where @xmath4 is a projective or linear unitary representation of @xmath0 , the matrix @xmath398 is given by its matrix elements @xmath399 with @xmath400 , @xmath401 , and @xmath402 are sets of invertible matrices labeled by @xmath291 .
since @xmath403 is also a projective or linear unitary representation of @xmath0 , we can absorb @xmath121 into @xmath4 and rewrite as @xmath404 we note that matrix elements @xmath405 is non - zero only when @xmath406 , @xmath407 and the full set of @xmath408 form a complete basis in the space of @xmath409 dimensional matrices .
@xmath402 do not necessarily form a representation of @xmath0 .
but the fixed point form of the matrices requires that @xmath410 be a projective representation , as on the one hand @xmath411 and on the other hand @xmath412 therefore @xmath413 for all @xmath414 .
however , the set of matrices @xmath408 form a complete basis in the space of @xmath409 dimensional matrices . therefore ,
@xmath415 and @xmath289 and @xmath288 form two projective representations @xmath416 with @xmath417 and @xmath418 let us rewrite as @xmath419 we note that @xmath420 where the matrices @xmath421 and @xmath422 are given by @xmath423 since the set of matrices @xmath408 form a complete basis in the space of @xmath409 dimensional matrices , we find @xmath424 putting back the factor of @xmath118 , we find that @xmath425
from the solution in section [ soluu ] , we know that the fixed point state symmetric under linear on - site symmetry of group @xmath0 takes the form @xmath426 where @xmath117 is an entangled pair between the right virtual qubit on site @xmath26 and the left virtual qubit on site @xmath79(see fig . [
fig : fp ] upper layer ) .
each entangled pair is invariant under a linear symmetry transformation of the form @xmath114,r}(g)\otimes u^{[k+1],l}(g)$ ] @xmath427,r}(g)\otimes u^{[k+1],l}(g ) |ep_{k , k+1}\rangle = |ep_{k , k+1}\rangle\ ] ] but @xmath114,r}(g)$ ] or @xmath428,l}(g)$ ] alone might not form a linear representation of @xmath0 .
they could in general be a projective representation of @xmath0 .
if @xmath114,r}(g)$ ] is a projective representation corresponding to class @xmath113 in @xmath116 , then @xmath428,l}$ ] must correspond to class @xmath115 .
@xmath113 does not vary from site to site and labels a particular symmetric fixed point state .
now we will show that symmetric fixed point states with the same @xmath113 can be connected through symmetric lu transformations and hence belong to the same phase while those with different @xmath113 can not and belong to different phases .
first , suppose that two symmetric fixed point states @xmath429 and @xmath15 are related with the same @xmath113 , i.e. @xmath430,r}_1(g)\otimes u^{[k+1],l}_1(g ) |ep_{k , k+1}\rangle_1 = |ep_{k , k+1}\rangle_1 \\ \nonumber u^{[k],r}_2(g)\otimes u^{[k+1],l}_2(g ) |ep_{k , k+1}\rangle_2 = |ep_{k , k+1}\rangle_2\end{aligned}\ ] ] where @xmath431 is an entangled pair of virtual spins on hilbert space @xmath432,r}_{1(2 ) } \otimes \mathcal{h}^{[k+1],l}_{1(2)}$ ] .
@xmath114,r}_{1(2)}(g)$ ] is a projective representation of @xmath0 corresponding to @xmath113 on @xmath432,r}_{1(2)}$ ] and @xmath428,l}_{1(2)}(g)$ ] a projective representation corresponding to @xmath115 on @xmath433,l}_{1(2)}$ ]
. we can think of @xmath434 and @xmath435 as living together in a joint hilbert space @xmath436,r}_{1 } \oplus \mathcal{h}^{[k],r}_{2})\otimes(\mathcal{h}^{[k+1],l}_{1 } \oplus \mathcal{h}^{[k+1],l}_{2})$ ] .
the symmetry representation on this joint hilbert space can be defined as @xmath437,r}(g ) \otimes u^{[k+1],l}(g ) \\ \nonumber = & ( u^{[k],r}_1(g ) \oplus u^{[k],r}_2(g))\otimes(u^{[k+1],l}_1(g ) \oplus u^{[k+1],l}_2(g))\end{aligned}\ ] ] as @xmath114,r}_1(g)$ ] and @xmath114,r}_2(g)$ ] ( also @xmath428,l}_1(g)$ ] and @xmath428,l}_2(g)$ ] ) both correspond to @xmath113 ( @xmath115 ) , their direct sum @xmath114,r}(g)$](@xmath428,l}(g)$ ] ) is also a projective representation corresponding to @xmath113(@xmath115 ) .
therefore , we have a linear representation of @xmath0 on each site @xmath26 , @xmath114,l}(g ) \otimes u^{[k],r}(g)$ ] and both @xmath434 and @xmath435 are symmetric under @xmath114,r}(g ) \otimes u^{[k+1],l}(g)$ ] . before ( upper ) and after ( lower ) a local unitary operation on the shaded region that does not break the symmetry .
white dots correspond to @xmath381 , the identity element in @xmath116 , black dots correspond to @xmath113 and gray dots correspond to @xmath115.,width=288 ] now we can perform a lu transformation on the joint hilbert space and rotate continuously between @xmath434 and @xmath435 .
that is , @xmath438 where @xmath439 , latexmath:[$|b\rangle = @xmath128 to @xmath203 . by doing this locally to each pair
, we can map @xmath429 to @xmath15 ( and vice verse ) with lu transformations without breaking the on - site symmetry of group @xmath0 .
therefore , @xmath429 and @xmath15 belong to the same phase if they are related with the same @xmath113 . on the other hand ,
if @xmath429 and @xmath15 are related to @xmath304 and @xmath305 respectively , we will show that they can not be connected by any lu transformation that does not break the symmetry .
suppose that @xmath304 is non - trivial , we start with @xmath429 and apply a local unitary operation @xmath19 to a finite region ( shaded in fig . [
fig : inequiv ] ) .
@xmath429 is composed of invariant singlets of symmetry group @xmath0 .
if @xmath19 does not break the symmetry , the resulting state should still be composed of singlets .
the singlet pairs outside of the shaded region are not changed while those overlapping with the shaded region can take any possible structure after the operation @xmath19 .
no matter what the change is , the right virtual spin on the site to the left of the region corresponding to @xmath304 should form a singlet with some degrees of freedom in the region . as the singlet is invariant under a linear representation of @xmath0
, these degrees of freedom in the region must form a projective representation of @xmath0 corresponding to @xmath441 .
these degrees of freedom could live on one site or distribute over several sites .
however , the sites only support linear representations .
therefore , there must be some remaining degrees of freedom on the same sites that correspond to @xmath304 .
these remaining degrees of freedom must form singlets again with other degrees of freedoms in the region that correspond to @xmath441 .
we can continue this argument until finally some degree of freedom in the region corresponding to @xmath441 connect with the left virtual spin on the site to the right of the region corresponding to @xmath304 and form a singlet . in fig .
[ fig : inequiv ] , we illustrate one possible structure of singlets after operation @xmath19 .
white dots correspond to @xmath381 , the identity element in @xmath116 , black dots correspond to @xmath304 and gray dots correspond to @xmath441 .
therefore , we can see that no matter what the symmetric lu operation might be on @xmath429 , singlet entangled pairs related to @xmath304 must connect head to tail and cover the whole length of the chain .
in other words , we can not shrink a chain of singlet entangled pairs related to non - trivial @xmath304 continuously to a point or change it to @xmath305 by acting on it locally and without breaking the symmetry . hence fixed point states with different @xmath113
can not be related to each other with symmetric lu transformation and hence belong to different classes .
a gapped ti state can be represented by a uniform mps .
after @xmath24 steps of @xmath150-block rg transformation , we obtain a mps described by matrices @xmath442 which is given by .
to describe a state that does not break the on - site linear symmetry , here @xmath442 is invariant ( up to a phase ) under @xmath443 on each site .
therefore@xcite , @xmath444 must be satisfied with some invertible matrix @xmath288 . here
we have dropped the rg step label @xmath24 ( except in @xmath118 ) .
each coarse grained site is a combination of @xmath445 original lattice sites and @xmath118 form a 1d representation of @xmath0 .
so if the number of sites has a form @xmath446 , then @xmath180 in will have a form @xmath447^q\end{aligned}\ ] ] for any value of @xmath337 .
now let us choose @xmath448 where @xmath449 and @xmath450 have no common factors .
the total system size becomes @xmath451 .
we can perform , instead , a @xmath452 step of @xmath453-block rg transformation , which leads to a 1d representation @xmath454 .
we find that @xmath180 in will have a form @xmath455^{q'}\end{aligned}\ ] ] where @xmath456 .
thus @xmath457^ { \prod_{i=1}^{r ' } n'_i}= [ \al^{(r')}(g ) ] ^ { \prod_{i=1}^{r } n_i } .\end{aligned}\ ] ] since @xmath449 and @xmath450 have no common factors , there must exist a 1d representation @xmath181 of @xmath0 , such that @xmath458^ { \prod_{i=1}^{r } n_i } , \ \ \
\al^{(r')}(g ) = [ \al(g)]^ { \prod_{i=1}^{r ' } n'_i}.\end{aligned}\ ] ] now becomes @xmath459^{q \prod_{i=1}^{r } n_i } = [ \al(g)]^l\end{aligned}\ ] ] which gives us .
we have been using the quantum circuit formulation to study ti systems and classify b - phases .
as the quantum circuit explicitly breaks translational symmetry , it is possible that each b - phase contains several different ti phases .
( on the other hand , states in different b - phases must belong to different ti phases . ) in this section , we will show that each b - phase actually corresponds to a single ti phase by establishing a ti lu transformation between states in the same b - phase .
we will use the time evolution formulation of lu transformation and find a smooth path of gapped ti hamiltonian whose adiabatic evolution connects two states within the same b - phase .
first , as an example , we consider the case of ti only and show that there is only one gapped ti phase .
each ti src mps is described(up to local change of basis ) by a double tensor @xmath68 which has a non - degenerate largest eigenvalue @xmath78 .
@xmath68 can be written as @xmath460 where @xmath461 is the eigenvector of eigenvalue @xmath78 and @xmath462 is of eigenvalue less than @xmath78 . in
the ` canonical form'@xcite , @xmath463 , @xmath464 .
obviously , @xmath465 is a valid double tensor and represents a state in fixed point form .
we can smoothly change @xmath68 to @xmath465 by turning down the @xmath462 term to @xmath128 from @xmath466 to @xmath467 as @xmath468 every @xmath469 represents a ti src mps state . to see this , note that if we recombine the indices @xmath60 as row index and @xmath61 as column index and denote the new matrix as @xmath325 , then both @xmath325 and @xmath470 are positive semidefinite matrices .
but then every @xmath471 is also positive semidefinite , as for any vector @xmath472 @xmath473 @xmath469 is hence a valid double tensor and the state represented can be determined by decomposing @xmath469 back into matrices @xmath474 .
such a decomposition is not unique but wlog , we can fix the decomposition scheme , so that @xmath474 vary continuously with time and reach the fixed point form at @xmath467(up to local change of basis ) .
the state represented @xmath475 hence also changes smoothly with @xmath476 and has a finite correlation length as all eigenvalues of @xmath469 expect for @xmath78 are diminishing with @xmath476 .
therefore , @xmath469 represents a smooth path in ti src mps that connects any state to a fixed point state(up to local change of basis ) . how do we know that no phase transition happens along the path ?
this is because for every state @xmath475 , we can find a parent hamiltonian which changes smoothly with @xmath476 and has the state as a unique gapped ground state.@xcite following the construction in , we choose a sufficiently large but finite @xmath477 and set the parent hamiltonian to be @xmath478 , where @xmath479 is the projection onto the support space of the reduced density matrix on site @xmath26 to @xmath480 at time @xmath476 .
note that this hamiltonian is translation invariant . for large enough @xmath477
, @xmath479 will always be @xmath37 dimensional . as the state changes continuously ,
its reduced density matrices of site @xmath26 to @xmath480 changes smoothly . because the dimension of the space does not change , @xmath479 also changes smoothly with time . moreover
, it can be shown that @xmath481 is always gapped as the second largest eigenvalue of @xmath469 never approaches @xmath78 @xcite .
therefore , by evolving the hamiltonian adiabatically from @xmath466 to @xmath467 , we obtain a local unitary transformation@xcite connecting any state to the fixed point form , and in particular without breaking the translation symmetry . if the system is ti and has on - site symmetry , we need to maintain the on - site symmetry while doing the smooth deformation . a ti src mps which is symmetric under on - site symmetry of group @xmath0
is described by matrices which satisfy @xmath482 for some invertible projective representation @xmath288 .
the double tensor @xmath68 hence satisfies @xmath483 where the group element label @xmath291 has been omitted .
being the non - degenerate one - dimensional eigenspace of @xmath68 , @xmath465 must be invariant under the same transformation , and so does @xmath462 .
therefore we have @xmath484 now we smoothly change the double tensor as in eqn.[et ] .
evidently , the symmetry condition eqn.[symm_e ] is satisfied for all @xmath476 . decompose @xmath469 back to matrices @xmath474
so that the represented state @xmath475 changes smoothly with time . denote the symmetry transformed double tensor as @xmath485 . as @xmath486
, there must exist a unitary operator @xmath487 , such that @xmath488 where @xmath487 is a linear representation of @xmath0 .
redefine @xmath489 , then @xmath490 as @xmath474 is chosen to be continuous with time , from the above equation we can see that @xmath491 is also continuous in time . on the other hand ,
@xmath491 form a linear representation of @xmath0 .
for all the cases we are interested in , the linear representations of @xmath0 are discrete .
therefore , as @xmath491 evolves smoothly with time , it can not change from one representation to another but only from one equivalent form to another which differ by a unitary conjugation .
that is , @xmath492 , with a continuous @xmath493
. we can incorporate @xmath493 into the matrices @xmath474 and define @xmath494 , so that @xmath495 is symmetric under @xmath4 for all @xmath476 . in the following discussion
, we will assume that such a redefinition is made and the symmetry operation of the system will always be @xmath496 .
therefore , the continuous evolution of @xmath469 from @xmath466 to @xmath467 corresponds to a continuous evolution of short range correlated states @xmath475 which is always symmetric under the same on - site symmetry @xmath4 , with the same phase factor @xmath497 and related to the same projective representation @xmath113 .
such a smooth path in symmetric state space corresponds to a smooth path in symmetric hamiltonian space .
construct parent hamiltonian as discussed previously .
because the state is symmetric under on - site @xmath4 , the support space on site @xmath26 to @xmath480 must then form a representation space for @xmath498 . therefore , it is easy to see that the parent hamiltonian , being a summation of projections onto such spaces , is also symmetry under on - site @xmath4 .
moreover , the hamiltonian remains gapped and ti . in this way
, we have found a smooth path of symmetric , in particular ti , hamiltonian whose adiabatic evolution connects any symmetric state labeled by @xmath191 and @xmath113 to the corresponding fixed point state(up to a local change of basis ) and hence establishing the symmetric ti lu equivalence between them . as we show in appendix [ equiv ] that fixed point states with the same @xmath191 and @xmath113
can be related by symmetric local unitary transformations to each other , we now complete the proof that _ for 1d spin systems with only translation and an on - site linear symmetry @xmath0 , all gapped phases that do not break the two symmetries are classified by a pair @xmath499 where @xmath500 label different types of projective representations of @xmath0 and @xmath192 label different 1d representations of @xmath0 .
_ similarly , if the system has translation and parity symmetry , we can establish the equivalence between states labeled by the same @xmath501 and @xmath502 in a translational invariant way ( see the discussion below ) .
the procedure is totally analogous to the on - site symmetry case , with the only difference that the symmetry condition for the matrices and double tensors become @xmath503 we find that _ for 1d spin systems with only translation and parity symmetry , there are four gapped phases that do not break the two symmetries . _ n. schuch , d. prez - garca and j. i. cirac , talk given on ` international workshop on density matrix renormalization group and other advances on numerical renormalization group methods ' , beijing , china , 2010 ( http://physics.ruc.edu.cn/meeting/20090915/3109.pdf ) . | quantum many - body systems divide into a variety of phases with very different physical properties .
the question of what kind of phases exist and how to identify them seems hard especially for strongly interacting systems . here
we make an attempt to answer this question for gapped interacting quantum spin systems whose ground states are short - range correlated .
based on the local unitary equivalence relation between short - range correlated states in the same phase , we classify possible quantum phases for 1d matrix product states , which represent well the class of 1d gapped ground states . we find that in the absence of any symmetry all states are equivalent to trivial product states , which means that there is no topological order in 1d .
however , if certain symmetry is required , many phases exist with different symmetry protected topological orders .
the symmetric local unitary equivalence relation also allows us to obtain some simple results for quantum phases in higher dimensions when some symmetries are present . |
this work is partially supported by the national natural sciences foundation of china under grant no.10405031 .
the author thanks prof .
m. di toro for helpful discussions .
50 h. a. bethe , rev .
62 , 801 ( 1990 ) i. bombaci , in isospin physics in heavy - ion collisions at intermediate energies , edited by b.a.li and w. udo schrder ( nova science publishers , inc .
, new york , 2001 ) , p35 a.w .
steiner , m. prakash , j.m .
lattimer , p.j .
ellis , phys .
( 2005 ) 325 b.a.brown , phys .
85(2000)5296 b.a.li , phys .
85(2000)4221 ; ibid.88(2002)192701 c.j.horowitz , j.piekarewicz , phys
lett.86(2001)5647 c.j.horowitz , j.piekarewicz , phys . rev .
c64(2001)062802(r ) ; ibid , c66(2002)055803 b.g.todd , j.piekarewicz , phys . rev .
c67(2003)044317 y.b .
shen , et.al .
phys.lett.b586(2004)225 w.z.jiang , y.l.zhao , phys .
b617(2005)33 l.w.chen , v.greco , c.m.ko , b.a .
li , phys .
90(2003)162701 m. di toro , a. drago , t. gaitanos , v. greco , a. lavagno , nucl - th/0602052 j.carriere , c.j.horowitz , j.piekarewicz , astrophys .
j 593(2003)463 w.z.jiang , y.l.zhao , z.y.zhu , s.f.shen , phys . rev .
c72 ( 2005)024313 s.aoki , et al prog . theor . phys
. 85(1991)1287 j.k.ahn , et al phys .
87(2001)132504 j.k.takahashi , et al phys .
87(2001)212502 j.schaffner , c. b.dover , a.gal , c.greiner , h.stcker , phys .
71(1993)1328 j.schaffner , c.b.dover , a.gal , c.greiner , d.j .
millener , and h.stcker , ann .
235 ( 1994 ) 35 d.javorsek ii , e.fischbach and d.elmore,phys .
d67(2003)034015 n.k.glendenning , phys.rept.342(2001)393 n.k .
glendenning , astrophys .
j. 293 ( 1985 ) 470 n.k .
glendenning , s.a .
moszkowski , phys .
67 ( 1991 ) 2414 j.schaffner-bielich , m.hanauske , h.stcker , w.greiner , phys .
89(2002)171101 j.schaffner-bielich , a.gal , phys .
m.prakash , i.bombaci , m.prakasha , p.j .
ellis , j.m .
lattimer and r.knorren , phys .
280(1997)1 j.mares,e.friedman , a.gal , and b.k.jennings , nucl .
b.d.serot , j.d.walecka , adv .
nucl . phys .
16(1986)1 ; int.j.mod.phys .
e6(1997)515 p.ring , prog . part .
nucl . phys .
37(1996)193 g.a.lalazissis , j.knig , p.ring , phys .
j.meng , h. toki , j.y.zeng , s.q.zhang , and s.g.zhou , phys .
c 65 ( 2002 ) 041302(r ) r.michaels , p.a.souder , g.m.urciuoli , spokespersons , jefferson laboratory experiment e-00 - 003 | 18.0pt the role of the isoscalar hyperon @xmath0 in probing the density dependence of the nuclear symmetry energy is studied in multi-@xmath0 hypernuclei , hyperon - rich matter and neutron stars in relativistic models .
relationships between the properties of three types of objects and the neutron thickness in @xmath1pb are established with respect to the isoscalar - isovector coupling that modifies the density dependence of the symmetry energy .
the exotic isotopes far from the neutron drip line can be stabilized by filling in considerable @xmath0 hyperons .
the difference of the binding energy of multi-@xmath0 hypernuclei from different models is attributed to different symmetry energies .
the isovector potential together with the neutron thickness in multi-@xmath0 hypernuclei investigated is very sensitive to the isoscalar - isovector coupling .
the large sensitivity of the @xmath0 hyperon fraction to the isoscalar - isovector coupling occurs at about @xmath2 in beta equilibrated hyperon - rich matter . in neutron stars with hyperonization ,
an on - off effect with respect to the isoscalar - isovector coupling exists for the neutron star radius .
= 15.5 cm = 22.5 cm = -0.0 cm = 7.1 mm = -2.9 mm 20.6pt the symmetry energy is pervasively related to the descriptions of nuclear structures and reactions in laboratories and astrophysics@xcite . however , the density dependence of the symmetric energy is still poorly known@xcite to date .
therefore , theoretical uncertainties exist in the understanding of a large number of properties of neutron stars and heavy nuclei@xcite , nuclei far from the @xmath3 stability@xcite , heavy ion collisions@xcite and consequently probable hadron - quark phase transition@xcite , etc .
many factors that influence the isovector potential directly or indirectly can modify the behavior of the density dependence of the symmetry energy .
recently the density dependence of the symmetry energy has been extensively explored through the inclusion of the isoscalar - isovector coupling terms in relativistic mean field ( rmf ) models@xcite .
these new terms enable one modify the neutron skin of heavy nuclei without changing the charge radius and a variety of ground - state properties that are well constrained experimentally .
on the other hand , the charge density distributions for some light exotic nuclei near drip lines that are not determined yet are sensitive to the isoscalar - isovector couplings@xcite .
hypernuclei that are produced through strangeness transfer reactions mostly contain only one hyperon , and just a few double @xmath0 hypernuclear events were observed@xcite . through heavy - ion collisions it may possibly create an environment which is favorable to the formation of metastable exotic multihypernuclear objects ( memos ) such as multi-@xmath0 hypernuclei .
schaffner et.al . showed that multi-@xmath0 hypernuclei were more strongly bound than normal nuclei using the rmf model@xcite .
recently , jevorsek et.al .
set limits on the existence and properties of exotic objects such as strangelets and memos , and provided limits as low as @xmath4 for m / z up to 120 through a reanalysis of related data@xcite . in bulk matter
, strangeness can be formed by virtue of the strong interaction , and hyperons may be important constituents of neutron stars .
many important effects of hyperonization in neutron stars have been found in the past ( for a review , see@xcite ) .
for instance , hyperonization reduces the maximum mass of neutron stars as much as 3/4@xmath5@xcite .
schaffner et.al . found that a phase transition to hyperon matter may even possibly occur in neutron stars@xcite . in this work , it is the aim to constrain the density dependence of the symmetry energy through the investigation of properties of multi-@xmath0 hypernuclei , hyperon - rich matter and neutron stars with hyperonization .
we will perform the investigation in rmf models considering the isoscalar - isovector coupling terms that soften the symmetry energy at large densities .
the modification in memos and hyperon - rich matter due to the isoscalar - isovector couplings is usually considered to be small due to the relatively weak meson - hyperon coupling . as pointed out in refs.@xcite that the memos may reach a very high baryon density ( @xmath6 ) like strangelets . in hyperon - rich matter
, the appearance of @xmath0 hyperons occurs at about twice normal density .
since the higher density relates to the larger isoscalar potentials ( or fields ) , the isoscalar - isovector couplings will induce large modifications . in hyperon - rich matter
, we will focus on the relation between the @xmath0 fraction and the isoscalar - isovector coupling .
as the determination of properties of neutron stars with hyperonization needs the equation of state of hyperon - rich matter , we may search for the observable consequence relating to the density dependence of the symmetry energy . as examples of memos
, we will consider multi-@xmath0 hypernuclei based on the ca isotope far off the normal neutron drip line at the fixed isospin . in this way
, we can observe the extension of the drip line , and explore the density dependence of the symmetry energy by shifting the isoscalar field . in this study , additional strange mesons @xmath7
( i.e. @xmath8 , 975mev ) and @xmath9 ( 1020mev ) are also included in rmf models to describe the strong @xmath10 attraction as in refs.@xcite .
the effective lagrangian density is given as follows & = & _ b[i_^-m_b+g_b - g_b _ ^-g_b__3 b_0^ + & & + _
-e(1+_c)_a^]_b - u(,^ , b_0^ ) + & & + ( _ ^-m_^2 ^ 2)- f_f^+ m_^2_^ + & & - b _ b^+ m_^2 b_0 b_0 ^ -a
_ a^ + l_y [ eq : lag1 ] and @xmath11 is for the strange meson - hyperon interactions and free fields of strange mesons _
y&= & _ y[g_^*y^ * - g_y _ ^]_y+ ( _ ^*^^*-m_^*^2^*^2 ) + & & - ( ^^-^^)(__-__)+ m_^2_^ [ eq : lag2 ] where @xmath12 , and @xmath13 are the fields of the baryon , scalar , vector , and charge - neutral isovector - vector mesons , with their masses @xmath14 , and @xmath15 , respectively .
@xmath16 is the field of the photon . @xmath17 and @xmath18 are the corresponding meson - baryon couplings .
@xmath19 is the third component of isospin pauli matrix for nucleons and @xmath20 for the @xmath0 hyperon .
@xmath21 is a constant relating to the baryon charge .
@xmath22 , @xmath23 , and @xmath24 are the strength tensors of the @xmath25 , @xmath26 mesons and the photon , respectively f_=__- _ _ , b_=_b_0 -_b_0 , a_=_a _ -_a_the self - interacting terms of @xmath27-mesons and the isoscalar - isovector ones are in the following u(,^ , b_0^)&=&g_2 ^ 3+g_3 ^ 4 -c_3(_^)^2 + & & -4g^2_n ( _ s g_n^2 ^ 2+_v g_n^2 _ ^)b_0b_0^[equ1 ] for the hyperon - rich matter , the chemical equilibration is established on weak interactions of baryons and leptons as in ref.@xcite .
we study chemically equilibrated and charge neutral hyperon - rich matter including baryons @xmath28 and leptons @xmath29 without @xmath30 hyperons , since @xmath30 hyperons do not appear@xcite according to the fact that the isoscalar potential changes sign in the nuclear interior and becomes repulsive based on recent @xmath31 atomic data @xcite . for the finite system ,
we consider the spherical case .
the solution detail can be referred to refs.@xcite , and it is not repeated here .
the pairing correlation of nucleons in nuclei is considered in the bcs approximation as in refs.@xcite .
we perform calculations with the nl3 @xcite and s271@xcite parameter sets .
these two models share the same binding energy per nucleon ( 16.24mev ) and incompressibility ( 271mev ) at the same fermi momentum @xmath32@xmath33 ( @xmath34@xmath35 ) .
the symmetry energy is given by@xcite a_4=+ [ sym]with @xmath36 and @xmath37 .
the density dependence of the symmetry energy is different between two models because of different @xmath38 , @xmath39 and @xmath40 ( @xmath41 for the nl3 and @xmath42 for the s271 ) .
the @xmath25 and @xmath43 meson couplings to the hyperons are given by the su(3 ) relations : @xmath44 , @xmath45 and @xmath46 .
the @xmath47 meson coupling to the hyperons is determined by reasonable hyperon potentials in nuclear matter : @xmath48mev and @xmath49mev .
the coupling constants @xmath50 and @xmath51 are taken to satisfy the su(6 ) relation : @xmath52 , and the @xmath53 is fitted to improve the @xmath10 interaction matrix element @xmath54@xcite . in ref.@xcite , the @xmath53 is taken in a range close to the @xmath55 .
we take the ratio of the scalar coupling constant to be @xmath56 in nl3 with a potential for the @xmath0 hyperon in @xmath57 matter @xmath58mev . in s271 ,
the same potential is taken to obtain the ratio @xmath59 . in this way
, the @xmath54 of @xmath60be is improved from 0.38 mev to 2.2 mev in nl3 , and from 0.2 mev to 3.9 mev in s271 , much close to the extracted value @xmath61 mev .
the @xmath62 is determined by the @xmath57 hyperon potential in @xmath57 matter @xmath63mev@xcite .
the @xmath64 tensor coupling is vanishing , and the @xmath65 tensor coupling is small but adjusted to simulate the vanishing spin - orbit splitting for the @xmath0 hyperon observed in @xmath66o . for simplicity
, we perform calculations for various isoscalar - isovector coupling @xmath67 s , and the @xmath68 is set as zero . for a given coupling @xmath67
, we follow ref.@xcite to readjust the @xmath69 coupling constant @xmath38 so as to keep an average symmetry energy fixed as 25.7 at @xmath70 @xmath33 . in doing so
, it was found in ref.@xcite that the binding energy of @xmath1pb is nearly unchanged for various @xmath67 s .
below , we will see how the total binding energy of multi-@xmath0 hypernuclei is modified by the @xmath67 .
the binding energy per baryon for various @xmath71ca+@xmath0 s systems is displayed in fig.[fig1 ] .
metastable minima appears with the rise of the @xmath0 hyperon number @xmath72 .
the additional @xmath10 attraction , simulated by the strange mesons , produces much larger binding energy at larger @xmath73 and shifts the minima outwards .
the difference of binding energies of multi-@xmath0 hypernuclei in nl3 and in s271 enlarges explicitly with the rise of @xmath72 .
since both models share the same saturation properties and take the same potential for the @xmath0 hyperon , this enlargement is attributed to the different density dependence of the symmetry energy . with the rise of @xmath72 , the central density of multi-@xmath0 hypernuclei increases .
for instance , the central density in @xmath71ca@xmath74 is about 1.43@xmath75 in nl3 which is much larger than 1.06@xmath75 in @xmath71ca .
the more different symmetry energy at the larger density domain , shown in the inset in fig.[fig1 ] , thus accounts for the large difference of the binding energy at larger @xmath73 .
fig.[fig2 ] shows that the two - neutron separation energy of @xmath76ca+@xmath0 s systems increases from the negative to the positive with the rise of @xmath72 . as the @xmath72 increases , the nucleon potential is enhanced by the attraction between hyperons and nucleons , being able to fill in more neutrons under the continuum . the neutron drip line can thus be largely extended .
for instance , the drip line of ca isotope can be extended from the neutron number 50@xcite to 82 as long as about 50 @xmath0 s are filled in .
the dependence of the separation energy on the isoscalar - isovector coupling @xmath67 reduces approximately with the increase of @xmath72 . in refs.@xcite , a large class of memos
are constructed by adding multi - hyperons in stable nuclei , while here we aim at the extension of the neutron drip line by filling in @xmath0 hyperons in unstable isotopes . as multi-@xmath77 hypernuclei
may exert their roles in extending the neutron drip line and binding baryons more strongly than in normal nuclei , they will potentially influence the astrophysical nucleosynthesis and the abundances of elements . to search for multi-@xmath0 hypernuclei with the @xmath72 up to 50
is possible but certainly difficult@xcite . here
, we just devote ourselves to investigating the sensitivity of properties of such exotic systems to the density dependence of the symmetry energy .
the sensitivity of the total binding energy to the @xmath67 is different for various @xmath71ca+@xmath0 s systems . for @xmath71ca+@xmath0 s systems that are unstable against the neutron emission at small @xmath72 and unstable against the @xmath0 hyperon emission at very large @xmath72 ,
the total binding energies can be modified by as much as tens of mev for various @xmath67 s . for stable @xmath71ca+@xmath0 s systems
whose binding energies are situated on or very close to the minima , shown in fig.[fig1 ] , the modification to the total binding energies by various @xmath67 s is quite small .
for instance , the total binding energy for @xmath71ca+50@xmath0 in nl3 is just modified by up to 2mev with various @xmath67 values .
a data to data correlation of the neutron thicknesses between @xmath1pb and stable multi-@xmath0 nuclei can be established as in refs.@xcite .
the neutron thickness of these stable multi-@xmath0 nuclei is very sensitive to the @xmath67 .
for instance , the uncertainty of the neutron thickness for @xmath71ca+40@xmath0 and @xmath71ca+50@xmath0 is about 3 times that for @xmath1pb .
the proton density distribution in these exotic system is also sensitive to the @xmath67 and one third of the uncertainty of the neutron thickness comes from the protons .
the large uncertainty of the neutron thickness is attributed to the addition of the @xmath0 hyperons , which will be elucidated below .
usually , one modifies the isospin dependent potential considering the shift of the isospin and the isovector field directly . here , we modify it by enhancing the isoscalar field ( @xmath25 field ) through filling in @xmath0 s without changing the isospin of the system , which is fulfilled by considering the isoscalar - isovector coupling . in fig.[fig4 ]
, the isovector potentials are displayed for @xmath71ca and @xmath71ca+@xmath78 in nl3 . in s271
, similar results for these two systems can be obtained .
the isovector potentials for various @xmath67 s in the exotic isotope are largely shifted by filling in considerable @xmath0 hyperons .
the uncertainty of the isovector potential against the @xmath67 in @xmath71ca is much larger in the larger radius region , compatible with the understanding that @xmath71ca is totally diffusive and unstable .
compared to results for @xmath71ca , the sensitivity of the isovector potential to the @xmath67 in @xmath71ca+@xmath78 is shifted to the central region , and at the large radius the convergence occurs , which is favored by the extension of the neutron drip line .
the results shown in fig.[fig4 ] for two exotic systems are related to different density profiles .
the central density in @xmath71ca+@xmath78 is about one third higher than that in @xmath71ca .
the deviation of the symmetry energy for various @xmath67 s enlarges with the rise of the density beyond the normal density or even much lower density@xcite . at the central region in @xmath71ca+@xmath78 ,
the isovector potential is more sensitive to the @xmath67 than that in @xmath71ca .
the isoscalar potential can also be shifted to a certain degree by the @xmath67 , whereas for the normal nuclei the shift is negligible .
the sensitivity of the isovector potential to the @xmath67 is responsible for the uncertainty of the neutron thickness .
the @xmath0 hyperon constituent is important in astrophysical bulk matter to explore the effects of the density dependence of the symmetry energy .
we plot the @xmath0 and @xmath79 fractions defined by @xmath80 in fig.[fig5 ] with the s271 based on a self - consistent calculation .
the large sensitivity of the hyperon fractions to the @xmath67 occurs at about @xmath2 where the @xmath0 hyperon starts to appear . since the density as high as @xmath2 is accessible through the heavy ion collisions with the colliding energy about 1gev per nucleon , the detection related to the hyperons , especially the @xmath0 s , in heavy ion collisions is helpful to probe the density dependence of the nuclear symmetry energy .
the starting density for the @xmath0 appearance is model dependent .
for instance , it is about 1.8@xmath81 in nl3 and 2.2@xmath81 in s271 . as shown in the inset in fig.[fig5 ] , a shift of the isovector potential starts to appear with the @xmath0 appearance , and the isovector potential differs more from that without hyperonization with the increase of the density . in multi-@xmath0
nuclei far from the normal drip line , the isovector potential is shifted by filling in @xmath0 s under the fixed isospin , while here the shift comes from the change of the isospin and the isoscalar field together . the convergent extent with respect to the @xmath67 exists for the hyperon fraction in the whole density domain
this indicates that it is favorable to include the isoscalar - isovector coupling terms for the model itself . at very high densities
, the convergence of hyperon fractions occurs .
this is attributed to the fact that the isospin is largely suppressed with the drop of the neutron fraction at high densities .
the isovector potential is thus small compared to the isoscalar potential , so that the isoscalar potential is little changed through the isoscalar - isovector coupling .
this implies that the hyperstars@xcite whose core is solely composed of hypermatter at high densities will be insensitive to the @xmath67 .
the large reduction of the maximum mass of neutron stars by hyperonization , firstly pointed out by glendenning ( see ref.@xcite and references therein ) , can be reproduced here . in nl3 ,
hyperonization in the core reduces the maximum neutron star mass by about 0.9@xmath5 from about 2.8@xmath5 , and by 0.75@xmath5 from about 2.4@xmath5 in s271 .
compared to models in which hyperons are absent , hyperonization in the neutron star whose mass is near the maximum mass increases the central baryon density by as much as 1@xmath81 in the region @xmath82 km . for a 1.4 solar - mass neutron star ,
the role of hyperonization is quite limited because of relative low core densities . using the oppenheimer - volkoff equations for bulk
matter in beta equilibrium , we thus investigate neutron stars with relative large masses : 1.8@xmath5 with the nl3 model and 1.62@xmath5 with the s271 model . in fig.[fig6 ] , the radii of neutron stars with respective masses in two models are displayed as a function of the neutron thickness for @xmath1pb .
the radius r of neutron star without hyperonization increases with the neutron thickness in @xmath1pb for a given parameter set , which is the same as in ref.@xcite . for neutron stars with hyperonization ,
the trend is similar , whereas the correlation between the r and the neutron thickness in @xmath1pb almost vanishes in a large region of the neutron thickness in @xmath1pb .
this property is consistent with the convergent extent shown in fig.[fig5 ] .
it indicates that for a neutron star with hyperonization the radius r nearly just depends on the on - off switch rather than concrete values of the isoscalar - isovector coupling in a given model .
an accurate measurement of the neutron radius for @xmath1pb at the jefferson laboratory@xcite determines the density dependence of the symmetry energy at low densities , as pointed out in ref.@xcite .
the measurement of the radius of the neutron star with hyperonization can discriminate two regions of the @xmath67 s to which the r is either very sensitive or insensitive at high densities .
various information through separate measurements may be combined together to constrain the isoscalar - isovector coupling that modifies the density dependence of the symmetry energy in the whole density region . in summary , we have studied the roles of the isoscalar @xmath0 hyperons in exploring the density dependence of the symmetry energy in multi-@xmath0 hypernuclei , hyperon - rich matter and neutron stars with rmf models .
relationships between the properties of three types of objects and the neutron thickness in @xmath1pb are established with respect to the isoscalar - isovector coupling that modifies the density dependence of the symmetry energy .
the exotic isotopes far from the neutron drip line can be stabilized by filling in considerable @xmath0 hyperons , and the neutron drip line is substantially extended outwards .
the difference of the binding energy of multi-@xmath0 hypernuclei from different models is attributed to different symmetry energies .
the isovector potential together with the neutron thickness in multi-@xmath0 hypernuclei situated on the new drip line that is far from the normal drip line is very sensitive to the isoscalar - isovector coupling . in beta
equilibrated hyperon - rich matter , we have investigated the sensitivity of the hyperon fractions to the isoscalar - isovector coupling based on the self - consistent calculation for hyperons and nucleons .
the large sensitivity of the @xmath0 hyperon fraction to the isoscalar - isovector coupling occurs at about @xmath2 .
the properties of neutron stars are investigated . an on - off effect with respect to the isoscalar - isovector coupling exists for the radius of neutron stars with hyperonization . besides the detection of @xmath0 hyperons in the heavy ion collisions ,
the measurements of the neutron star radius and neutron thickness in @xmath1pb could help set up the constraints on the density dependence of the symmetry energy . |
in this paper we propose how to utilize a recent idea of guerin _ et al .
_ @xcite who applied the well - known dykhne - davis - pechukas ( ddp ) method @xcite for optimization of adiabatic passage in a two - state system .
in order to adapt this approach to stirap , we reduce the three - level raman system to effective two - state systems in two limits : on exact resonance and for large single - photon detuning .
the optimization , which minimizes the nonadiabatic transitions and maximizes the fidelity , leads to a particular relation between the pulse shapes of the driving pump and stokes fields .
the probability amplitudes in a two - state system @xmath0 ^{t}$ ] satisfy the schrdinger equation , @xmath1where the hamiltonian in the rotating - wave approximation ( rwa ) reads b.shore @xmath2 .
\label{h2}\]]the detuning @xmath3 is the difference between the transition frequency @xmath4 and the carrier laser frequency @xmath5 .
the time - varying rabi frequency @xmath6 describes the laser - atom interaction , where @xmath7 is the electric dipole moment for the @xmath8 transition and @xmath9 is the laser electric field envelope . for the derivation of the transition probability we shall need the adiabatic basis , i.e. the basis of the eigenstates of the hamiltonian ( [ ha ] ) .
we summarize below the basic definitions and properties of this basis .
the probability amplitudes in the diabatic and adiabatic bases are connected via the rotation matrix @xmath10 ,
\]]as @xmath11where the column - vector @xmath12^{t}$ ] comprises the probability amplitudes of the adiabatic states @xmath13 and @xmath14 .
these amplitudes satisfy the transformed schrdinger equation , @xmath15where the transformed hamiltonian is given by @xmath16 , \end{aligned}\]]where the overdots denote time derivatives . for the reader convenience
we write the expresion for the nonadiabatic couplig @xmath17 in terms of @xmath18 and @xmath19 . using the definition given by eq.([theta ] ) easely can be seen that @xmath20 in terms of the mixing angle @xmath21 , defined as @xmath22 the eigenstates of @xmath23 read [ adiabatic states ] @xmath24the time dependences of the adiabatic states @xmath13 and latexmath:[$% @xmath21 , whereas the bare ( diabatic ) states @xmath26 and @xmath27 are stationary .
the energies of the adiabatic states are the eigenvalues of @xmath23 , @xmath28 .\ ] ] the splitting between them is given by @xmath29hereafter we will consider level crossing models . because the rabi frequency @xmath18 vanishes at large times , @xmath30 , and because the detuning @xmath19 sweeps from minus to plus infinity , @xmath31 , the mixing angle @xmath21 rotates clockwise from @xmath32 to @xmath33 , and the composition of the adiabatic states changes accordingly .
asymptotically , each adiabatic state becomes uniquely identified with a single diabatic state , [ diabatic - adiabatic ] @xmath34because of the level crossing each adiabatic state connects _ different _ bare states at @xmath35 and @xmath36 . in the adiabatic limit ,
the system starts in state @xmath26 and follows the adiabatic state @xmath37 . hence adiabatic evolution and level crossing lead to complete population transfer .
it is important to note that the probability of transition in the adiabatic basis @xmath38 is equal to the probability of no transition in the diabatic basis , @xmath39 we will continue with the description of the dykhne - davis - pechukas ( ddp ) method , which gives the adiabatic probability @xmath38 , and we shall use eq .
( [ pp ] ) to find the diabatic probability @xmath40 .
a useful and very accurate technique for obtaining the final transition probabilities is the dykhne - davis - pechukas method , or ddp method , first introduced by dykhne and given a rigorous mathematical formulation later by davis and pechukas @xcite .
the basic idea of the ddp method is that , in the adiabatic limit , the two - state coupling is universal , independent of a given model , and the contributions to the t transition probability @xmath40 between the adiabatic states are given by the complex plane zeros of the adiabatic eigenenergies , in the form of an exponential .
dykhne - davis - pechukas ( ddp ) approximation , @xcite , which provides the asymptotically exact transition probability @xmath40 can be also used to estimate the non - adiabatic effects @xcite .
the ddp formula reads @xmath41where @xmath42is an integral over the eigenenergy splitting @xmath43 .
the point @xmath44 is called the transition point and it is defined as the ( complex ) zero of the quasienergy splitting , @xmath45which lies in the upper half of the complex @xmath46-plane ( i.e. , with im@xmath47 ) .
equation ( [ dp-1 ] ) gives the correct asymptotic probability for nonadiabatic transitions provided : ( i ) the quasienergy splitting @xmath43 does not vanish for real @xmath46 , including at @xmath48 ; ( ii ) @xmath43 is analytic and single - valued at least throughout a region of the complex @xmath46-plane that includes the region from the real axis to the transition point @xmath49 ; ( iii ) the transition point @xmath44 is well separated from the other quasienergy zero points ( if any ) and from possible singularities ; ( iv ) there exists a level ( or stokes ) line defined by @xmath50which extends from @xmath35 to @xmath36 and passes through @xmath44 . as has been pointed out already by davis and pechukas @xcite , for the landau - zener model , which possesses a single transition point , the ddp formula ( [ dp-1 ] ) gives the exact transition probability , not only in the adiabatic limit but also in the general case .
this amazing feature indicates the relevance of the ddp approximation . in the case of more than one zero points in the upper @xmath46-plane , davis and
pechukas @xcite have suggested , that eq .
( [ dp-1 ] ) can be generalized to include the contributions from all these @xmath51 zero points @xmath52 in a coherent sum .
this suggestion has been later verified joye93,suominen92oc , suominen92pra .
the generalized ddp formula has the form @xmath53where the @xmath54 factors are defined by @xmath55and they usually take values @xmath56 or @xmath57 . here
@xmath17 accounts for the nonadiabatic coupling between the adiabatic states , with @xmath58 . in principle , eq . ( [ dp - n ] ) should be used when there are more than one zero points lying on the lowest stokes line ( the closest one to the real axis ) and should include only the contributions from these zeroes .
the contributions from the farther zeroes are exponentially small compared to the dominant ones and may therefore be neglected .
it is shown in @xcite that dykhne - davis - pechukas ( ddp ) method davis76 can be used to examine the adiabatic limit of population transfer in two - level models driven by a chirped laser field .
in @xcite the final population transfer for different trajectories in the parameter space in the adiabatic limit is analyzed .
after using the scaled time @xmath59 , where the parameter @xmath60 under the limit @xmath61 can be viewed as adiabatic limit , we can write the new scaled schrodinger equation eq.([sch - eq ] ) for the two - state system . in eq.([h2 ] ) , we can parameterize the trajectories defined from @xmath18 and @xmath19 as a function of time , by assuming a given smooth pulse shape function @xmath62 , which has its maximum for @xmath63 .
this pulse shape function is related to the coupling by @xmath64 where @xmath65 is the two - state peak rabi frequency . in reason to have the constant eigenenergy splitting @xmath43 , accordingly to eq.([rabi - ts ] ) the detunig is defined by @xmath66 the parametrization given by eq.([rabi - ts ] ) and eq.([delta - ts ] ) implies @xmath67 \lambda ^{2}(t ) } \label{e - ts}\ ] ] using the ddp method , eq.([tc - def ] ) and eq.([e - ts ] ) give @xmath68 for particular class of analytic functions , defined with @xmath69 , the following condition is fulfilled @xmath70 using eq.([transition points - ts ] ) for this class of models defined with @xmath71 the difference between @xmath65 and @xmath72 to tends to zero @xmath73 is necessary and sufficient condition the imaginary part of the transition points to tends to infinity @xmath74 from eq.([d(tc ) ] ) can be seen that @xmath75 therefore , from eq.([dp-1 ] ) follows that the dominant nonadiabatic correction given by the ddp formula vanishes for the level lines defined by eq.([rabi - ts ] ) , eq.([delta - ts ] ) and @xmath76 as have been pointed in @xcite in the adiabatic regime , the optimum level lines can be seen as a boundary between decreasing and oscillating regimes for the nonadiabatic correction .
as we have expalined , gurin _ et al .
_ @xcite have used the ddp method to optimize the adiabatic passage between two states , assuming that the probability for nonadiabatic losses could be determined by the brhaviour of the transition points @xmath77 .
they have proposed to suppress the nonadiabatic losses altogether by choosing the rabi frequency @xmath18 and the detuning @xmath19 such that there are _ no transition points_. this condition is obviously fulfilled if the quasienergy splitting is _ constant _ , @xmath78the later condition also manifests the choice of a detuning and rabi frequency defined with eqs .
( [ delta - ts ] ) and ( [ rabi - ts ] ) .
easely can be seen that the same detuning and rabi frequency functions could be parameterized as @xmath79 , \qquad \omega ( t)=\omega _ { 0}\cos \left [ \frac{\pi } { 2}f(t)\right ] , \label{omega - delta } \\
-1=f(-\infty ) \leqq f(t)\leqq f(\infty ) = 1 , \label{f(t)}\end{gathered}\ ] ] with @xmath80 being an arbitrary monotonically increasing function with the above property . according to the ddp method such models
do not have transition points and lead to vanishing nonadiabatic corrections .
the optimization based on ddp is not exact in the sense that it is performed by using approximate technique .
it is interesting to calculate the corrections to this ddp optimization .
although we are able to design models that yield according to the ddp , vanishing nonadiabatic corrections there is no way to calculate the transition probability for these models again using ddp method in the diabatic basis .
ddp approximation comprise nonadiabatic corrections not only from the first - order perturbation theory in the adiabatic basis but adding the contributions from all orders via correct prefactor .
even so , using ddp not in the diabatic but in the first adiabatic basis is instructive .
this is also a way to calculate the deviation form the ddp optimization and to reveal the nature of the oscillation behavior for the transition probability for models that are designed to minimize the nonadiabatic corrections @xcite .
we note that the the probability of transition in the adiabatic basis @xmath38 is equal to the probability of no transition in the diabatic basis and the both are related via ( [ pp ] ) . using the parametrization ( [ omega - delta ] ) for the schrdinger equation in the adiabatic basis , the transformed hamiltonian up to phase transformation
is given by @xmath81 , \label{ha2}\ ] ] where the nonadiabatic coupling is given by ( [ non - ad - coup ] ) .
nonetheless a particular level crossing model does not have a transition points in the diabatic basis , in the first adiabatic basis , such a model depending on the function @xmath80 would have transition points .
this means that applying a ddp method in the first adiabatic basis instead of the diabatic one , one could analyze the nonadiabatic corrections of the optimized adiabatic passage . as a particular example we will consider a level crossing model , @xmath82 , \qquad \omega ( t)=\omega _ { 0}\cos \left [ \frac{\pi } { 2}\text{erf}(t / t)% \right ] . \label{erf}\ ] ] this model in the adiabatic basis
is related to the gaussian model , which is seen from the hamiltonian ( [ ha2 ] ) .
\frac{\sqrt{\pi } } { t\text { } } \exp \left [ -\left ( t / t\right ) ^{2}\right ] & -\omega _ { 0}% \end{array}% \right ] \mathbf{a}(t),\ ] ] the analytic estimation of the transition probability for the gaussian model could be performed using ddp .
we will briefly review details of such calculation , but the reader could find a similar calculation in more details in @xcite . for the gaussian model ( [ model ] ) , there are infinitely many transition points in the upper half - plane . in terms of the dimensionless time @xmath84 ,
they are given by [ transition points ] @xmath85where @xmath86 and @xmath87 for @xmath88 , we have [ transition points small ] @xmath89hence , as @xmath90 decreases , the transition points approach the imaginary axis and in the limit @xmath91 coalesce ( logarithmically ) with their counterparts in the second quadrant . as we have mentioned the transition probability for the models choosen to satisfy the ddp optimization condition ( [ trajectory ] ) show oscillating behavior , although they should yield optimized addiabatic passage .
this is due to the fact that ddp is approximate method .
a particular model without transition points in the diabatic basis , generally has a transition point in the first adiabatic basis .
as in the gaussian model the contributions from this transition points lead to oscilations , according to ddp formula ( dp - n ) .
it is important to note the relation between the asymptotic bahaviour of the transition points for @xmath88 and the adiabatic limit . from the definition of the model ( [ erf ] ) is clear that the limit @xmath92 can be seen as the adiabatic limit , so @xmath93 play role of the adiabatic parameter . from the definition ( [ alpha ] )
we see that @xmath94 is proportional to the adiabatic parameter and the limit @xmath95 is the adiabatic limit .
according to the asymptotic behavior of the transition points , the transition point from the first quadrant logarithmically coalesce with their counterpart in the second quadrant.and approach the imaginary axis . since in the limit @xmath91 , which is perfect adiabatic regime , we do not have a coherent contribution to the ddp formula ( [ dp - n ] ) from two transition points , no oscillations will be seen .
this simple analysis shows that even in the first adiabatic basis , the correct asymptotic behaviors of the optimized adiabatic passage is achieved . because for the gaussian model ( [ model ] ) there are infinitely many transition points , the most accurate transition probability is expected to be given by the generalized ddp formula ( [ dp - n ] ) . the dominant contributions to the sum in this formula originate from the two transition points closest to the real axis , @xmath96 and @xmath97 . for simplicity
, we neglect the contributions from all others and retain only the terms from these two points . because @xmath98 and
because @xmath99 is an even function of time , it is easy to show that @xmath100 that is @xmath101 and @xmath102 . hence it is sufficient to calculate only one of these integrals and we choose @xmath103 for this purpose . because the imaginary part of the ddp integral @xmath104 is the same for the two transition points @xmath97 and @xmath96 [ cf .
( d - d ) ] , these points lie on the same stokes line , defined by eq .
( stokes line ) .
this stokes line extends from @xmath105 to @xmath106 , which is a necessary condition for the validity of the ddp approximation davis76,joye91 . with the arguments presented above ,
the problem is reduced to the calculation of the ddp integral @xmath107 the estimation of this integral will be our main concern hereafter in this section . [
[ asymptotic - behavior - of - the - ddp - integral - for - small - protectalpha ] ] asymptotic behavior of the ddp integral for small @xmath108 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + for small @xmath90 ( @xmath88 ) we expand the integrand in eq .
( d(tc ) ) by using the taylor expansion , and perform term - by - term integration .
this integration is justified within the circle @xmath109 , where the series ( [ sqrt(1+x ) ] ) is uniformly convergent .
we choose the path of integration to be the straight line from @xmath110 to @xmath111 and parameterize this path as @xmath112 ( @xmath113 ) .
it is easy to see that @xmath114 along this path . indeed , @xmath115because @xmath116 and @xmath113 . by using the relation
@xmath117we find that @xmath118 , \label{d - small - series}\end{aligned}\ ] ] it is shown in @xcite , one can derive derive a uniform approximation to the ddp integral @xmath119 , eq .
( d(tc ) ) , by choosing an appropriate integration contour .
[ [ imaginary - part - of - the - ddp - integral ] ] imaginary part of the ddp integral + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + for the imaginary part of the ddp integral ( [ d(tc ) ] ) we have @xmath120 and @xmath121 .
the advantage of this choice is that the approximation ( [ imd ] ) , besides providing the exact result for @xmath122 is also very accurate in some vicinity of this important point . on the other hand , eq .
( [ imd ] ) has the following asymptotics [ imd - limit ] @xmath123these expressions agree with eqs .
( [ imd - small ] ) and ( [ imd - large ] ) , except for the factor @xmath124 , which is insignificant in the limits @xmath125 and @xmath88 [ since @xmath126 for @xmath127 and similarly for @xmath88 ] .
this factor becomes significant for intermediate @xmath90 , where , however , the accuracy of eq .
( [ imd ] ) improves until , as explained above , it becomes exact for @xmath129 .
[ [ real - part - of - the - ddp - integral ] ] real part of the ddp integral + + + + + + + + + + + + + + + + + + + + + + + + + + + + + for the real part of the ddp integral ( [ d(tc ) ] ) we have @xmath130 .\ ] ] the integral @xmath131 is approximated as @xmath132 ^{2}-1}}.\]]and @xmath133the second integral @xmath134.is approximated as @xmath135 ^{2}+\pi ^{2}}+\ln \frac{\alpha ^{2}}{\mu ( 2-\mu ) } } , \]]where @xmath136 , and @xmath137 in order to sum the contributions from various ddp integrals we need the factors @xmath54 , eq . ( [ gamma - k ] ) .
one finds after simple algebra that @xmath138 collecting the results we find @xmath139 \sin ^{2}\left [ { \text{re}}\mathcal{d}(\tau _ { 0}^{+})\right ] .
\label{p - ddp}\]]in @xcite , is shown that including the contributions from all transition points one can verify the following expresion for the transition probability @xmath140 } { \cosh ^{2}\left [ { \text{im}}\mathcal{d}(\tau _ { 0}^{+})\right ] } .
\label{p - ddp - sech}\ ] ] for four values of the detuning , @xmath141 .
the exact results obtained by numerical integration of the schrdinger equation are shown by dots and the approximation by solid lines.,width=283 ] equation ( [ p - ddp - sech ] ) provides a very accurate description of the transition probability @xmath38 .
this approximation is plotted on fig .
[ fig - rabi ] as a function of the peak rabi frequency @xmath142 for four different values of the detuning @xmath143 . as @xmath142 increases , rabi - like oscillations
are observed , with both amplitude and frequency matched very well by our approximation ( [ p - ddp - sech ] ) .
the ddp based approximation for the transition probability can be derived in the case of absent transition points by using the same ddp technique but in the first adiabatic basis .
a very reasonable question is how we can derived transition probability whenever a small deviation from optimized pulses takes place.insted of the optimized pulses given by eq.([erf ] ) we introduce @xmath144 , \qquad \omega ( t)=\omega _ { 0}\cos \left [ \frac{\pi } { 2}\text{% erf}(t / t)\right ] \label{erf - mu}\ ] ] where @xmath145 stands for small parameter .
if we apply ddp for such model , for @xmath146 as have been shown ddp fails to describe transition probability .
this is true not only for @xmath146 but for some vicinity of this point . as a function of rabi frequency @xmath65 .
black curve shows the pulses given in eq.([erf ] ) , red curve shows pulses given with eq.(erf - mu ) for @xmath147,width=283 ] figure [ probgr ] displays in log scale 1 - transition probability @xmath148 i.e. @xmath149 for the adiabatic optimizad pulses given in eq.([erf ] ) and pulses given with eq.(erf - mu ) as a function of peak rabi frequency @xmath65 .
we have examined the optimization via ddp method in the superadibatic basis .
according to the ddp method models that do not have transition points would lead to vanishing nonadiabatic corrections .
this would be the essence of the optimization based on ddp technique .
we have shown that this condition is not sufficient for perfect adiabtic optimization .
this is due to the approximate origin of the ddp method itself .
nevertheless ddp has been derived in order to take into account higher order adiabatic corrections , even for models that should yield according to ddp perfect adiabatic evolution(i.e .
there are no transition points ) within the the next adiabatic basis consecutive optimization is possible
. z. kis and f. renzoni , phys .
rev . a * 65 * , 032318 ( 2002 ) ; x. lacour , s. gurin , n. v. vitanov , l. p. yatsenko and h. r. jauslin , opt . commun . * 264 * , 362 ( 2006 ) ; c. wunderlich , t. hannemann , t. krber , h. hffner , c. roos , w. hnsel , r. blatt and f. schmidt- kaler , j. mod
. opt . * 54 * , 1541 ( 2007 ) .
p. w. shor , 37th symposium on foundations of computing 5665 ( ieee computer society press , washington dc , 1996 ) ; a. steane , rep .
. phys . * 61 * , 117 ( 1998 ) ; e. knill , nature * 434 * , 39 ( 2005 ) ; j. benhelm , g. kirchmair , c. f. roos and r. blatt , nature phys . * 4 * , 463 ( 2008 ) . | this paper we propose how to apply the dykhne - davis - pechukas ( ddp ) method for optimization of adiabatic passage in a two - state system in the second adiabtic basis . |
a _ quandle _ ( sometimes called a _ distributive groupoid_)@xcite is a non - associative algebraic structure which is particularly connected to the study of classical knots . in particular
a quandle may be defined as a set @xmath0 together with an left - invertible binary operation which we will write by exponentiation @xmath1 obeying the following relations : @xmath2 @xmath3 note that left invertibility is equivalent to also requiring that the equation @xmath4 with variable @xmath5 should have exactly one solution .
these relations are analogues of the reidemeister moves for classical knots . in particular , eq . [ q1 ] reflects the first reidemeister move , left invertibility reflects the second , and eq .
[ q2 ] reflects the third .
for this reason , they are particularly suited to the study of knots .
joyce@xcite showed in particular that there is a quandle naturally associated to any knot in @xmath6 which is a nearly complete ( up to mirror - reversal ) invariant of the knot .
the quandle associated to a classical knot is a special case of joyce s quandle associated to any codimension two knotting .
we briefly review this construction here , referring the reader to joyce s exposition for further details .
+ let @xmath7 be a path - connected manifold , and let @xmath8 be a codimension two knotting ( neither @xmath7 nor @xmath8 need to be spheres ) .
let @xmath9 be a tubular neighborhood of @xmath8 .
then @xmath10 is a manifold with boundary .
one component of @xmath11 will be @xmath12 .
let @xmath13 be the subgroup of @xmath14 which is the image of the homomorphism induced by the injection @xmath15 .
we refer to this as the _ peripheral _ subgroup .
note that @xmath13 is defined only up to conjugation ; any of its conjugates are also peripheral . +
now suppose that in @xmath13 there is a preferred element @xmath16 , the _
meridian_. in particular for any codimension two knot , the meridian is canonically defined up to conjugation .
we may now define the quandle @xmath0 associated to the pair @xmath17 and the chosen meridian .
the elements of @xmath0 are homotopy classes of paths that start at the basepoint of @xmath7 ( that is , the basepoint chosen for the fundamental group ) and end in @xmath12 , where homotopies are required to preserve these two conditions .
the quandle operation on two elements of @xmath18 is defined geometrically as illustrated in fig .
in particular , @xmath19^{[q']}$ ] is the homotopy class of paths containing the path @xmath20 , where the overbar indicates that the path is followed in reverse .
it is a striaghtforward exercise to show that this is well - defined and meets the definition of a quandle operation . , while the black circles represent points on @xmath12.[qop ] ] note that there is an action @xmath14 action on @xmath18 : @xmath19g$ ] is the equivalence class including the homotopy class of the path @xmath21 .
we follow the notation of joyce for this action .
given a group @xmath22 , a subgroup @xmath13 , and an element @xmath16 in the center @xmath23 , joyce defines a quandle as follows .
the underlying set is the set of right cosets of @xmath13 , @xmath24 .
we define the quandle operation such that @xmath25 .
this is well - defined , because if @xmath26 , then @xmath27 for some @xmath28 .
then @xmath29 where the last equality follows from @xmath30 .
we will denote a quandle constructed in this way by @xmath31 , or simply @xmath24 when there is a canonical choice for @xmath16 ( as there is when @xmath16 is a meridian and @xmath13 a peripheral subgroup ) .
the work in this section is largely a generalization of the work of joyce@xcite for the case of classical knots .
+ consider a codimension two pair @xmath32 , and the tubular neighborhood @xmath9 of @xmath8 .
then we can define the quandle @xmath33 geometrically as in the first section . on the other hand we can define the quandle @xmath22 to be the quandle @xmath34 for p some arbitrarily chosen peripheral subgroup of @xmath35 , and @xmath16 the meridian in @xmath13 .
+ now @xmath36 acts on @xmath0 by setting @xmath19g$ ] to be the equivalence class including the homotopy class of the path @xmath21 .
@xmath16 is an element of the fundamental group , but we may also define an analogous element @xmath37 in the fundamental quandle to be the quandle element such that @xmath38 is the meridian of @xmath13 . every element @xmath39 may be obtained as @xmath40 for some element @xmath41.[transq ] * proof : * we choose representatives of @xmath42 whose endpoints on @xmath43 agree .
then the path @xmath44 represents the same quandle class as @xmath45 .
however , the path @xmath46 starts and ends at the basepoint , and hence represents some element of the fundamental group .
thus ( abusing the notational difference between paths and their equivalence classes ) @xmath47.@xmath48 + this shows in addition that the group action on @xmath0 is transitive .
@xmath49 iff @xmath50 for some @xmath28 .
* proof : * if @xmath50 then it is straightforward to see that @xmath49 . on the other hand
if @xmath49 , then there is a map of the disk into @xmath7 as shown in fig .
[ tri ] .
to @xmath51 with the endpoint of the path tracing out a path @xmath52 on @xmath43.[tri ] ] but this disk shows that the path which follows @xmath53 is contractible .
however , @xmath54 is in the peripheral group , and we have our result.@xmath48 + a similar argument shows that the stabilizer of @xmath40 is @xmath55 . for elements in that conjugate peripheral group are easily seen to stabilize @xmath40 . in particular if @xmath56 , then @xmath57 . on the other hand if some some element of the fundamental group stabilizes @xmath40 we may build another homotopy disk as in the previous lemma .
as the stabilizer does not play a role in our following argument we leave the details of the construction as an exercise .
+ let us now turn our attention to the quandle @xmath22 which is built from the fundamental group and the peripheral information .
our goal is to establish the following result : [ main ] @xmath58 as quandles ; hence the quandle can be recovered from the fundamental group and peripheral information .
we begin by noting that the underlying set of @xmath22 is just the set of right cosets of @xmath59 .
therefore @xmath59 naturally acts on @xmath22 , and this action is obviously transitive .
indeed , every element of @xmath22 is of the form @xmath60 for some @xmath61 .
furthermore , @xmath62 iff @xmath50 for some @xmath28 .
+ we now define a quandle morphism @xmath63 sending @xmath64 .
this is well defined and invertible by our previous lemmas , so we need only show that it is a quandle map as well .
now @xmath65 this is mapped to @xmath66 which is precisely @xmath67 . that the inverse map is also
a morphism of quandles follows similarly .
therefore this is an invertible quandle morphism , and hence an isomorphism of quandles , establishing thm .
joyce has also proved a theorem for classical knots stating that the triple @xmath68 can be reconstructed from @xmath69 .
this is because @xmath59 is isomorphic to @xmath70 , the free group on @xmath69 quotiented by taking all the relations on @xmath69 and treating the quandle operation as conjugation in the group .
we write an element @xmath39 as @xmath71 when we wish to designate them as elements of @xmath70 .
note that @xmath70 is canonically isomorphic to @xmath59 ( up to a choice of basepoint ) , by sending @xmath71 to the homotopy class of a wirtinger generator , @xmath72 $ ] .
then in @xmath70 one may take @xmath16 to be any of the elements of the generators defined .
@xmath70 acts on @xmath0 by @xmath73 .
if an element of @xmath70 has two presentations , @xmath74 and @xmath75 , this implies that those products of wirtinger generators are homotopic rel basepoints , and this homotopy passes to a homotopy of @xmath76 to @xmath77 .
it is then straightforward to check that this action of @xmath70 is just the geometrically defined action of @xmath59 on @xmath0 .
then @xmath13 will be the stabilizer of @xmath78 under this action , and so the triple can be recovered from the quandle .
+ however , this only permits us to recover the triple when all the relations on @xmath79 can be written down in terms of conjugation
. if @xmath79 has no such presentation , then this will not be possible .
checking this condition is complicated , however , by the fact that @xmath0 will generally be infinite , and so the presentation we obtain will not be a finite presentation .
nonetheless , for higher - dimensional knots in spheres , the above argument shows : for codimension two knots in spheres of any dimension , the quandle and the triple @xmath68 can both be calculated from the other .
more generally this holds for codimension two knots in simply connected spaces.[hid ] thm .
[ main ] does imply that any invariant based upon quandles for any knot should be interpretable as an invariant based upon classical knot invariants .
thus , for example , the quandle cocycle invariants@xcite are determined by the triple @xmath68 , even for higher - dimensional knots .
this becomes even more powerful when it is recalled that the peripheral group of a sphere is cyclic , and hence is determined by @xmath16 . on the other hand , since in general @xmath80 for some normal subgroup @xmath81 , this indicates that for higher - dimensional knots which as manifolds have large fundamental groups , the quandle has the potential to capture more information than for those with smaller fundamental groups .
+ eisermann@xcite has in fact shown that the quandle cocycle invariants of classical knots is a specialization of certain colourings of their fundamental groups .
his construction makes use of the full peripheral structure of the classical knot ( that is , the longitude and the meridian ) , and hence does not generalize immediately to higher dimensions .
however , in light of our result here , we pose the question of whether a similar construction might not be possible in higher dimensions . + as a last application of our result , we give the following theorem .
recall that _
virtual _ knots are a combinatorial generalization of knot diagrams introduced by kauffman@xcite , and welded knots are a quotient of virtual knots given by adding an additional permitted move , first explored for braids in @xcite . when the diagrams are allowed to have two endpoints instead of being closed we obtain virtual ( welded ) _
_ arcs__@xcite .
* proof : * satoh@xcite has defined a map @xmath83 on welded knots and arcs ( hence on virtual knots and arcs ) which maps each knot or arc to a surface knot , and shown that @xmath83 preserves the quandle and fundamental group .
in addition , a direct calculation shows that @xmath13 and @xmath16 are preserved by @xmath83@xcite .
but the latter determine the quandle for the surface knot , and hence also for the virtual or welded knot or arc .
+ to see that the quandle determines the triple @xmath82 , it suffices to check this for surface knots , as was done in thm .
[ hid].@xmath48 50 j.s .
carter , s. kamada , and m. saito , _ surfaces in 4-space _ , springer ( 2004 ) .
m. eisermann , _ knot colouring polynomials _ , pacific journal of mathematics * 231 * 2 ( 2007 ) 305 - 336 .
r. fenn , r. rimanyi , c. rourke , _ the braid - permutation group _ , topology * 36 * ( 1997 ) 123 - 135 .
l. h. kauffman , _ virtual knot theory _
j. combinatorics * 20 * ( 1999 ) 663 - 691 .
d. joyce , _ a classifying invariant of knots , the knot quandle _
, journal of pure and applied algebra * 23 * ( 1982 ) 37 - 65 .
v. manturov , _ knot theory _ , chapman and hall / crc crc press ( 2004 ) .
s. matveev , s.v .
_ distributivnye gruppoidy v teorii uzlov _ ,
sbornik * 119 * 1 ( 1982 ) 78 - 88 ( in russian ) .
english version : _ distributive grouppoids in knot theory _ ,
sbornik * 47 * 73 - 83 ( 1984 ) .
s. satoh , _ virtual knot presentation of ribbon torus - knots _ , j. knot theory ramifications * 9 * ( 2000 ) , 531542 .
b. winter , _ on the classification problem for ribbon torus knots _ , to appear in j. knot theory ramifications , preprint arxiv:0711.1638 [ math.gt ] . | joyce has shown that the fundamental quandle of a classical knot can be derived from consideration of the fundamental group and the peripheral structure of the knot , and also that the group and much of the peripheral structure can be recovered from the quandle .
we generalize these results to arbitrary dimensions , and also to virtual and welded knots and arcs . |
given a piecewise monotonic map @xmath0 of an interval @xmath1 , one measure of the ` complexity ' of the map is the topological entropy @xmath2 , given by the exponential growth rate of the number of monotonicity intervals of iterates of the map .
it turns out that varying the map may produce discontinuities of the topological entropy . as a simple ,
yet illustrative example , consider for @xmath3 , the scaled farey map @xmath4{\rightarrow}[0,1]$ ] given by @xmath5$},\\ a\frac{1-x}{x } & \text{for $ x\in ( 1/2,1]$}. \end{cases}\ ] ] it is not difficult to see that if @xmath6 then @xmath7 .
for @xmath8 the map has a closed invariant set on which it is topologically conjugate to the tent map , so @xmath9 .
thus , as @xmath10 passes a half , the sudden appearance of a full tent map produces a discontinuity of the topological entropy . on the other hand
, it is known that for certain one - parameter families of smooth maps the topological entropy is even hlder continuous ( see , for example , @xcite ) .
a somewhat milder way of changing the map is to fix the map itself but to change the interval of definition , that is to introduce ` holes ' in the map . in dynamical terms , if a point in the domain of definition hits the hole under iteration of the map , it is not allowed to evolve any further .
the motivational idea is a physical system for which mass leaks through an opening in phase space .
this set - up has received considerable attention over the past decades ( see @xcite for a snapshot of recent developments ) .
most of the efforts have been directed to studying the escape rate of mass through the hole as well as showing the existence of conditionally invariant measures with good statistical properties for various classes of maps , as well as conditions on the hole , typically that the hole is either part of a markov partition for the map ( see , for example , @xcite ) or that the hole is ` small ' ( see , for example , @xcite ) . in a similar vein
, the behaviour of the escape rate as the size of the hole shrinks to zero has also been studied in some detail ( see , for example , @xcite ) . in this article , we shall focus on the regularity of the topological entropy @xmath11 of a fixed piecewise monotonic expanding interval map @xmath12 as a function of the hole @xmath13 , without any assumptions on the size or position of @xmath13 . roughly speaking
we shall show that if @xmath14 is a neighbourhood of @xmath15 and @xmath16 is a family of holes varying sufficiently regularly in the space of holes , then , under a mild expansivity condition on @xmath0 , the function @xmath17 is continuous at @xmath18 .
moreover , if @xmath19 , then @xmath17 is even hlder continuous at @xmath18 with hlder exponent at @xmath18 depending on @xmath20 . in order to provide a more precise formulation of these results , we require some more notation . [ def : map ]
let @xmath1 be a compact interval and let @xmath21 be a finite collection of disjoint open subintervals of @xmath1 .
we say that @xmath22 is a _ piecewise monotonic @xmath23-map _ with _ initial partition @xmath21 _
if for each @xmath24 the restriction @xmath25 of @xmath0 to @xmath26 is strictly monotonic and extends as a @xmath23-function to the closure @xmath27 of @xmath26 .
suppose now that we are given a piecewise monotonic @xmath23-map @xmath0 with initial partition @xmath21 .
for @xmath28 we write @xmath29 for the collection of _ cylinder sets of level @xmath30 _ and observe that @xmath31 is well - defined and strictly monotonic on each element of @xmath32 . in order to avoid trivialities
we shall assume throughout this article that @xmath32 is non - empty for all @xmath30 .
let @xmath33 denote the derivative of @xmath31 and @xmath34 lebesgue measure on @xmath1 .
we set @xmath35 and define @xmath36 the number @xmath37 quantifies expansivity of the map . in the particular case where @xmath0 is markov - map @xmath22 is said to be _ markov _ if for every @xmath24 the closure of its image @xmath38 is a union of closures of elements of the initial partition @xmath21 . ]
, it follows that @xmath39 is uniformly bounded from below , so the number @xmath37 measures the growth rate of the maximal expansion of @xmath31 .
let us now make the notion of a hole more precise .
for @xmath40 we write @xmath41 for the collection of sets @xmath42 which may be written as a union of at most @xmath43 closed subintervals of @xmath1 .
we then define @xmath44 to be our ` space of holes ' and equip it with the following distance function @xmath45 note that this distance is a pseudometric , that is , it is a _ bona fide _ metric , except that the distance between two holes may vanish without the holes coinciding .
note also that it does not distinguish the number of intervals constituting a hole . given a map @xmath22 and a hole @xmath13 as above
, we will be interested in the dynamics of the map restricted to the complement of @xmath13 .
more precisely , we consider @xmath0 on the initial partition @xmath46 and let @xmath47 denote the corresponding collection of cylinder sets of level @xmath30 . in passing ,
we note that elements of @xmath48 are not necessarily intervals in the usual sense .
the ( non - negative ) becomes eventually trivial .
here , we choose to set @xmath49 in this case , as this assures continuity of the topological entropy , as we shall see in the following . ]
topological entropy of the restricted map is then given by the exponential growth rate of the number of elements in @xmath48 @xmath50 where @xmath51 .
by @xcite , this coincides with other definitions of the topological entropy .
associated with the restricted map we also have a transfer operator acting on @xmath52 , the space of functions of bounded variation on @xmath1 , @xmath53 it turns out that if @xmath54 then @xmath55 is an isolated eigenvalue ( in fact the largest in modulus ) of @xmath56 , and hence also a pole of the resolvent of @xmath56 . as we shall see
shortly , the order of this pole plays a role for the degree of regularity of the topological entropy of @xmath0 at @xmath13 . our main result on the regularity of the topological entropy as a function of the hole , to be proved in section [ sec : htop ] ,
can now be formulated as follows .
[ theorem : main ] suppose that @xmath22 is a piecewise monotonic @xmath23-map with @xmath57 .
let @xmath14 be a neighbourhood of @xmath15 and suppose that @xmath58 is a family of holes which is lipschitz continuous in @xmath59 and for which the number of holes is uniformly bounded .
then @xmath60 is continuous at @xmath18 .
furthermore , if @xmath61 then @xmath60 is hlder continuous at @xmath18 with local hlder exponent ( see definition [ def : holder ] ) satisfying @xmath62 where @xmath63 is the order of the pole of the resolvent of @xmath64 at @xmath65 .
it is rather curious that the local hlder exponent of the topological entropy at a particular hole depends itself on the value of the topological entropy at that hole .
a similar behaviour of the topological entropy , albeit not for maps with holes , but for one - parameter deformations of the tent map , has already been conjectured by isola and politi in the early nineties ( see @xcite ) .
our numerical simulations also indicate that the factor @xmath63 can not be omitted in the above formula .
the theorem above is in fact a special case of a more general theorem with similar hypotheses and similar conclusions , but with the behaviour of the topological entropy as a function of the hole replaced by the behaviour of the topological pressure as a function of the potential ( see corollary [ coro : main ] ) .
the proof of the theorem above and its generalisation relies on a spectral perturbation theorem of keller and liverani @xcite , which has opened up a rich seam of applications in various areas .
in fact , our proof relies on a refinement of the keller - liverani theorem ( see corollary [ coro : ourcoro ] ) which is of interest in its own right and elucidates the role of the pole of the resolvent of the perturbed operator . as an illustration of the theorem above we consider the topological entropy of the doubling map @xmath66 on the unit interval with a uniformly left - expanding hole , that is , we are interested in the regularity of the function @xmath67)\,.\ ] ] it is not difficult to see that @xmath68 , so the first part of the theorem above immediately implies that @xmath69 is continuous , thus giving a new proof of a result originally due to urbaski ( *
* theorem 1 ) .
moreover , as we shall show in section [ sec : ex ] , the order of the pole of @xmath70}$ ] at @xmath71 is one for each @xmath72 , so @xmath69 is hlder continuous at each @xmath72 with local hlder exponent satisfying @xmath73 the above lower bound for the local hlder exponent was recently obtained by carminati and tiozzo @xcite , using a different approach of a combinatorial flavour , which , as a bonus , also yields that there is equality in ( [ eq : ct ] ) for all @xmath72 at which @xmath69 is not locally constant .
in fact , they are able to show that the local hlder exponent of the topological entropy of the @xmath74-adic map @xmath75 with @xmath76 with uniformly left - expanding hole equals the topological entropy divided by @xmath77 for all points at which the topological entropy is not locally constant .
this paper is organised as follows . in section [ sec : setup ]
we consider piecewise monotonic interval maps and study the corresponding transfer operators with general weights on spaces of functions of bounded variation
. the main goal will be to obtain a lasota - yorke inequality for the transfer operator ( see proposition [ prop : lyi ] ) , which is sufficiently uniform in order for the keller - liverani theorem to apply .
section [ sec : kl ] is devoted to proving our refinement of the keller - liverani theorem ( see corollary [ coro : ourcoro ] ) .
in the following section [ sec : holder ] , we define the notion of the pressure @xmath78 of a piecewise monotonic interval map @xmath0 with a given weight @xmath79 in a form suitable for our applications and show its equivalence to other definitions ( see theorem [ thm : bkpressure ] ) .
we then go on to prove our main result on the regularity of @xmath78 as a function of @xmath79 ( see corollary [ coro : main ] ) . in section [ sec : htop ]
we specialise the results of the previous section to prove our theorem [ theorem : main ] on the regularity of the topological entropy as a function of the hole . in section
[ sec : ex ] we apply the results of the previous section to the doubling map with left - expanding hole , for which we are able to show that the order of the pole is always equal to one .
we also consider the doubling map with a sliding hole of fixed size , showing that a double pole can suddenly occur .
let @xmath80 be a piecewise monotonic @xmath23-map .
given a bounded function @xmath81 we define the _ ruelle transfer operator of @xmath0 with weight @xmath79 _ by @xmath82 for any bounded @xmath83 . here
, @xmath84 denotes the inverse of the restriction of @xmath0 to @xmath26 .
it is not difficult to see that the @xmath30-th power of @xmath85 can be written @xmath86 where @xmath87 denotes the inverse of @xmath31 restricted to @xmath88 and @xmath89 is given by @xmath90 we shall see presently that , for suitable @xmath79 , the transfer operator @xmath85 has nice spectral properties on the space @xmath91 of functions of bounded variation , the definition of which we briefly recall . let @xmath34 denote lebesgue measure on @xmath92 . for @xmath93
measurable we write @xmath94 to denote the banach space of ( equivalence classes ) of @xmath34-integrable functions on @xmath95 with the usual norm .
suppose now that @xmath96 is a bounded interval .
if @xmath97 is an ordinary function we write @xmath98 for the total variation of @xmath99 , while for @xmath100 the total variation is given by @xmath101 the space of functions of bounded variation is now defined by @xmath102 when equipped with the norm @xmath103 it becomes a banach space , which is compactly embedded in @xmath104 ( see , for example ( * ? ? ?
* lemma 5 ) or ( * ? ? ? * theorem 1.19 ) ) . for other characterisations of this space ,
see , for example @xcite or ( * ? ? ?
* theorem 2.3.12 ) .
we briefly mention @xmath91 spaces over more general sets , which are defined as follows .
let @xmath105 be a finite collection of mutually disjoint open intervals and @xmath106 .
for @xmath107 , we let @xmath108 and define @xmath109 it turns out that , for suitable @xmath79 , the transfer operator @xmath85 of a piecewise monotonic @xmath23 map @xmath22 is a continuous endomorphism of @xmath52 with discrete peripheral spectrum ( see @xcite ) .
moreover , we shall see that the peripheral spectrum is stable under suitable perturbations of @xmath79 .
this will follow from an application of the keller - liverani perturbation theorem @xcite . in order to apply it we require a lasota - yorke inequality for @xmath85 , which is suitably uniform in @xmath79 .
in our derivation we shall follow the approach in @xcite , where a @xmath91-@xmath110 lasota - yorke inequality is proved . for reasons that will become apparent later on
, we require a @xmath91-@xmath111 lasota - yorke inequality , similar to the one by rychlik @xcite , but with more explicit control of the coefficients .
we start by quickly summarising some useful properties of variation . in order to declutter notation it will be useful to introduce the following shorthand : for @xmath112 a real interval and @xmath113 we write @xmath114 for its @xmath110-norm .
[ lem : varlem ] let @xmath112 and @xmath115 be bounded non - empty intervals and let @xmath116 . a. [ varlem : a ] if @xmath117 , then @xmath118 b. [ varlem : b ] if @xmath117 , then @xmath119 c. [ varlem : c ] if @xmath120 is a homeomorphism , then @xmath121 d. [ varlem : d ] we have @xmath122 e. [ varlem : e ] if @xmath123 , then @xmath124 f. [ varlem : f ] if @xmath125 are subintervals of @xmath112 with mutually disjoint interiors then @xmath126 these are all well - known , apart , perhaps , from ( e ) , which is proved as follows . if @xmath127 is a version of @xmath128 and @xmath129 we have @xmath130 so @xmath131 from which the assertion follows .
we shall now derive a number of auxiliary results , including a uniform lasota - yorke inequality for the action of the transfer operator on functions of bounded variation . here and in the following we use @xmath33 to denote the derivative of @xmath31 .
[ lem : varlg ] [ lem : intlg ] let @xmath22 be a piecewise monotonic @xmath23-map and let @xmath132 .
for @xmath28 and @xmath88 write @xmath133 a''_n(z)&= { { \left | { g_n}\right | } _ { z,\infty}}{{\left | { d(t^n)}\right | } _ { z,\infty}}\ , .
\label{eq : aanz}\end{aligned}\ ] ] then for any @xmath134 we have @xmath135 using ( [ varlem : b ] ) , ( [ varlem : c ] ) , and ( [ varlem : e ] ) of lemma [ lem : varlem ] we have @xmath136 but , by ( [ varlem : d ] ) of lemma [ lem : varlem ] we have @xmath137 and the first assertion follows by observing that @xmath138 for the second we use a change of variables formula to obtain @xmath139 we are now able to deduce the following @xmath91-@xmath111 lasota - yorke inequality . [ prop : lyi ] let @xmath22 be a piecewise monotonic @xmath23-map and let @xmath140
. then we have for any @xmath141 @xmath142 @xmath143 where , using the notation from the previous lemma , @xmath144 for any @xmath134 and @xmath88 , we have by lemma [ lem : intlg ] @xmath145 so @xmath146 which proves ( [ eq : l1b ] ) .
similarly , @xmath147 so , using ( [ varlem : f ] ) of lemma [ lem : varlem ] , @xmath148 which proves ( [ eq : lyi ] ) . in the following we need a control on the growth of the above coefficients @xmath149 and @xmath150 which is uniform in @xmath79 .
in order to achieve this we will use the following notation . define for @xmath141
( recalling that @xmath151 by our standing assumption ) : @xmath152 and let @xmath153 where , as before , @xmath33 denotes the derivative of @xmath31 . 1 .
the first two limits can be shown to exist by a simple sub - additivity argument .
a short calculation shows that @xmath154 .
3 . the quantity @xmath155 measures the growth rate of the maximal expansion of @xmath31 .
although it is easier to calculate than @xmath37 it will not play any further role in our results .
we note , however , that it is closely related to @xmath37 , as we shall see presently . since @xmath156 we obviously have @xmath157 .
there is equality when @xmath0 has _
big images _ , that is , if @xmath158 note that any piecewise monotonic @xmath23-map which is markov has big images .
the following lemma establishes uniform bounds for the constants occurring in the lasota - yorke inequality .
[ lem : vargn ] suppose that @xmath22 is a piecewise monotonic @xmath23-map and that @xmath132 .
let @xmath159 and define the quantities ( tacitly assuming that @xmath160 ) @xmath161 then for @xmath28 and @xmath88 @xmath162 the first equation is obviously a consequence of ( [ m - bound ] ) .
let @xmath28 and @xmath88 with @xmath163 for some @xmath164 . then , using ( [ varlem : e ] ) of lemma [ lem : varlem ] , we have @xmath165 but @xmath166 , so @xmath167 and the assertion follows . the following estimate will provide a perturbative bound when varying the weight .
[ lem : lg_kl3 ] let @xmath22 be a piecewise monotonic @xmath23-map .
then for any @xmath168 and any @xmath134 we have @xmath169 using the change of variables formula we have , for @xmath24 , @xmath170 but @xmath171 and by ( [ varlem : d ] ) of lemma [ lem : varlem ] @xmath172 so the inequality follows since @xmath173
let @xmath174 be a banach space and @xmath175 a bounded linear operator . in the following ,
we denote by @xmath176 the _ resolvent set _ of @xmath177 @xmath178 and by @xmath179 the _ spectrum _ of @xmath177 .
recall that the _ discrete spectrum _ @xmath180 consists of all isolated eigenvalues of @xmath177 with finite algebraic multiplicity , and that its complement in @xmath181 is known as the _ ( browder ) essential spectrum _ @xmath182 .
spectral radius _ of @xmath177 will be denoted by @xmath183 and the _ essential spectral radius _ of @xmath177 by @xmath184 , that is , @xmath185 finally , given @xmath186 and @xmath187 we write @xmath188 and @xmath189 the keller - liverani theorem applies in the following situation .
suppose that the banach space @xmath174 is equipped with a second norm @xmath190 , which is weaker than the original norm in the sense that @xmath191 for every @xmath192 .
given a bounded linear operator @xmath175 we write @xmath193 for the norm of @xmath177 considered as an operator from @xmath174 to @xmath194 . furthermore , suppose that @xmath195 $ ] is a set of parameter values where @xmath196 is a positive constant and that we are given a family @xmath197 of bounded linear operators on @xmath174 satisfying the following three properties .
* there are constants @xmath198 and @xmath199 such that for every @xmath200 @xmath201 and @xmath202 * for every @xmath200 , if @xmath203 and @xmath204 then @xmath205 does not belong to the residual spectrum of @xmath206 .
* there is a monotonic upper - semicontinuous function @xmath207 with @xmath208 and @xmath209 [ rem : itm ] note that condition ( kl2 ) follows from condition ( kl1 ) provided that the embedding @xmath210 is compact , since then @xmath211 by a theorem of ionescu tulcea and marinescu ( see @xcite for the original paper or @xcite for a more contemporary exposition ) .
the keller - liverani theorem can now be stated as follows .
[ thm : kl ] suppose that @xmath197 is a family of bounded linear operators satisfying conditions ( kl1 ) , ( kl2 ) , and ( kl3 ) above .
fix @xmath212 and @xmath186 . then there is a constant @xmath213 such that @xmath214)\,.\ ] ] moreover , if @xmath215 is a compact subset of @xmath216 , then there is a constant @xmath217 such that @xmath218)\,,\ ] ] where @xmath219 see ( * ? ? ?
* theorem 1 ) .
suppose now that @xmath220 with @xmath221 is an isolated point of the spectrum of @xmath222 .
fixing @xmath223 with @xmath224 we can choose @xmath186 so that @xmath225 , where @xmath226 denotes the closed disk with radius @xmath227 and centre @xmath205 .
the keller - liverani theorem now implies that for all @xmath228 sufficiently small , the positively - oriented boundary @xmath229 belongs to @xmath230 implying that the spectral projections @xmath231 exist for all @xmath228 sufficiently small ( e.g. for all @xmath232 $ ] ) .
in fact , more is true .
[ coro : klcoro ] suppose that @xmath197 is a family of bounded linear operators satisfying conditions ( kl1 ) , ( kl2 ) , and ( kl3 ) above and suppose that @xmath205 is an isolated spectral point of @xmath222 with @xmath221 . fix @xmath223 with @xmath224 .
then there is a constant @xmath233 such that for every @xmath234 $ ] there are constants @xmath235 and @xmath236 with the following properties : @xmath237)\,,\ ] ] @xmath238)\,,\ ] ] @xmath239)\,.\ ] ] see ( * ? ? ?
* corollary 1 ) . for the applications we have in mind we require the following refinement of the previous results , providing a quantitative bound for the behaviour of a peripheral eigenvalue @xmath240 of @xmath222 under perturbations which are small in the norm @xmath241 .
it turns out that the behaviour is governed by the size of the largest jordan block of @xmath222 corresponding to @xmath240 , or , equivalently , by the order of the pole of the resolvent of @xmath222 at @xmath240 , just as in standard perturbation theory ( see , for example , ( * ? ? ?
* theorem 6.7 ) ) .
[ coro : ourcoro ] suppose that @xmath197 is a family of bounded linear operators satisfying conditions ( kl1 ) , ( kl2 ) , and ( kl3 ) above and suppose that @xmath205 with @xmath221 is a pole of the resolvent of @xmath222 of order @xmath63 .
fix @xmath223 with @xmath224 .
then there is a constant @xmath242 such that for every @xmath243 $ ] there are constants @xmath244 and @xmath245 with the following properties : @xmath246 ) \,,\ ] ] @xmath247)\,,\ ] ] where @xmath219 fix @xmath223 with @xmath224 .
choose @xmath248 so that @xmath249
. now fix @xmath250 .
write @xmath251 and let @xmath252 $ ] .
the first assertion now follows , since we have @xmath253 by ( [ kl : coro3 ] ) of the previous corollary . in order to prove the remaining assertion
write @xmath254 \,dz\,,\ ] ] and observe that @xmath255 is a bounded linear operator on @xmath256 . in order to conclude the proof we shall bound @xmath257 from above and below .
we start with the upper bound . by theorem [ thm : kl ]
we have with @xmath258 @xmath259 for the lower bound , we first use the fact that @xmath205 is a pole of order @xmath63 of @xmath260 together with analytic functional calculus for @xmath206 ( see , for example , ( * ? ? ?
* theorem vii.3.10 ) ) to conclude that @xmath261 let now @xmath262 .
then there is a non - zero @xmath263 with @xmath264 and @xmath265 .
thus @xmath266^pf ' = ( { \lambda}'-{\lambda})^pf ' = ( { \lambda}'-{\lambda})^pp_{\epsilon}^{({\lambda},\delta)}f'\,.\ ] ] since @xmath267 we may , with no loss of generality , assume that @xmath268 .
then , using the previous equation together with ( [ kl : coro2 ] ) of the previous corollary , we have @xmath269 combining ( [ kl : ourcoro : upper ] ) and ( [ kl : ourcoro : lower ] ) we have @xmath270 and ( [ kl : ourcoro3 ] ) follows by setting @xmath271 .
in this section we prove our main abstract result on the regularity of the pressure .
before we start with a definition of pressure suitable for our purposes , we require the following notation . for @xmath112 a finite union of open subsets of @xmath112
we write @xmath272 for the space of real non - negative weights on @xmath112 of bounded variation .
suppose that @xmath22 is a piecewise monotonic @xmath23-map .
let @xmath273 and let @xmath274 denote the corresponding transfer operator .
we then call @xmath275 the _ pressure of @xmath0 with weight @xmath79_. an alternative expression for the pressure is given in the following theorem .
[ thm : bkpressure ] let @xmath22 be a piecewise monotonic @xmath23-map and @xmath276 . if either 1 .
@xmath277 is continuous for each @xmath24 and @xmath21 is generating , or 2 .
@xmath277 is constant for each @xmath24 ( @xmath21 need not be generating ) then @xmath278 the first case is proved as in ( * ? ? ?
* theorem 3 ) . in the second case
we note that @xmath279 is constant for each @xmath280 .
the set @xmath281 is a subset of the so - called hofbauer tower so has cardinality @xmath282 ( see , for example , ( * ? ? ?
* section 3 ) ) . for each @xmath283 , pick @xmath284 . then @xmath285 so @xmath286 .
it follows that @xmath287 .
the reverse inequality follows as in the proof of ( * ? ? ?
* theorem 3 ) .
before we continue we summarise some consequences of the results in section [ sec : setup ] .
we start with the following well - known facts already contained in @xcite , the simple proofs of which we give for the convenience of the reader .
[ lem : ress ] let @xmath22 be a piecewise monotonic @xmath23-map and @xmath132 . then @xmath288 fix @xmath289 .
proposition [ prop : lyi ] and lemma [ lem : vargn ] imply that the transfer operator @xmath85 satisfies a lasota - yorke inequality of the form @xmath290 with @xmath291 since the embedding @xmath292 is compact , it follows , for example by @xcite , that @xmath293 .
since @xmath289 was arbitrary , the assertion is proved .
let @xmath22 be a piecewise monotonic @xmath23-map and @xmath273 .
if @xmath294 , then @xmath295 is a pole of the resolvent of @xmath85 and in particular an eigenvalue of @xmath85 .
this follows from a positivity argument .
consider the real banach space @xmath296 and observe that @xmath297 is a total cone , that is , a cone the span of which is dense in @xmath298 . since @xmath299 the transfer operator leaves this cone invariant .
moreover , since @xmath294 , the previous lemma implies that @xmath300 , so the resolvent of @xmath85 must have a pole on the circle centred at @xmath301 with radius @xmath302 . using ( * ? ? ?
* appendix 2.4 ) ) it now follows that @xmath302 is a pole of the resolvent of @xmath85 on @xmath298 and the assertion of the lemma is proved since the complex banach space @xmath52 is equivalent to any standard complexification of @xmath298 .
this also follows from ( * ? ? ?
* theorem 2 ) . the above proof , however , is more direct .
[ def : holder ] when @xmath303 and @xmath304 are metric spaces , a function @xmath305 is said to be _ hlder continuous at @xmath306 _ if there is @xmath307 $ ] such that @xmath308 in case such an @xmath309 exists , we call @xmath310 } \ , : \ , { \limsup_{s\to t } \frac { d_y(h(s),h(t))}{d_x(s , t)^\alpha}<\infty } \ , \right \ } } \ ] ] the _ local hlder exponent of @xmath69 at @xmath18_. in the following we consider a fixed piecewise monotonic @xmath23-map @xmath22 and a family @xmath311 of weights
. we shall be interested in the regularity of @xmath312 .
the following is our abstract main result .
[ thm : main ] suppose that @xmath22 is a piecewise monotonic @xmath23-map with @xmath57 .
let @xmath15 and let @xmath14 be a neighbourhood of @xmath18 .
suppose that @xmath311 is a family of non - negative weights on @xmath313 of bounded variation and that there is @xmath314 with the following properties ( see lemma [ lem : vargn ] for definitions ) : * @xmath315 and @xmath316 ; + * @xmath317 whenever @xmath318 ; + * @xmath319 , where @xmath320 is a monotonic upper - semicontinuous function defined in a neighbourhood of @xmath301 with @xmath321 . if @xmath322 , then for every @xmath323 there is a constant @xmath324 and a neighbourhood @xmath325 of @xmath18 such that @xmath326 where @xmath63 is the order of the pole of the resolvent of @xmath327 at @xmath328 . + fix @xmath329 .
we may choose constants satisfying @xmath330 and @xmath331 such that @xmath332 .
now fix @xmath333 with @xmath334 . by possibly shrinking @xmath14 we may , by ( i ) and ( ii ) , assume that @xmath335 and @xmath336 .
we thus have @xmath337 moreover , by lemma [ lem : vargn ] we have @xmath338 for all @xmath339 and all @xmath141 , @xmath88 . as @xmath340
we may absorb the factor @xmath30 and conclude that there is a constant @xmath341 not depending on @xmath59 such that @xmath342 furthermore , by the definition of @xmath37 there is a constant @xmath343 such that @xmath344 hence we also have @xmath345 we can thus bound the constants @xmath149 and @xmath150 occurring in proposition [ prop : lyi ] by @xmath346 where @xmath347 and @xmath348 , which shows that condition ( kl1 ) of the keller - liverani theorem is satisfied with @xmath349 and @xmath350 for the family @xmath351 with @xmath352 a small neighbourhood of @xmath301 .
condition ( kl2 ) follows from remark [ rem : itm ] together with the compactness of the embedding @xmath353 . by lemma [ lem : lg_kl3 ]
, there is a constant @xmath354 such that @xmath355 so condition ( kl3 ) of the keller - liverani theorem is also satisfied . since @xmath356 is the largest real eigenvalue of @xmath357 we have @xmath358 . in order to see this , fix @xmath359 and let @xmath360 .
observing that by ( [ kl : eq : spec ] ) of theorem [ thm : kl ] we have @xmath361 whenever @xmath362 as @xmath363 is contained in a @xmath227-neighbourhood of @xmath364 .
similarly , by ( [ kl : coro3 ] ) of corollary [ coro : klcoro ] we have @xmath365 for @xmath366 since there exists an eigenvalue of @xmath357 in a @xmath227-neighbourhood of @xmath367 . to conclude the proof we again set @xmath368 and observe that , by corollary [ coro : ourcoro ] , we can choose a @xmath186 and a neighbourhood @xmath325 of @xmath18 so that @xmath369 for all @xmath370 and @xmath371 where @xmath372 and @xmath373 is the constant in corollary [ coro : ourcoro ] .
[ coro : main ] under the hypotheses of the previous theorem : a. [ coro : main : a ] if @xmath374 is continuous at @xmath18 , then so is the map @xmath375 ; b. [ coro : main : b ] if @xmath376 and @xmath374 is lipschitz continuous at @xmath18 then @xmath377 is hlder continuous at @xmath18 with local hlder exponent satisfying @xmath378 where @xmath63 is the order of the pole of the resolvent of @xmath327 at @xmath328 . part ( b ) follows immediately from ( [ holderineq ] ) in our previous theorem by taking suitable limits .
in order to show ( a ) it suffices to prove that if @xmath379 then also @xmath380 .
hence , assume that @xmath379 and let @xmath381 . the spectrum of @xmath382 is contained in the disk @xmath383 . choosing @xmath384 and @xmath385 so that @xmath386 in theorem [ thm : kl ] , inclusion ( [ kl : eq : spec ] ) implies that if @xmath362 then @xmath363 is contained in the disk @xmath387 .
so @xmath388 for such @xmath59-values and the claim follows .
we shall now apply the results of the previous section to interval maps with holes and , in particular , prove theorem [ theorem : main ] from the introduction . in the following ,
we let @xmath389 be a fixed piecewise monotonic @xmath23-map .
the space of holes equipped with a pseudometric was defined in the introduction , see equations ( [ def : holespace ] ) and ( [ def : holedist ] ) .
first we note that the relation between the pressure and the ( non - negative ) topological entropy of the dynamical system with holes is as follows .
. then @xmath391 .
given the weight @xmath392 , it is clear that @xmath393 takes values one or zero only . for @xmath88
, we have @xmath394 precisely when @xmath26 contains a non - empty cylinder for the restricted dynamics , say @xmath395 .
in fact , the support of @xmath393 restricted to @xmath26 is the closure of @xmath396 .
thus @xmath397 and by theorem [ thm : bkpressure ] it follows that when @xmath48 is non - empty for every @xmath28 @xmath398 and the assertion follows in this case . if on the other hand @xmath48 is empty for some @xmath28 , then @xmath399 is empty for all @xmath400 , hence @xmath49 .
moreover , @xmath401 for all @xmath400 , so @xmath402 and the assertion follows in this case as well .
[ proof of theorem [ theorem : main ] : ] since the weight @xmath403 takes the values 0 and 1 only , we observe that @xmath404 @xmath405 so the family of weights is @xmath111-lipschitz continuous in @xmath59 . if the number of holes is uniformly bounded by @xmath406 and the cardinality of @xmath21 is @xmath74 then the variation norm of @xmath407 is uniformly bounded by @xmath408 .
we therefore have @xmath409 whenever @xmath410 and @xmath411 , both uniformly in @xmath59 .
the conditions in theorem [ thm : main ] are thus satisfied with @xmath412 .
the conclusions follow by applying corollary [ coro : main ] to @xmath413 .
this result is perhaps a bit surprising , given the rather brutal nature of introducing a hole into the dynamics .
note that @xmath414 may drop discontinuously from @xmath301 to @xmath415 , which happens when all points escape .
this is the reason for defining the topological entropy as in the introduction .
$ ] as a function of @xmath10 .
right : estimated local hlder constant @xmath416 as a function of @xmath10 assuming @xmath417 using different mesh - sizes.,title="fig : " ] $ ] as a function of @xmath10 .
right : estimated local hlder constant @xmath416 as a function of @xmath10 assuming @xmath417 using different mesh - sizes.,title="fig : " ] in this section we will illustrate and compare hlder continuity of families of interval maps numerically and theoretically .
a non - trivial difficulty in applying our theorems numerically is that an explicit value for the hlder exponent requires knowledge of the order @xmath63 of the pole associated with the leading eigenvalue . in our first example below
we exhibit a family in which the leading eigenvalue is simple for all parameter values , which in particular implies that @xmath417 .
we then provide an example exhibiting a double pole and show how this is reflected in numerical experiments .
we consider the map @xmath418 , defined on @xmath419 .
. we will be interested in the topological entropy of @xmath421)$ ] , that is , @xmath0 restricted to the set @xmath422 .
as shown in the previous section we may calculate this entropy as the leading eigenvalue of a transfer operator .
there are essentially two ways to do so .
the first is to consider the map @xmath0 as defined on @xmath423 , and using as weight @xmath79 the indicator function of @xmath95 .
the transfer operator then takes the form @xmath424 we note that the map has big images as @xmath425 for every @xmath28 , @xmath280 .
we also have @xmath426
. we may apply theorem [ theorem : main ] to see that @xmath69 is hlder continuous at every @xmath72 with the local hlder exponent satisfying @xmath427 for some @xmath428 .
we claim that @xmath417 for each @xmath72 .
this is also indicated by numerical experiments illustrated in figure [ fig:1 ] . in that figure
we show our numerical estimates of @xmath421)$ ] on the left , and on the right the local hlder constant under the assumption that @xmath417 given by @xmath429 the ( numerically obtained ) limit superior seems to be neither 0 nor infinity , indicating that our estimate for the local hlder exponent is _
optimal_. now , in order to prove that @xmath417 it suffices to show that the leading eigenvalue is simple .
we shall do so by considering an alternative description of the system , using a hofbauer tower ( @xcite , see also ( * ? ? ?
* section 3 ) ) , which we also used to compute our numerical estimates .
we start out by writing @xmath430 , @xmath431 , and @xmath432 . set @xmath433 and consider the locally constant weight @xmath434 .
for simplicity of notation we shall in the following tacitly omit intervals arising from @xmath435 from the hofbauer tower as they will not contribute to the spectral properties of the transfer operator on the tower . instead of working with cylinder sets , that is , intersections of preimages of the @xmath436 s , the idea is to deal with intersections of forward iterates of these intervals . in our case
, this boils down to studying the forward orbit of the point @xmath10 . to this end
, we define the sequence @xmath437 by setting @xmath438 and @xmath439 $ ] for @xmath440 , where @xmath441 for @xmath442 $ ] .
the sequence contains finitely many values if either @xmath443 for some @xmath444 , which happens when the orbit of @xmath10 ` escapes ' , or if @xmath10 is pre - periodic for @xmath0 .
otherwise the orbit is infinite .
let us first consider the case when the orbit is infinite , so in particular we have @xmath445 and @xmath446 for all @xmath447 .
we set @xmath448 and define the ` flats ' @xmath449 for @xmath447 .
we have the following ` transition ' rules giving rise to a transition matrix @xmath450 ( elements other than those mentioned all being zero ) : @xmath451 note that the above lines describe two cases : if @xmath452 , then there is exactly one transition , namely from @xmath453 to @xmath454 ; if @xmath455 , then two different transitions are possible , namely the one from @xmath453 to @xmath454 and an additional one from @xmath453 to the bottom flat @xmath456 . for later use ,
let @xmath457 denote the set of indices for which there is a transition from the @xmath444-th flat to the bottom flat .
note that @xmath458 and that @xmath95 contains at least one more index , the latter being a consequence of the fact that @xmath0 is expanding .
the hofbauer tower is now defined as the disjoint union of flats @xmath459 with the tower we associate a directed graph @xmath460 obtained from the transition matrix .
a possible transition graph is sketched in figure [ fig : graph ] . in the case of @xmath461 by ( * ? ? ?
* theorem 2 and lemma 4.1 ) , the multiplicity of the leading eigenvalue @xmath240 of the transfer operator @xmath462 equals the order of the leading pole @xmath463 of the zeta - function associated with the tower , which is given by @xmath464 where @xmath465 is the total number of periodic points in @xmath460 of period @xmath444 ( not necessarily prime ) .
now , by hofbauer ( * ? ? ?
* theorem 1 as well as lemmas 3 and 4 ) this periodic orbit counting zeta - function has the same poles in the unit disk as the zeros of a determinant calculated in the following way : call @xmath466 a simple cycle if @xmath467 and all the @xmath468 s are distinct . also denote by @xmath469 the length of such a cycle .
then @xmath470 where the sum is over all @xmath471-tuples ( with @xmath472 ) of disjoint simple cycles of @xmath460 .
when a transition matrix @xmath450 is of finite size @xmath473 the formula follows easily from the standard expansion of @xmath474 in terms of permutations and rewriting permutations as products over distinct cycles . in the case of unbounded matrix size
we refer to @xcite which explains how to take a limit of finite matrix truncations , that is , levels in the hofbauer tower . combining the above two results we have :
let @xmath475 with @xmath476 .
then @xmath477 is a zero of @xmath74 iff @xmath478 is an eigenvalue of the transfer operator .
moreover , if this is the case the order of the zero equals the ( algebraic ) multiplicity of the eigenvalue .
[ thm : hofbauer ] in order to apply the theorem above to the present situation we requite a simple lemma on the nature of the zeros closest to the origin of certain power series .
[ lem : dlem ] let @xmath479 denote a sequence with @xmath480 such that @xmath481 for at least one @xmath447
. then @xmath482 is holomorphic in the open unit disk .
moreover , there is a unique @xmath483 which is a simple zero of @xmath227 and all other zeros of @xmath227 are strictly larger in modulus . by the cauchy - hadamard theorem ,
the function @xmath227 is holomorphic in the open unit disk .
furthermore , we have @xmath484 with equality iff @xmath477 is real and positive .
since @xmath485 and @xmath486 for @xmath477 real and sufficiently close to @xmath487 there is @xmath483 with @xmath488 .
moreover , since @xmath489 when @xmath490 , any other zero must be of absolute value strictly larger than @xmath223 , and noting that @xmath491 the order of the zero must be one . returning to our specific case
when the orbit of @xmath10 is infinite we obtain the following .
suppose that the orbit @xmath492 is infinite .
then the hofbauer determinant ( [ eq : hofbauer ] ) is given by @xmath493 and is holomorphic in the open unit disk .
there is a unique @xmath483 which is a simple zero of @xmath74 and all other zeros of @xmath74 are of strictly larger modulus .
the spectral radius of the transfer operator equals @xmath494 and is a simple eigenvalue .
all other spectral values are strictly smaller in modulus than @xmath494 .
cycles are to be distinct in the above sum and _ any _ cycle in our transition graph has to contain @xmath456 so there are no two distinct non - trivial cycles .
since every cycle is of the form @xmath495 with @xmath496 we immediately obtain the expression for the hofbauer determinant .
the remaining assertions now follow from theorem [ thm : hofbauer ] and lemma [ lem : dlem ] by observing that @xmath497 , that @xmath458 and that @xmath498 is non - empty .
we now turn to the case where the hofbauer tower is finite .
recall that this happens precisely if @xmath10 is pre - periodic for @xmath0 or if the orbit escapes .
in either case there are @xmath43 and @xmath499 with @xmath500 ; note that if the orbit escapes we have @xmath501 .
thus there is no element @xmath502 but we have the additional transition @xmath503 .
we then have the following simple cycles : 1 . for @xmath504 with @xmath496 ,
@xmath505 is of length @xmath506 .
2 . one additional cycle of the form @xmath507 and of length @xmath508 . in this case
there is also the possibility of disjoint @xmath509-tuples for each @xmath510 with @xmath511 of the form @xmath512 the hofbauer determinant ( [ eq : hofbauer ] ) then takes the form @xmath513 where @xmath514 we may thus conclude : suppose that the orbit @xmath492 is finite .
then the hofbauer determinant ( [ eq : hofbauer ] ) is given by @xmath515 for some @xmath43 and @xmath516 with @xmath499 and @xmath517 gven by ( [ eq : tildem ] ) .
the function @xmath74 is holomorphic in the open unit disk .
moreover , there is a unique @xmath483 which is a simple zero of @xmath74 and all other zeros of @xmath74 are of strictly larger modulus .
the spectral radius of the transfer operator equals @xmath494 and is a simple eigenvalue .
all other spectral values are strictly smaller in modulus than @xmath494 .
this follows from the calculation above together with theorem [ thm : hofbauer ] and lemma [ lem : dlem ] .
summarising , we have shown the following .
for any @xmath72 the transfer operator @xmath462 has a leading eigenvalue which is simple .
in particular , the topological entropy @xmath518 of the doubling map with hole @xmath519 $ ] satisfies @xmath520 one of the dynamically simplest examples exhibiting a double pole of the resolvent is again furnished by the doubling map @xmath66 .
this time , however , we introduce a hole from @xmath10 to @xmath521 , that is , we shall consider @xmath522)$ ] as a function of @xmath10 . for the particular value @xmath523
the hole is from @xmath524 to @xmath525 .
figure [ fig:2 ] shows this map indicating orbits of boundary points .
$ ] showing orbits of the hole boundaries . ] the estimated entropy @xmath522)$ ] as a function of @xmath10 is shown in figure [ fig:3 ] on the left , exhibiting a ` dip ' at @xmath523 . in order to understand this dip , note that the intervals @xmath526 , @xmath527 , and @xmath528 make up the initial partition . introducing @xmath529 and @xmath530
we have the following transitions ( see figure [ fig : double ] for the corresponding transition graph ) : @xmath531 in the double pole case ] one verifies that @xmath532 is a double eigenvalue of the corresponding transition matrix @xmath533 ( see figure [ fig:3 ] right ) and that @xmath534 is one - dimensional .
this implies that the resolvent has a double pole at @xmath535 .
theorem [ theorem : main ] applied at @xmath523 shows that the entropy @xmath536)$ ] is hlder continuous with a local exponent at least @xmath537 which is consistent with a numerical estimate for the exponent ( not shown ) . )
$ ] as a function of @xmath10 .
note the dip for @xmath523 .
right : the two largest eigenvalues of the transfer operator.,title="fig : " ] ) $ ] as a function of @xmath10 . note the dip for @xmath523 .
right : the two largest eigenvalues of the transfer operator.,title="fig : " ]
the research in this article was carried out while the first author was visiting the dpartement de mathmatiques d at the universit paris - sud during research leave from queen mary , university of london .
both authors are grateful to viviane baladi , gerhard keller and henk bruin for valuable feedback during the preparation of this article , as well as to an anonymous referee , whose comments led to a considerable simplification of the arguments in section [ sec : ex ] .
j. guckenheimer . the growth of topological entropy for one - dimensional maps . in z.
nitecki and c. robinson , _ global theory of dynamical systems ( proc .
conf . , northwestern univ . ,
evanston , ill . , 1979 )
_ , lecture notes in mathematics 819 , berlin , springer , 1980 , pp .
216223 . c. liverani and v. maume - deschamps .
lasota - yorke maps with holes : conditionally invariant probability measures and invariant probability measures on the survivor set .
. h. poincar probab .
statist . _
* 39 * ( 2003 ) 385412 . | we study the dependence of the topological entropy of piecewise monotonic maps with holes under perturbations , for example sliding a hole of fixed size at uniform speed or expanding a hole at a uniform rate .
we show that under suitable conditions the topological entropy varies locally hlder continuously with the local hlder exponent depending itself on the value of the topological entropy . |
here we formulate our main result in the context of measurable spaces .
two main cases are of interest : when the space of ontic states @xmath18 is discrete , and when it is absolutely continuous . the discrete case was treated in the main text .
it has the advantage of notational simplicity , but is probably difficult to justify as it is known that reproducing the predictions of quantum theory requires that the space of ontic states @xmath18 be continuous @xcite .
the absolutely continuous case is more natural , but the notation and proofs are slightly more cumbersome . in the following we first introduce a general measure theoretic framework for ontic states , and
then restrict to the absolutely continuous case .
the state space @xmath18 is assumed to be a ( probability ) measure space with a fixed @xmath103-algebra @xmath104 of events .
we assume the existence of a reference measure @xmath105 on @xmath18 . for a measurable set @xmath106 , @xmath107 , where @xmath108 is the indicator function for @xmath109 ( 1 on @xmath109 and 0 elsewhere ) .
each preparation @xmath2 induces a conditional probability measure @xmath110 defined on @xmath104 . from the lebesgue decomposition theorem @xcite
we infer that the measure @xmath110 can be decomposed as @xmath111 where the measures @xmath112 and @xmath113 are respectively singular and absolutely continuous with respect to @xmath105 .
recall that a measure @xmath114 is concentrated on a set @xmath115 if for any measurable set @xmath116 , @xmath117 , and that two measures are singular with respect to each other if they are concentrated on disjoint sets .
furthermore the singular part @xmath112 can be decomposed into a discrete singular part which consists of a sum of dirac measures ( `` @xmath17 functions '' ) and a singular continuous part .
in the remainder we consider the case where @xmath110 only consists of an absolutely continuous part .
this is the assumption that is generally made in the literature .
the case where there is only a discrete singular part was discussed in the main text .
note that our results can be extended to the case where either the absolutely continuous or the discrete singular part are non - zero .
our results do not apply to the case where there is only a singular continuous part .
we denote by @xmath118 the set on which @xmath105 is concentrated . the fact that the measure @xmath110 is absolutely continuous with respect to @xmath105 ( @xmath119 whenever @xmath120 ) implies that there is an @xmath105-measurable function @xmath121 on @xmath18 such that for all @xmath122 @xmath123 in integration theory the function @xmath121 is called the radon - nikodym derivative @xcite .
we further suppose that the probability of obtaining result @xmath4 given measurement @xmath5 and ontic state @xmath1 is given by an @xmath105-measurable function @xmath124 .
then we have @xmath125 we can define the support @xmath126 of the function @xmath121 as the complement of the set on which it vanishes . since @xmath127 there will be no loss of generality if we restrict the support to @xmath118 .
( note that if the measure is not absolutely continuous , we can not introduce in general the concept of `` support '' of the measures @xmath110 as it requires a topology on @xmath18 which we do not assume ) .
given the above framework , we now state and prove the results presented in the main text in the case of absolutely continuous measures .
this includes the case where @xmath128 with @xmath105 the standard measure on @xmath129 .
a statistical model is @xmath0-epistemic if there exist two preparation procedures @xmath130 and @xmath131 giving two distinct quantum states such that @xmath132 .
a @xmath0-epistemic model is @xmath17-continuous if , given the preparation q of the pure quantum state @xmath8 , there is a set @xmath133 ( which can depend on @xmath2 ) such that for all states @xmath134 satisfying @xmath135 , @xmath136 and @xmath137 . note that this implies that @xmath138 .
a model is continuous , if it is @xmath17-continuous for some @xmath27 .
why is @xmath17-continuity a `` natural '' condition ?
informally the reason is that except in rare occasions physical quantities depend continuously on their parameters .
discontinuities are generally attributable to mathematical idealisation . to illustrate this in the present case ,
suppose that we can write @xmath139 , where @xmath8 is the pure state produced by @xmath2 , and @xmath12 are additional variables specifying the preparation .
furthermore suppose that @xmath140 all belong to compact subsets of real euclidean space @xmath129 , and that the probability distribution @xmath141 is a continuous function of all its parameters .
then it is uniformly continuous , and hence @xmath142continuous for some @xmath27 .
let the dimension of the quantum state space be @xmath16 .
we consider the same @xmath16 independent non - orthogonal states @xmath143 as in the main text .
suppose the model is @xmath17-continuous for @xmath36 .
then there exists a set @xmath144 for all @xmath37 with @xmath138 .
define a new function @xmath145 .
then we have @xmath146 it remains to show that @xmath147 is strictly positive . to this end
we note that for each function @xmath148 there must be a positive integer @xmath149 such that the set @xmath150 satisfies @xmath151 .
otherwise @xmath152 , contrary to our hypothesis .
let @xmath153 .
then for the set @xmath154 we have @xmath155 for all @xmath156 and hence @xmath157 .
we conclude that @xmath158 we thus arrive at the contradiction @xmath159 .
consider a preparation @xmath2 of a physical system and any subset of ontic states @xmath160 that occurs with non - zero probability @xmath161 .
consider @xmath53 identical copies of the preparation devices , which when used yield preparation @xmath54 .
then the product set @xmath162 occurs with non - zero probability : @xmath163 .
note that according to this definition the space of ontic states for @xmath53 independent preparations may be larger than the product @xmath164 .
this does not affect the definition of separability .
the proof of theorem 2 in the case of absolutely continuous measures is a simple combination of the arguments given in the main text in the case of discrete singular measures , and the proof of theorem 1 for the case of absolutely continuous measures .
hardy recently proposed a no - go theorem for @xmath0-epistemic models for a single system that satisfies _ ontic indifference _ , or _ restricted ontic indifference _ ( see definitions below ) @xcite . here
we show that ontic indifference is an extremely strong condition for @xmath0-epistemic models , since it implies that there are ontic states @xmath1 that are shared by all quantum states in the hilbert space .
it is therefore immediately in contradiction with the predictions of quantum theory since this implies that orthogonal quantum states have overlapping ontic distributions .
restricted ontic indifference does not suffer from this problem , but on the other hand it breaks the unitary invariance of hilbert space by promoting one state to a specific status .
we show how the proof methods we use to rule out @xmath17-continuous @xmath0-epistemic models can be easily transposed to provide no - go results for @xmath0-epistemic models obeying restricted ontic indifference . any quantum transformation on a system which leaves unchanged any given pure state , @xmath165 , can be performed in such a way that it does not affect the underlying ontic states @xmath166 in the ontic support of that pure state . by the ontic support of a given state , @xmath165 ,
one means the set @xmath167 of ontic states @xmath1 which might be prepared when the given pure state is prepared .
ontic indifference implies the following fact : if two distinct states @xmath168 share the same ontic state @xmath1 , that is @xmath166 and @xmath169 , and if @xmath73 is a unitary transformation that leaves @xmath165 invariant , @xmath170 , then @xmath171 belongs to the ontic support of @xmath172 .
this follows from considering the following 4 preparations : a specific preparation @xmath173 of @xmath165 , the preparation @xmath173 followed by unitary @xmath73 ( that leaves @xmath165 invariant ) , a specific preparation @xmath174 of @xmath175 , and the preparation @xmath174 followed by unitary @xmath73 ( which prepares @xmath172 ) .
then we note that preparation @xmath173 sometimes yields the ontic state @xmath1 , that @xmath73 leaves @xmath1 invariant , that preparation @xmath174 sometimes yields the ontic state @xmath1 , hence preparation @xmath174 followed by @xmath73 sometimes yields the ontic state @xmath1 , and therefore @xmath171 .
this observation has the following consequence , which implies directly that no @xmath0-epistemic model can reproduce the predictions of quantum theory ( since it implies in particular that orthogonal quantum states share a common real state @xmath1 ) . if any two distinct states @xmath176 share the same ontic state @xmath1 , that is @xmath177 and @xmath178 , and if ontic indifference holds , then all states in the hilbert space share the ontic state @xmath1 , that is @xmath166 for all @xmath165 .
to prove this consider the following sets @xmath179 ( @xmath180 ) of states , where @xmath181 and @xmath182 there exist @xmath183 and a unitary @xmath73 such that @xmath184 , @xmath185 .
that is , @xmath179 is the set of states that can be obtained from @xmath186 by a unitary that leaves one of the states in @xmath186 invariant . by hypothesis , both states in @xmath187 share the same ontic state @xmath1 .
hence by the above observation , all states in @xmath188 for all @xmath189 , also share the state @xmath1 . now consider the case of a 2-dimensional hilbert space . if @xmath190 are at an angle @xmath191 from each other ( as measured on the bloch sphere ) , then it is easy to show ( just draw the successive sets @xmath192 that the size of @xmath179 increases exponentially , such that when @xmath193 , @xmath179 comprises the whole bloch sphere .
the case of a 2-dimensional hilbert space helps to visualize what is happening , but is not a restriction , since the linear span of the set @xmath194 is the whole hilbert space . any quantum transformation on a system which leaves unchanged the pure state @xmath165 can be performed in such a way that it does not affect the underlying ontic states , @xmath166 , in the ontic support of that pure state .
the quantum predictions are incompatible with @xmath0-epistemic models satisfying the assumptions of restricted ontic indifference and separability , and such that there exists a state @xmath195 sharing a common ontic state @xmath1 with @xmath29 , where @xmath29 is the particular state in the definition of restricted ontic indifference .
let @xmath196 be an integer such that @xmath197 .
add an ancilla @xmath198 to the space such that the dimension of the full hilbert space is greater or equal to @xmath16 . by the assumption of separability the states @xmath199 and @xmath200 share a common ontic state @xmath1 .
choose a basis such that @xmath201 , and coefficients @xmath202 , @xmath203 such that @xmath204 , @xmath205 , @xmath206 ( this is always possible since @xmath207 ) .
define the states @xmath208 .
( note that these states coincide with the states @xmath209 used in the main text if @xmath210 , @xmath211 ) .
the observation above implies that all the states @xmath212 share the ontic state @xmath1 , since @xmath213 . we then obtain a contradiction with the predictions of quantum theory exactly as in the proof of theorem 1 given in the main text .
m. g. a. paris and j. rehacek ( eds . ) , lect .
notes phys .
* 649 * , 1 , springer ( 2004 ) .
y. aharonov and l. vaidman , , phys .
a * 178 * , 38 ( 1993 ) . c. m. caves , c. a. fuchs , and r. schack , phys .
a * 65 * , 022305 ( 2002 ) . c. brukner and a. zeilinger , in `` time , quantum , information '' , edited by l. castell and o. ischebeck , springer ( 2003 ) .
r. w. spekkens , phys .
a * 75 * , 032110 ( 2007 ) .
n. harrigan and r. w. spekkens , found .
* 40 * , 125 ( 2010 ) .
m. leifer , the quantum times * 6*(3 ) ( 2011 ) .
p. g. lewis , d. jennings , j. barrett , and t. rudolph , phys . rev . lett . * 109 * , 150404 ( 2012 ) . m. f. pusey , j. barrett , and t. rudolph , nat . phys .
* 8 * , 476 ( 2012 ) .
m. k. patra , l. olislager , f. duport , j. safioui , s. pironio , s. massar , arxiv:1306.0414 ( 2013 ) .
o. j. maroney , how statistical are quantum states ? .
arxiv preprint arxiv:1207.6906 ( 2012 ) s. m. leifer and o. j. maroney , arxiv preprint arxiv:1208.5132 ( 2012 ) s. aaronson , a. bouland , l. chua , g. lowther , arxiv:1303.2834 ( 2013 ) l. hardy , international journal of modern physics b , 27 , no 01n03 ( 2013 ) l. hardy , studies in history and philosophy of science part b : studies in history and philosophy of modern physics * 35 * , 267 ( 2004 ) .
a. montina , phys .
a * 77 * , 022104 ( 2008 ) .
a. montina , phys .
a * 83 * , 032107 ( 2011 ) . | the quantum state @xmath0 is a mathematical object used to determine the probabilities of different outcomes when measuring a physical system .
its fundamental nature has been the subject of discussions since the inception of quantum theory : is it _
ontic _ , that is , does it correspond to a real property of the physical system ? or is it _ epistemic _ , that is , does it merely represent our knowledge about the system ? assuming a natural continuity assumption and a weak separability assumption , we show here that epistemic interpretations of the quantum state are in contradiction with quantum theory .
our argument is different from the recent proof of pusey , barrett , and rudolph and it already yields a non - trivial constraint on @xmath0-epistemic models using a single copy of the system in question . _
introduction .
_ quantum theory textbooks usually start from the hypothesis that to every physical system corresponds a mathematical object a ray in hilbert space called the _
quantum state_. they then go on to deduce the multitude of quantitative predictions that make quantum theory so successful .
but does the quantum state correspond to a real physical state or does it merely represent an observer s knowledge about the underlying reality ?
a major reason for doubting the reality of the quantum state is that it can not be observed directly : it can only be reconstructed indirectly by lengthy state estimation procedures @xcite . furthermore , an epistemic interpretation of the quantum state could provide an intuitive explanation for many counterintuitive quantum phenomena and paradoxes , such as the measurement postulate and wavefunction collapse @xcite . to formulate with precision the above question ,
we assume , following @xcite , that every quantum system possesses a real physical state ( also called ontic state ) , denoted @xmath1 , which is independent of the observer .
when a measurement is performed on the system , the probabilities to get different outcomes are determined by @xmath1 .
if an ensemble of such systems is prepared , different members of the ensemble may be found in different states @xmath1 .
a preparation procedure @xmath2 therefore corresponds in general to a probability distribution @xmath3 over the real states .
the probability to obtain the outcome @xmath4 when preparation @xmath2 is followed by measurement @xmath5 is @xmath6 .
such a model will reproduce the quantum predictions if @xmath7 , where @xmath8 is the quantum state assigned by quantum theory to the preparation @xmath2 and @xmath9 is the quantum operator describing the measurement .
we can now distinguish two classes of models of the above type .
a model is said to be @xmath0__-ontic _ _ if the preparation of distinct pure quantum states always give rise to distinct real states .
that is , for every @xmath1 either @xmath10 or @xmath11 if the preparations @xmath2 and @xmath12 correspond to different quantum states @xmath13 . in this case , every real state @xmath1 is compatible with a unique pure quantum state .
the quantum state is `` encoded '' in @xmath1 and we can consider it to represent a real property of the system , akin , e.g. , to the total energy of a system in classical physics @xcite . in the second class of models , known as @xmath0__-epistemic _ _ models , preparation of distinct pure quantum states may result in the same real state @xmath1 .
formally , there exists preparations @xmath2 and @xmath12 corresponding to distinct quantum states @xmath13 such that both @xmath14 and @xmath15 for some @xmath1 . in this case , the quantum state is not uniquely determined by the underlying real state and has a status analogous , e.g. , to the liouville distribution in statistical physics .
while non - trivial @xmath0-epistemic models exist in any fixed dimension @xmath16 @xcite , such models are necessarily highly contrived .
indeed , pusey , barrett , and rudolph ( pbr ) have recently shown that the predictions of @xmath0-epistemic models are in contradiction with quantum theory under the assumption , termed _ preparation independence _ , that independently prepared pure quantum states correspond to product distributions over ontic states @xcite . in the present work ,
we derive two alternative no - go theorems for @xmath0-epistemic models based on a natural assumption of continuity .
our approach shows that already at the level of a single system there exist strong constraints on @xmath0-epistemic models .
furthermore , our first no - go theorem readily translates in a simple experimental test , an implementation of which has been reported in @xcite using high - dimensional attenuated coherent states of light traveling in an optical fibre .
constraints on @xmath0-epistemic models at the level of single quantum systems have also been obtained in @xcite using an assumption termed _
ontic indifference_. this assumption is in fact closely related to the one presented here . in section 2 of the supplemental material @xcite , we show how to use our approach to recover , in a simple and clear way , those of @xcite .
arguments using single quantum systems have also been used in @xcite to show that @xmath0-epistemic model can not be maximally epistemic " ( in a sense defined in @xcite . _
no - go theorems for @xmath0-epistemic models .
_ the key motivation behind our result is that @xmath0-epistemic models should satisfy a form of continuity .
indeed , we assign an ontic status to @xmath0 if a variation of @xmath0 necessarily implies a variation of the underlying reality @xmath1 , and we assign it an epistemic status if a variation of @xmath0 does not necessarily imply a variation of @xmath1 .
it is then natural to assume a form of continuity for @xmath0-epistemic models : a slight change of @xmath0 induces a slight change in the corresponding ensemble of @xmath1 s in such a way that at least some @xmath1 s from the initial ensemble will also belong to the perturbed ensemble .
we use a slightly stronger form of continuity which asserts that there are real states @xmath1 in the initial ensemble that will remain part of the perturbed ensemble , no matter how we perturb the initial state , provided this perturbation is small enough .
models that violate this condition are presumably very contrived . formally this continuity condition is defined as follows ( see fig . [
fig : model ] for a depiction of the difference between @xmath0-ontic and @xmath17-continuous @xmath0-epistemic models ) .
-ontic and @xmath17-continuous @xmath0-epistemic models .
depicted is the space @xmath18 of ontic states , as well as the support of the probability distribution @xmath19 for preparation @xmath20 associated to distinct pure states @xmath21 , @xmath22 . in @xmath0-ontic models ( left ) distinct quantum states
give rise to probability distribution @xmath19 with no overlap .
in @xmath17-continuous @xmath0-epistemic models ( right ) , states that are close to each other ( such as @xmath23 and @xmath24 ) all share common ontic states .
however states that are further from each other ( such as @xmath25 and @xmath26 ) do not necessarily have common ontic states @xmath1 . ]
let @xmath27 and let @xmath28 be the ball of radius @xmath17 centered on @xmath29 , i.e. , @xmath28 is the set of states @xmath30 such that @xmath31 .
we say that a model is @xmath17-continuous if for any preparation @xmath2 , there exists an ontic state @xmath1 ( which can depend on @xmath2 ) such that for all preparations @xmath12 corresponding to quantum states @xmath32 in the ball @xmath33 centered on the state @xmath34 , we have @xmath15 . note that for notational simplicity we formulate our results in the case where the set @xmath35 of real states is finite or denumerable .
the generalisation of theorems 1 and 2 below to measurable spaces is given in section 1 of the supplemental material @xcite .
this generalisation is important since reproducing the predictions of even a single qubit requires an infinite , and probably even uncountably infinite , number of real states , see @xcite for evidence to this effect
. note also that the above definition introduces a connection between the overlap of quantum states and the overlap of distributions in the ontic space of @xmath1s .
this is extremely natural if we do not introduce a privileged direction in hilbert space ( i , e , a preferred basis ) , since then the properties of psi - epistemic models can only depend on the geometry of the hilbert space .
our first result is a constraint on @xmath17-continuous models for single systems .
there are no @xmath17-continuous models with @xmath36 reproducing the measurement statistics of quantum states in a hilbert space of dimension @xmath16 .
consider @xmath16 preparations @xmath20 ( @xmath37 ) corresponding to distinct quantum states @xmath38 all contained in a ball of radius @xmath17 .
by definition of a @xmath17-continuous model , there is at least one @xmath1 for which @xmath39 and thus @xmath40 this last quantity can be viewed as a measure of the extent to which distributions over real states overlap in the neighborhood of a given quantum state .
it was also introduced in @xcite where it was shown to be related to the variational distance between the distributions @xmath19 .
suppose now that a measurement @xmath5 yielding one of the possible outcomes @xmath41 is made on each of the prepared systems .
a @xmath17-continuous model then makes the prediction @xmath42 according to quantum theory , however , there exist states in a hilbert space of dimension @xmath16 contained in a ball of radius @xmath43 such that the left - hand side of eq .
( [ eq : nogo ] ) is equal to @xmath44 . to show this ,
let @xmath45 be a basis of the hilbert space .
consider the @xmath16 distinct states @xmath46 .
these states are all at mutual distance @xmath47 from the state @xmath48 .
let the measurement @xmath5 be the measurement in the basis @xmath49 .
then @xmath50 for all @xmath37 and thus @xmath51 .
( these states and measurements were considered in the @xmath52 case in @xcite ) .
note that the above result also applies if we only require @xmath17-continuity to hold around some fixed quantum states rather than for all states in hilbert space .
interestingly , the @xmath0-epistemic model of @xcite for hilbert spaces of dimension @xmath16 is @xmath17-continuous around a specific state with a value of @xmath17 saturating the above bound .
though one expects a @xmath0-epistemic model to be @xmath17-continuous for some value of @xmath17 , the bound derived in theorem 1 may a priori seem arbitrary .
this motivates the following definition .
a @xmath0-epistemic model is continuous if there exists a non - zero @xmath27 such that it is @xmath17-continuous .
our second result shows that there are no @xmath0-epistemic models that are both continuous and which satisfy the following separability assumption .
a similar condition was independently introduced in @xcite , where it is called compactness " .
though weaker than the preparation independence assumption explicitly used by pbr , it is already sufficient to derive their main result .
let @xmath2 be the preparation of a physical system yielding with non - zero probability @xmath14 the real state @xmath1 .
a model is separable if @xmath53 independent copies @xmath54 of the preparation devices yield with non - zero probability @xmath55 a system in the joint real state @xmath56 , for any positive integer @xmath53 .
separable continuous @xmath0-epistemic models can not reproduce the measurement statistics of quantum states in a hilbert space of dimension @xmath57 .
the idea of the proof is to fix an arbitrarily small @xmath27 , and choose specific states @xmath58 , all within the ball of radius @xmath17 . because of @xmath17-continuity
, these states share a common ontic state . using separability , the states @xmath59 also share a common ontic state . by taking @xmath53 large enough
, the distance between the tensor products @xmath59 becomes large enough that we can apply theorem 1 .
more in detail we proceed as follows .
consider @xmath57 preparations @xmath20 corresponding to the @xmath16 distinct states @xmath60 with @xmath37 , @xmath61^{1/n}}$ ] and @xmath62^{1/n}}$ ] .
it is easily checked that these states are normalised @xmath63 , have mutual scalar product @xmath64^{1/n}$ ] for @xmath65 , and are all at distance @xmath66 from the state @xmath67 , where @xmath68 . in a @xmath69-continuous model ,
these states share at least a common real state @xmath1 .
the separability assumption then implies that the states @xmath59 also share a common real state , and thus that @xmath70 . by the same argument as in theorem 1
, it then follows that if a measurement @xmath5 yielding one of the possible outcomes @xmath41 is performed on each of these systems , the quantity @xmath71 .
note now that the @xmath16 @xmath53-fold copies @xmath59 are normalised and have mutual scalar product @xmath72 for @xmath65 .
there therefore exists a unitary transformation @xmath73 in the subspace @xmath74 spanned by the @xmath16 states @xmath59 that carries out the transformation : @xmath75 where @xmath76 for some basis @xmath49 .
the states @xmath38 are identical to the states used in the proof of theorem 1 .
it follows that there exists a @xmath16-outcome measurement @xmath5 in @xmath77 which applied on the states @xmath59 gives the same statistics as the measurement in the basis @xmath49 applied on the states @xmath38 of theorem 1 .
we can therefore find a measurement such that @xmath78 and thus @xmath79 in contradiction with the prediction of a @xmath69-continuous separable @xmath0-epistemic model .
we have thus shown that one can exclude @xmath17-continuous separable models with @xmath80 for any positive integers @xmath16 and @xmath53 . for large @xmath53
this bound behaves as @xmath81 , with @xmath82/(2d)$ ] , thereby implying by taking @xmath53 arbitrarily large that no @xmath0-epistemic model for hilbert spaces of dimension @xmath83 can satisfy both the assumptions of continuity and separability .
it is interesting to compare how these no - go theorems could be used in practice to test @xmath0-epistemic models ( see @xcite for actual tests ) .
experimental tests that rule out @xmath0-epistemic models for smaller values of the continuity parameter @xmath17 are clearly stronger .
we thus consider how resources scales as @xmath84 .
if we use the construction of theorem 1 , then we need to use systems of dimension @xmath85 , and the resources needed to test @xmath0-epistemic models increase as @xmath86 .
if we use theorem 2 , and take e.g. the dimension @xmath52 , then one needs to prepare three states @xmath87 , where the number of copies of each state is @xmath88 .
again the resources needed increase as @xmath86 .
finally , we could also test @xmath17-continuous @xmath0-epistemic models using the construction given in pbr @xcite . in this case
we need to prepare @xmath89 distinct states , each of which is a product state of @xmath53 qubits , with @xmath90 .
the resources required for the application of the pbr construction therefore grow exponentially in @xmath91 .
experimental tests based on theorems 1 and 2 thus seem much easier than those based on the pbr construction .
note that our theoretical arguments and pbr s one rely on the fact that certain quantum probabilities are exactly equal to zero .
however , these arguments are robust against small deviations from these predictions , as expected in an experimental implementation where noise is inevitably present . indeed , eq .
( [ eq : nogo ] ) implies that the observed value @xmath92 provides an upper bound on the overlap @xmath93 of the ontic distributions @xmath19 .
a small value of @xmath94 therefore translates into a strong constraint on continuous @xmath0-epistemic models , since it implies that these distributions have only a small common overlap .
similarly @xmath95 in theorem 2 upper bounds the overlap @xmath96 of the @xmath53-copy joint distributions @xmath97 .
this last quantity can simply be related to the single - copy overlap @xmath98 if we further make the preparation independence assumption of pbr that joint distributions @xmath99 are product of individual distributions , which then implies @xmath100 .
note that a comparison of the sensitivity of the different tests to experimental noise is possible , but goes beyond the scope of the present work .
it would require a detailed modeling of the state preparation and measurement procedures .
_ discussion .
_ in his seminal paper on the probabilistic interpretation of quantum theory , born gave the wavefunction a functional interpretation : a mathematical object from which the probabilities of different measurement outcomes can be determined @xcite .
but the fundamental nature of this object , a real physical wave or a summary of our knowledge about physical systems , is a question that has divided physicists ever since .
a precise formulation of these two alternatives , opening the way to clear - cut answers , was provided by harrigan and spekkens @xcite .
if the wavefunction corresponds to a real , ontic , property of physical systems , the preparation of a system in different pure quantum states should always result in different physical states .
if , on the other hand , the wavefunction has an epistemic status , such preparations should sometimes result in the same underlying physical state .
pbr have recently introduced a no - go theorem that , given certain assumptions , rules out this latter possibility @xcite , thus awarding ontic status to the wavefunction .
this result can also be seen as a constraint on the structure of possible extensions or generalizations of quantum theory .
if they reproduce the quantum predictions and satisfy these assumptions , then such theories can only _ supplement _ the wavefunction @xmath0 with additional variables @xmath101 , i.e. , a system should be described by a physical state of the form @xmath102 .
the theorem of pbr relies on two main assumptions .
the first , which is also unquestioned in the present work , is that a system has a real and objective state @xmath1 that is independent of the observer .
the second is an assumption of preparation independence " , which states that independently prepared systems are described by independent product distributions over real states .
it can be replaced by the weaker separability assumption used here . in the present work
, we reached the same conclusion as pbr using a simple argument that relies on a natural assumption of continuity .
this notion of continuity captures the intuition that in a model where the quantum state is epistemic , a small variation of @xmath0 does not necessarily imply a variation of the underlying real state @xmath1 .
we derived a fundamental limit on the degree of continuity of @xmath0-epistemic models , as parametrized by a quantity @xmath17 , already at the level of single quantum systems ( theorem 1 ) . combining our continuity assumption with a separability assumption
, we then showed that no @xmath0-epistemic model can reproduce all the predictions of quantum theory ( theorem 2 ) .
besides their simplicity and the fact that they already constrain @xmath0-epistemic models for single quantum systems , an interest of our results is that they are easy to implement experimentally .
such an experimental test based on theorem 1 has been reported in @xcite . _
acknowledgments . _
m. p. would like to thank rob spekkens for helpful discussions .
we thank anonymous referees for useful comments .
we acknowledge financial support from the european union under project qcs , from the frs - fnrs under project diqip , and from the brussels - capital region through a bb2b grant . |
the organic charge - transfer compound tdae - c@xmath0 ( where tdae is tetrakis - dimethylamino - ethylene ) is a ferromagnet with the curie transition temperature of @xmath3 16 k ( ref .
this is the highest temperature onset of ferromagnetic behavior for a purely organic material .
it is customarily accepted that the system is an isotropic heisenberg ferromagnet with an extremely small anisotropy field@xcite@xcite .
the tdae molecule , a strong electron donor , transfers one electron to the lowest unoccupied molecular orbital ( lumo ) of c@xmath0 in a similar way to that found in alkali metal c@xmath0 compounds .
it has been established that the single - charged c@xmath0 alkali salts , a@xmath4c@xmath0 ( a= k , rb , cs ) , reveal metallic properties in a wide temperature range@xcite .
although , analogously to the alkali metal c@xmath0 compounds , the valence band in tdea - c@xmath0 originates from the triply degenerated @xmath5 orbital of the c@xmath0 molecule , the tdae - c@xmath0 system was found to be non - metallic@xcite@xcite .
the non - metallic behavior of tdae - c@xmath0 was explained assuming the jahn - teller effect and mott - hubbard localization@xcite@xcite .
magnetic susceptibility measurements@xcite@xcite@xcite as well as electron spin resonance ( esr ) results@xcite@xcite showed that the tdae - c@xmath0 system has only one @xmath6 magnetic moment per chemical formula unit . from the @xmath7-factor analysis
it was concluded that the spins are mainly localized on c@xmath8 ( ref .
the tdae@xmath2 donors should also carry non - paired electrons .
however , these centers remain esr silent , which is probably due to the mechanism of the spin - singlet pairing resulting from the dimerization shift of the neighboring tdae@xmath2 molecules@xcite .
two magnetic variants of tdae - c@xmath0 were discovered . as grown crystals ( @xmath9-phase )
reveal the antiferromagnetic ( afm ) behavior , whereas annealed crystals ( @xmath10-phase ) show the transition to the ferromagnetic ( fm ) state@xcite .
surprisingly , in x - ray diffraction ( xrd ) , these two structures look almost indistinguishable around the room temperature when the c@xmath0 balls rotate freely .
however , the difference is observed when freezing of the rotation of the c@xmath0 balls occurs at low temperatures@xcite@xcite . in the fm phase
, a new bragg reflection develops , suggesting lowering of symmetry and ordering of the c@xmath0 balls@xcite .
this reflection is not observed in the afm phase , which suggests the existence of a disordered ( glassy ) state of the c@xmath0 orientations .
nuclear magnetic resonance ( nmr ) data imply that in the fm phase the electron spin of c@xmath8 are partially delocalized on the tdae molecule , thus raising a possibility for the super - exchange interaction@xcite . in a simple model
, one would expect that the super - exchange mechanism would promote the afm behavior .
nevertheless , the observed ferromagnetic properties are explained in terms of the hubbard model with the antiferro - orbital order of the jahn - teller distorted ( jtd ) c@xmath0 balls@xcite .
indeed , nmr observations of the jt distortion and the asymmetric charge distribution on the c@xmath0 balls have recently been reported@xcite . however , the nmr results of arcon et al.@xcite suggest that the @xmath10-phase contains the c@xmath8 molecules with both ferromagnetic and antiferromagnetic configurations . in that case , the spontaneous magnetization should appear when the concentration of the fm configuration is high enough for the appearance of an infinite ferromagnetic cluster through percolation mechanism@xcite@xcite .
recently , mizoguchi and co - workers@xcite reported polymerization of tdae - c@xmath0 under pressure of @xmath11 kbar . the polymerized phase ( @xmath12-phase ) remains stable even after releasing the pressure
the polymerization process occurs along the c - axis and it has been suggested that the linear polymers can be formed due to a [ 2 + 2 ] cycloaddition process . in this article , we report the @xmath1 diagram of the stability of the polymeric tdae - c@xmath0 structure .
the effects of the high - hydrostatic pressure on the physical properties of both monomeric and polymeric phases were investigated .
we studied the physical properties of the @xmath12-phase at low and high temperatures and the effect of the temperature - induced depolymerization .
the properties of the tdae - c@xmath0 polymer are compared with those of other bonded fullerene structures .
single crystals of the tdae - c@xmath0 were prepared by the diffusion method as reported in ref .
the single crystals dimensions were , typically , of @xmath13 mm@xmath14 .
the presence of the fm phase was checked by the magnetization measurements .
= 0.40 ambient pressure electron spin resonance ( esr ) measurements were performed in the temperature range of @xmath15 k using a bruker esp300e x - band spectrometer . in the temperature range of @xmath16
k the esr spectra were acquired using a standard bruker te@xmath17 cavity that was equipped with an oxford instrument , model esr900 , gas - flow cooling system . in the upper temperature range ( @xmath18 k )
we used a bruker er4114ht high - temperature cavity system . the magnetic field and the microwave frequency
were calibrated using a commercially available nmr gauss - meter and a frequency counter , respectively .
the esr line intensities were calibrated using a secondary standard sample , a small speck of dpph ( 2,2-dipenyl-1-picrylhydrazyl from sigma ) .
the high - pressure esr measurements were performed using a high - pressure system that was recently developed at the epfl .
the high - pressure esr probe was designed as an interface to our bruker esp300e x - band spectrometer .
the probe consists of two sub - assemblies : 1 ) the microwave resonant structure containing the double - stacked dielectric resonator ( dr ) and 2 ) the miniature sapphire - anvil pressure cell ( sac ) .
the sac is ruby - calibrated ; thus the hydrostatic pressure can be monitored in situ by detecting the pressure - induced shift of the red fluorescence of cr@xmath19 ions in a small crystal of ruby .
the commercially available `` daphne '' oil was used as a pressure - transmitting medium . in this work ,
the maximum applied pressure was of 9 kbar . for performing low temperature measurements ,
the high - pressure probe was inserted into the cf-1200 oxford instrument gas - flow cryostat operating in the @xmath20 k temperature range .
the details of the dr - based high - pressure esr probe will be published shortly elsewhere @xcite .
the esr line intensities and the @xmath7-factor were calibrated using an additional reference sample , a polycrystalline mno / mgo , which was positioned in the active zone of the microwave resonant structure ( close to the gasket of the sac ) .
figure [ figure1 ] shows the temperature evolution of the esr - probed spin susceptibility and the esr linewidth of the ferromagnetic phase of tdae - c@xmath0 measured at ambient pressure . in the ferromagnetic region , below @xmath21 k , the linewidth can not be determined precisely due to the strong distortion of the esr line shape .
the line shape distortion is probably due to the non - homogeneity of the local internal fields in different ferromagnetic domains of the crystal .
the pressure dependence of the esr linewidth ( peak - to - peak , @xmath22 ) of tdae - c@xmath0 at ambient temperature is shown in fig .
[ figure2]a , whereas fig .
[ figure2]b shows the pressure dependence of the @xmath7-factor .
the initial ( ambient pressure ) linewidth of @xmath23 g slowly decreases with increasing pressure to 15 g at @xmath24 kbar .
then , at @xmath25 kbar , a transition to the polymeric phase is clearly seen as a sudden drop in the esr linewidth ( fig .
[ figure2]a ) .
this is also accompanied by an abrupt change ( increase ) in the @xmath7-factor ( fig .
[ figure2]b ) .
the phase transition is irreversible and the polymeric phase remains stable after releasing the pressure .
these results are in a partial agreement with the esr measurements performed at lower microwave frequencies by mizoguchi et al.@xcite .
the major difference is , however , that the critical pressure of suppression of the ferromagnetic transition in our measurements is rather lower and coincides with the pressure of the complete polymerization . in the monomeric tdae - c@xmath0
the esr linewidth is defined by dipolar interaction with additional narrowing introduced by the exchange interaction . accordingly , as can be seen in fig .
[ figure2]a , the monomer s linewidth slowly narrows with pressure approaching 7 kbar .
this is due to the larger overlap of the electronic wave functions , which leads to an enhanced exchange interaction . at ambient pressure ,
the @xmath7-factor value for the monomeric phase of tdae - c@xmath0 is 2.0006 and is distinctively closer to the @xmath7-factor of the fullerene anion c@xmath8 ( @xmath26 ) than to the @xmath7-factor value that is characteristic for the tdae@xmath2 cation ( @xmath27 ) .
this is due to the fact that the esr signal originates from the electrons that are mainly localized on the c@xmath0 balls .
the spins on the tdae@xmath2 radicals are esr - silent , which is probably due to a slight dimerization of tdae molecules , thus yielding a spin - singlet configuration . with increasing pressure , the @xmath7-factor linearly increases towards the value of the tdae@xmath2 cation .
this implies that the unpaired spin density is spreading towards the tdae molecule with increasing pressure . at @xmath28 7
kbar , a sudden narrowing of the esr line is visible due to polymerization . the linewidth drops by two orders of magnitude , reaching @xmath22(polymer)= 0.5 g. the narrow linewidth and the currie type temperature dependence of the spin susceptibility at this pressure suggest that the polymeric phase is non - metallic . upon polymerization
the @xmath7-factor value rises reaching 2.0024 and becomes almost pressure independent .
mizoguchi et al.@xcite suggested that the polymerization process decouples the previously esr - silent spins related to the tdae@xmath2 , which should lead to an effective doubling of the total number of spins .
the observed pressure dependence of esr susceptibility at room temperature ( fig .
[ figure2]c ) does not support this suggestion . at this stage , a small increase ( of @xmath29 ) of the esr susceptibility upon polymerization can be explained by the differences in the weiss temperatures between the two phases .
figure [ figure3 ] shows the temperature dependence of the esr linewidth and the @xmath7-factor for three different pressures .
the temperature dependence of the linewidth and the @xmath7-factor at applied pressure of @xmath30 2.6 kbar is shown in fig .
[ figure3]a and fig . [ figure3]b , respectively . as can be seen in fig .
[ figure3]a , below the characteristic temperature , @xmath31 k , in addition to the esr line of tdae - c@xmath0 monomer ( full circles ) , a new line appears ( open circles ) .
this new esr line can be assigned to the polymeric phase while taking into account the @xmath7-factor and esr linewidth evolution with pressure .
the intensity of the polymeric line does not exceed 15% of the intensity of the monomeric line .
the temperature dependence of the @xmath7-factor and esr linewidth at @xmath30 4.7 kbar is shown in fig .
[ figure3]c and fig .
[ figure3]d , respectively . at this pressure , a partial polymerization starts at higher temperatures , around @xmath32 k. the intensity of the polymeric line is now much more pronounced . for both applied pressures , the monomeric phase undergoes the ferromagnetic phase transition , whereas the polymeric phase does not . at pressure of @xmath30 7.0 kbar ,
the sample is fully polymerized already at room temperature ( fig .
[ figure3]e and fig.[figure3]f ) .
no ferromagnetic phase transition is visible down to 5 k. at this applied pressure , the esr susceptibility follows a simple curie law , without a detectable weiss constant .
the esr linewidth is very narrow ( @xmath33 g ) and is almost independent of temperature , whereas the @xmath7-factor slightly changes with temperature .
= 0.45 the stability of the polymer phase in the @xmath1 parameter space is depicted in fig .
[ figure4 ] .
the polymerization temperature ( @xmath34 ) has a linear dependence with applied pressure , where the proportionality constant .
the ratio of the polymeric and monomeric esr line intensities depends not only on pressure , but also on temperature . to deduce the exact structural dynamics of the polymer formation an additional structural study is needed .
the pressure dependence of the ferromagnetic transition temperature ( @xmath35 ) of the monomeric phase is depicted in fig .
[ figure5 ] . to determine @xmath35
, we can not compare the resonance field shift with the conventional bloch s law , because the resonance field has pronounced temperature dependence even above @xmath35 ( see fig .
[ figure3 ] ) .
this shift is due to the demagnetization effects , which depend on the sample shape@xcite .
concomitantly , we define @xmath35 as the onset temperature of the broadening of a linewidth distribution of the monomer - related esr features ( see : inset to fig .
[ figure5 ] ) .
this linewidth distribution broadening is due to the growth of an internal field below @xmath35 and the mosaicity of the crystal .
the observed pressure dependence of @xmath35 is similar to the parabolic dependence reported in ref .
@xcite . however , in contrast to their esr data acquired at low microwave frequencies and fields , the ferromagnetic transition observed in this work vanishes simultaneously with polymerization .
this result can be expected and explained in the framework of the theory by kawamoto and coworkers@xcite , because the onset of polymerization prevents the antiferro ordering of the orbital of the jtd fullerene molecules .
the polymeric phase remains stable even after the pressure is released .
for the polymeric phase , the temperature dependences of the esr parameters are shown in fig .
[ figure6 ] . as can be seen from comparison of the results presented in fig.[figure6 ] and in fig .
[ figure3]a and fig .
[ figure3]b , the low - temperature properties of the polymeric phase at ambient pressure are very similar to those observed at @xmath30 7 kbar .
the temperature dependence of the inverse esr susceptibility , @xmath36 , of the polymeric phase is shown in fig .
[ figure6]a .
since we are dealing with localized spins , this type of plot ( @xmath36 vs. t ) is the most informative , directly yielding information on the curie constant and the weiss temperature of the system .
the spin susceptibility below the room temperature reveals one spin @xmath37 1/2 per chemical formula unit ( @xmath38 1 ) and a simple curie behavior . as seen in fig .
[ figure6]a , the inverse susceptibility departs from the simple curie behavior at 260 k. this transition region is relatively broad and extends up to ca .
320 k. above 320 k , the esr susceptibility reveals the doubling of the number of spins per chemical formula .
this phenomenon can be understood in terms of re - appearance of the previously hidden spins of the tdae@xmath2 radicals .
the weiss temperature does not vanish anymore , having the value of @xmath39 k. this change of behavior at high temperatures is also seen in the temperature dependence of the linewidth ( fig .
[ figure6]b ) . at lower temperatures
the @xmath22 is almost constant , with a slight tendency to decrease with increasing temperature , whereas above the room temperature it changes its slope and starts to increase more rapidly .
the temperature dependence of the @xmath7-factor ( fig . [ figure6]c ) reveals a similar , distinctive change of behavior above the room temperature .
= 0.40 the polymeric phase remains stable up to the depolymerization temperature , @xmath40 k. above this temperature , both the @xmath22 and the @xmath7-factor recover their characteristic values for the ferromagnetic phase ( fig .
[ figure7 ] ) .
the depolymerization process is an irreversible transition . at the depolymerization temperature ,
both polymeric and ferromagnetic esr features are present , thus pointing to the coexistence of the two phases .
it also suggests that this phase transition does not seem to be an abrupt one . as in the case of ferromagnetism ,
the polymerization process might be influenced by percolation mechanism , as the samples of apparently lesser quality exhibit a bit lower depolymerization temperature . at low temperatures ,
the depolymerized crystals reveal the same temperature dependences of the esr parameters as those observed for the ferromagnetic crystals .
the @xmath7-factor shifts remarkably below 15 k , which indicates that the depolymerized crystal undergoes the ferromagnetic phase transition .
in the intermediate pressure range , below @xmath25 kbar , the tdae - c@xmath0 monomer partially polymerizes on cooling .
both the polymerization temperature ( @xmath34 ) and the relative ratio of the polymeric to the monomeric fractions depend on pressure ( fig .
[ figure4 ] ) .
preliminary analysis@xcite has suggested that the polymer would have a linear [ 2 + 2 ] cycloadduct bonding structure with a similar intra - chain ball distance as in the case of rb@xmath41c@xmath0 ( ref .
the most prominent difference between the systems is a much larger inter - chain distance in the case of tdae - c@xmath0 , which is due to the large size and anisotropic ( steric ) properties of the tdae interstitials .
= 0.40 rich phase diagrams have already been reported for several c@xmath0-related compounds .
in particular , interesting phase transitions were found for rb@xmath41c@xmath0 .
depending on cooling rate and quenching , it can form either monomeric , dimeric@xcite or polymeric phases@xcite@xcite .
nevertheless , the coexistence of the two phases has never been observed for rb@xmath41c@xmath0 .
in contrast , the coexistence of two phases at low temperatures was observed in the case of single - bonded linear polymers of na@xmath42rbc@xmath0 system@xcite . in that case , the partial polymerization can be explained by steric effects and by disorder resulting from different possible directions of the bond formation .
in contrast , tdae - c@xmath0 is a strongly anisotropic structure with the only one possible direction of the bond formation ( c - axis ) .
hence , the partial polymerization is probably governed by the orientation disorder of jtd - c@xmath0 molecules .
the polymerization temperature , @xmath34 , is much lower than the freezing temperature of the c@xmath0 molecule rotation for a given pressure .
this suggests that the freezing of molecular orientations is not sufficient for initiating the polymerization process and there exists a kinetic barrier for this phase transition .
it seems to be natural that different kinetic barriers can characterize various relative orientations of the jtd - c@xmath0 molecules . as the remaining monomeric phase still shows a pronounced ferromagnetic order , most probably , the fm molecular configuration has the highest energetic barrier .
it follows from our measurements that the depolymerization of tdae - c@xmath0 occurs at the higher temperature than in the case of one - dimensional a@xmath41c@xmath0 polymers ( a= k , rb , cs ) , where , depending on the compound , @xmath43 varies in the range of @xmath44 k ( ref .
it seems that @xmath43 of tdae - c@xmath0 is rather comparable to that found for the two - dimensional polymer na@xmath45c@xmath0 ( @xmath46 k)@xcite .
therefore , the polymeric chains of tdae - c@xmath0 seem to form much more stable structures than the double - bonded polymeric chains of the rb@xmath41c@xmath0 system .
if the intra - chain bounds of both tdae - c@xmath0 and rb@xmath41c@xmath0 were of the same nature ( isostructural ) , one would expect for them similar temperature stability .
the observed discrepancy in the temperature stability can be ascribed to a potential structural difference in the two polymeric structures .
alternatively , the previously not investigated effects of dopant molecule and the inter - chain coupling can be important for the polymeric chain stability .
a precise structural analysis is needed to answer these questions .
below room temperature ( @xmath47 k ) , the esr spin susceptibility of the polymerized @xmath48phase follows a simple curie law with one spin @xmath6 per chemical formula .
this indicates that spins probed by the esr experiment are localized .
moreover , in this temperature range , the esr linewidth for the polymerized @xmath48phase is very narrow ( @xmath33 g ) and almost temperature independent , which suggests strong exchange interactions between the spins .
the theoretical band calculations , however , predict that in the absence of electron correlations , the isolated single - charged , double - bonded linear polymer should be metallic with a half - filled band@xcite . also , the same property holds well for single - bonded linear polymers encountered in na@xmath42ac@xmath0 systems@xcite .
indeed , all the other charged linear polymers of the c@xmath0 compounds , discovered to date , are metallic in a wide temperature range , with possible ground - state instabilities , such as spin - density wave@xcite .
therefore , the tdae - c@xmath0 polymeric phase seems to be unique .
the inter - chain coupling is very important in alkali fullerides a@xmath41c@xmath0 , influencing their dimensionality .
however , according to erwin et al .
@xcite , a direct inter - chain coupling interaction can be neglected in tdae - c@xmath0 based on the large inter - chain separation found in these systems . assuming a simple model
, one would expect the tdae - c@xmath0 polymer should be a strongly anisotropic metal .
the effective strong localization observed in the @xmath48phase of tdae - c@xmath0 might suggest that its actual polymer topology essentially differs from that of rb@xmath41c@xmath0 . alternatively
, the localization effect might originate from the possible enhancement of the effective coulomb repulsion at c@xmath0 sites , due to the influence of the tdea@xmath2 radicals . in such a mott - hubbard localization scheme
, one would expect a non - vanishing weiss constant .
in contrast , the presence of the weiss constant has not been detected in the low - temperature @xmath48phase .
mizoguchi et al.@xcite suggested that the absence of the weiss parameter could be explained by re - activation of the positive exchange coupling with the spins of tdae@xmath2 radicals .
this , in turn , would fortuitously cancel the negative inter - c@xmath0 spin coupling leading to a diminishing weiss constant .
nevertheless , in this work , we did not observe the complete recovery of the tdae@xmath2 spins in the polymeric phase at the room temperature , as claimed by mizoguchi et al .
indeed , the appearance of tdae - spins at elevated temperatures is accompanied with development of the antiferromagnetic weiss constant . assuming that polymeric chains in the @xmath12-tdae - c@xmath0 have rather small sizes and are disordered
, one can apply a model of the random - exchange afm heisenberg 1d chains . in this case , the weiss temperature would be absent , whereas the esr susceptibility should be proportional to @xmath49 , where @xmath50 . fitting this model to our data yields @xmath51 .
clearly , this result does not support the above - mentioned model . for right now , a plausible reason for non - appearance of the weiss temperature in the polymer phase remains unclear .
as can be seen in fig .
[ figure6]a , in the polymeric @xmath48phase , a complete recovery of the tdae@xmath2-related spins is not observed at ambient temperature .
the recovery of the spins occurs rather at higher temperatures , with full development at 320 k. the dynamics of the tdae@xmath2-related spins is probably connected to small movements of the tdae@xmath2 molecules leading to the dimerization shift@xcite .
in conclusion , we have investigated the polymerization mechanism in tdae - c@xmath0 ferromagnetic system and the physical properties of the polymeric @xmath12-phase .
the complete polymerization at room temperature is observed at the pressure of 7 kbar . at the same pressure ,
the ferromagnetic transition is suppressed .
partial polymerization is observed at lower pressures and temperatures , and the stability of the polymeric phase was established in the @xmath1 parameter space .
the high depolymerization temperature suggests that the polymeric chains are much more stable than in the case of double - bonded linear polymers of the rb@xmath41c@xmath0 systems . to deduce the exact structural properties of the polymer , an additional high - resolution x - ray diffraction study is needed . moreover , the observed strong localization of spins in the polymeric tdae - c@xmath0 is in contradiction with conclusions from a simple theoretical reasoning and it is not comparable to any other charged linear polymer of the c@xmath0 . above 320 k ,
the previously silent spins on the tdae - c@xmath0 are revealed .
the observed decoupling of the tdae@xmath2 spins is accompanied by the appearance of the weiss constant . in summary ,
we conclude that the polymerization phenomenon in tdae - c@xmath0 is very important to better understanding the ferromagnetic properties of the system .
however , it can also shed new light towards our general understanding of the ground state electronic properties of the linear fullerides polymers .
this work has been supported in part by a grant - in - aid for scientific research on the priority area `` fullerenes and nanotubes '' , from the japanese ministry of education and culture ( t.k .
, m.f . , and k.o . ) and by the polish kbn grant # 2-po3b-090 - 19 ( a. s. ) .
the support of the swiss national scientific foundation is also greatly acknowledged ( s. g. and l. f. ) . | the high - pressure electron spin resonance ( esr ) measurements were preformed on tdae - c@xmath0 single crystals and stability of the polymeric phase was established in the @xmath1 parameter space . at 7 kbar
the system undergoes a ferromagnetic to paramagnetic phase transition due to the pressure - induced polymerization .
the polymeric phase remains stable after the pressure release .
the depolymerization of the pressure - induced phase was observed at the temperature of 520 k. below room temperature , the polymeric phase behaves as a simple curie - type insulator with one unpaired electron spin per chemical formula .
the tdae@xmath2 donor - related unpaired electron spins , formerly esr - silent , become active above the temperature of 320 k and the curie - weiss behavior is re - established . |
the feedback - control method presented in this paper mimics the `` run - and - tumbling '' of _ escherichia coli _ but combined with active steering , thus it is a simple but more efficient method to transport microscopic swimmers under thermal fluctuations .
the added ability of deterministic `` active '' reorientation achieves more efficient transportation of particle than the natural `` run - and - tumbling '' .
it also has several advantages over other conventionally used micro - manipulation techniques : laser tweezers for example uses high power laser that could damage fragile samples and can be tricky to use as particles often jump out of the confining potential@xcite .
+ we have also addressed the problem of optimizing the feedback and observed some interesting insights .
the active reorientation decouples the abm and the reorientation process , in contrast to passive reorientation . due to this decoupling
, we showed that the optimal acceptance angle is a function of @xmath14 .
remarkably , since the timescale of passive reorientation is determined by @xmath15 , which scales in the cubic order of the radius , our method becomes particularly effective when the particles are relatively large .
for instance in the case of a particle of diameter @xmath16 , @xmath17 in water whereas it can be as small as @xmath18 s in our experiments , as shown in fig .
[ u0_omega ] , leading to more than 10 times enhancement of peclet number . for even smaller particles ,
the gain of active reorientation becomes less significant as @xmath19 approaches @xmath20 , though the magnitude of @xmath20 is tunable by the applied electric field .
+ although the optimal tolerance angle is determined by @xmath14 , our theory also predicts the robustness of the proposed algorithm .
this is guaranteed by the fact that @xmath7 has a quite shallow shoulder towards large @xmath0 , at least for the range of parameters relevant to this experiment . in real world , individual particle may possess variable @xmath20 or experience an inhomogeneous @xmath19 from the environment .
it may also be possible that the exact position of the target might not be known , or estimation of the orientation might not be precise , contributing to a poor resolution of @xmath21 .
the final pclet number however , is weakly affected by these noises thanks to the broad tolerance of optimal @xmath0 . that allowed us to control the motion of particles in spite of the strong variability in their behavior .
+ the key of our method lies in our finding of the peculiar rotational motion of janus particle that can be switched on and off by changing the parameters of the electric field .
although the mechanism of this rotation is not yet fully understood , experimental measurements of @xmath20 showed that it was proportional to @xmath22 , implying that the torque @xmath23 as well as the force @xmath24 may originate from an asymmetric flow field around the particle generated by iceo .
the other parameter , @xmath25 , has been poorly explored in the framework of icep .
further experimental studies , as well as theoretical works , should be addressed .
our results also suggested that geometrical factors such as chirality of the particle can play an critical role in determining the swimming behavior of the particle .
an interesting challenge would be to find a way to artificially fabricate `` brahma particles '' , which as the hindu god would have four `` heads''@xcite .
these swimmers would have two well - designed axis of asymmetry , one used to propel the particle and the other to induce `` switchable '' rotations .
our experiment provides the first proof - of - principle demonstration of such an idea .
our results encourage further quest towards engineering functional artificial swimmers .
+ w. f. paxton , p. t. baker , t. r. kline , y. wang , t. e. mallouk and a. sen _ catalytically induced electrokinetics for motors and micropumps .
_ , journal of the american chemical society 128 ( 2006 ) , pp.148818 .
w. paxton , k. kistler , c. olmeda , a. sen , s. st .
angelo , y. cao , t. mallouk , p. lammert and v. crespi _
catalytic nanomotors : autonomous movement of striped nanorods .
_ , journal of the american chemical society 126 ( 2004 ) , pp.1342431 .
b. qian , d. montiel , a. bregulla , f. cichos , and h. yang _ harnessing thermal fluctuations for purposeful activities : the manipulation of single micro - swimmers by adaptive photon nudging .
_ , chemical science 4 ( 2013 ) , pp.1420 .
f. kmmel , b. ten hagen , r. wittkowski , i. buttinoni , r. eichhorn , g. volpe , h. lwen and c. bechinger _
circular motion of asymmetric self - propelling particles .
_ , physical review letters 110 ( 2013 ) , pp.198302 .
t. cordes , w. moerner , m. orrit , s. sekatskii , s. faez , p. borri , h. prabal goswami , a. clark , p. el - khoury , s. mayr , j. mika , g. lyu , d. cross , f. balzarotti , w. langbein , v. sandoghdar , j. michaelis , a. chowdhury , a. j. meixner , n. van hulst , b. lounis , f. stefani , f. cichos , m. dahan , l. novotny , m. leake and h. yang _
plasmonics , tracking and manipulating , and living cells : general discussion .
_ , faraday discussions 184 ( 2015 ) , pp.451473 ( see pp.464 - 465 ) . c. peng , i. lazo , s. v. shiyanovskii and o. d. lavrentovich _ induced - charge electro - osmosis around metal and janus spheres in water : patterns of flow and breaking symmetries _ , physical review e , 90 ( 2014 ) , pp.051002 .
k. maeda , h. , onoe , m. , takinoue , and s. takeuchi _ controlled synthesis of 3d multi - compartmental particles with centrifuge - based microdroplet formation from a multi - barrelled capillary _ , advanced materials , 24 ( 2012 ) , pp.13401346 .
we used polystyrene spheres of diameter m. a droplet of a solution of theses polystyrene spheres is then dragged on a glass slide by a linear motor at the appropriate speed to obtain a monolayer of particles@xcite .
we do not need a perfect crystal in our case but it is important that there is no particle on top of each other . using thermal evaporation , one of their hemispheres is then coated by thin layers of chromium and gold with nm and nm .
the other hemisphere facing the glass slide remains bare polystyrene so that the particles have two hemispheres with different polarizabilities .
after the coating process , the particles are detached from their substrate using mild - sonification and suspended in ion - exchange water .
the observation of the particles at high magnification shows that they are almost always chiral ( as shown in the inset of fig .
[ scheme_janus ] ) , which is due to a fast slightly slanted evaporation process .
note that the amount of janus particles exhibiting rotations at low field amplitude increases when the metal layers is quite thick .
for example , particles with nm rarely exhibit rotations . making chiral janus particles thus requires to deposit large enough quantities of metal .
a droplet of a suspension of janus particles is then put in between two ito electrodes sandwiching a spacer made of stretched parafilm of about m. using a function generator connected to the ito slides , we apply a vertical ac electric field to the solution such that .
to prevent the particles from sticking to the bottom electrode , we apply a surface treatment to the ito glass slides : the slides are exposed to a strong plasma for several minutes and are then immersed in a 5 % solution of pluronic f-127 ( a non - ionic copolymer surfactant ) for more than one hour .
they are then washed with water to remove the excess of pluronic . by coating the surface of the electrodes with this surfactant
, we significantly reduce the risks of adhesion .
however , this problem still remains one of the biggest experimental difficulties . throughout the experiment
, particles were imaged using 10x , na=0.3 objective lens mounted on a standard inverted microscope .
particles were illuminated with an incoherent light source , and the transmitted light was captured using ccd camera with 512x512 pixels at the frame rate of 100 fps .
tracking and feedback was done by a home - built labview program .
real - time tracking was initiated manually by feeding the program with the position of the particle of interest .
then for subsequent frames captured by camera , the small roi around the target particle was extracted , then thresholded to obtain a binary image .
the center of the mass was calculated from this binary image , and the updated particle coordinate was passed down to the next acquisition loop . at the same time , coordinate information was sent to the feedback loop , where it calculated the angle .
feedback loop directly communicates with the function generator via usb connection , and updated the appropriate control parameters .
the particles move at a constant speed @xmath26 , rotate at the frequency @xmath20 and are subjected to translational and rotational noises @xmath27and @xmath28 respectively . if we assume that their motion is overdamped , we can thus write the following system of langevin equations : @xmath29 this model had already been studied to get an analytical expression for the mean - square displacement of l - shaped artificial swimmers@xcite but here , we will focus on the autocorrelation function instead .
the second equation can easily be integrated to get @xmath30 @xmath28 being a white noise , we have @xmath31 according to eq .
[ [ gaussian ] ] , @xmath32 is a sum of gaussian variables and is therefore a gaussian itself . as we just calculated its first and second moments , we can deduce the expression of the probability density @xmath33 and the green function @xmath34 time evolution of the auto - correlation function for different frequencies @xmath25 .
the symbols are experimental measurements and the lines of the same colors correspond to a fit using expression [ vcorr ] .
we have three fitting parameters : @xmath20 , @xmath26 and @xmath19.,width=7 ] the velocity autocorrelation function of a particle at time @xmath35 is given by @xmath36 \cdot \left [ u_0 \boldsymbol{\hat{u}}(t+\tau ) + d_t \boldsymbol{\xi}_t(t+\tau ) \right ] \right\rangle\\ \\ = 4 d_t \delta(\tau ) + u_0 ^ 2 \langle \boldsymbol{\hat{u}}(t ) \cdot \boldsymbol{\hat{u}}(t+\tau ) \rangle \\ \\ = 4 d_t \delta(\tau ) + u_0 ^ 2 \left\langle \left ( \begin{array}{l } \cos\phi(t)\\ \sin\phi(t ) \end{array } \right ) \cdot \left ( \begin{array}{l } \cos\phi(t+\tau)\\ \sin\phi(t+\tau ) \end{array } \right ) \right\rangle \\ \\ = 4 d_t \delta(\tau ) + u_0 ^ 2 \left\langle \cos\phi(t)\cos\phi(t+\tau ) + \right . \\
\\ \hspace{0.5 cm}\left .
\sin\phi(t)\sin\phi(t+\tau ) \right\rangle .
\end{array } \label{averages}\ ] ] here , @xmath37 represents an ensemble average given , for an arbitrary function @xmath38 by @xmath39 injecting eqs .
[ [ dens_prob ] ] and [ [ green ] ] into [ [ ensemble ] ] , we can calculate the two averages of eq .
[ [ averages ] ] and find as a final expression for the velocity auto - correlation function @xmath40 strictly speaking , this expression diverges at @xmath41 but the system of langevin eqs .
[ [ lang_sys ] ] is actually only valid at times greater than the typical collision time of the heat bath .
we can thus neglect the first term .
if we use the expressions of the average velocity of the particles @xmath26 and of the angular velocity @xmath20 , we recover eq .
[ [ vcorr ] ] .
the agreement with the experimental results is excellent as can be seen on fig .
[ correlations ] . | even though making artificial micrometric swimmers has been made possible by using various propulsion mechanisms , guiding their motion in the presence of thermal fluctuations still remains a great challenge .
such a task is essential in biological systems , which present a number of intriguing solutions that are robust against noisy environmental conditions as well as variability in individual genetic makeup . using synthetic janus particles driven by an electric field
, we present a feedback - based particle guiding method , quite analogous to the `` run - and - tumbling '' behavior of _ escherichia coli _ but with a deterministic steering in the tumbling phase : the particle is set to the `` run '' state when its orientation vector aligns with the target , while the transition to the `` steering '' state is triggered when it exceeds a tolerance angle @xmath0 .
the active and deterministic reorientation of the particle is achieved by a characteristic rotational motion that can be switched on and off by modulating the ac frequency of the electric field , first reported in this work .
relying on numerical simulations and analytical results , we show that this feedback algorithm can be optimized by tuning the tolerance angle @xmath0 .
the optimal resetting angle depends on signal to noise ratio in the steering state , and it is demonstrated in the experiment .
proposed method is simple and robust for targeting , despite variability in self - propelling speeds and angular velocities of individual particles .
the physics of active suspensions made significant progress during the past decades and it is now possible to build artificial microscopic particles able to self - propel in a fluid .
the range of possible applications of such swimmers is wide , with fascinating perspectives : targeted drug delivery@xcite , bottom - up assembly of very small structures@xcite , mixing or automatic pumping in microfluidic devices@xcite , design of new microsensors and microactuators in mems@xcite or artificial chemotactic systems@xcite to name a few .
a lot of man - made microscopic swimmers fall into the category of `` janus '' particles which share the same property : an asymmetric structure inducing a breaking of symmetry of the interactions with the surrounding fluid resulting in a self - propelling force .
several physical phenomena can be at the origin of this force : local temperature gradients induced by a defocused laser beam@xcite ( thermophoresis ) , enzimatic catalysis of chemical reactions by a coated surface@xcite or electrostatic interactions between surface charges and the ions of the solution@xcite ( induced - charge electrophoresis or icep ) .
+ if several methods are known to generate self - propelling forces for janus particles , guiding their motion remains a challenge .
the biggest difficulty consists in controlling their orientation , a particularly delicate task when working with microscopic objects subjected to thermal fluctuations .
swimmers need to resist rotational diffusion by fixing or steering their orientation to reach specified targets or follow given trajectories .
experimental works showed that it was possible to lock the orientation of catalytic nanorods made of ferromagnetic materials using magnetic fields@xcite .
another interesting method involves visualizing the orientation of the particle at every moment and turn on the self - propelling force only when it is directed to the right direction@xcite . in that approach ,
the reorientation process is `` passive '' in a sense that the experimentalist waits for rotational diffusion to correct the orientation of the particle .
+ in this paper , we use janus particles driven by icep and introduce a new method to control their trajectory with an `` active '' reorientation process .
this new concept consists in switching between two distinct modes of motion exhibited by the particles : a self - propelling state and a regular rotation state .
such rotations had already been observed experimentally with l - shaped self - propelling swimmers moving by thermophoresis@xcite but the origin and characteristics of these rotations are very different here .
the janus particle under feedback control exhibits a motion quite similar to the `` run - and - tumbling '' behavior observed for the bacteria _ _
escherichia coli__@xcite . however , the reorientation is not random but deterministic , which might be compared to the adaptive steering found in evolved organisms , e.g. phototaxis in _ _ volvox carteri__@xcite .
such a `` hybrid '' strategy enables a high efficiency while minimizing the complexity of the implementation . in the first part of this article
, we will describe in detail the two different behaviors exhibited by our janus particles .
based on these properties , the second part will be devoted to the experimental implementation of the proposed particle guiding method . finally in the third part , we will present numerical simulations and analytical calculations showing how it is possible to optimize the feedback process .
[ cols="^,^ " , ] _ volvox _ on the other hand is a large multicellular organism which carries photoreceptors and thousands of flagella ( @xmath1 , @xmath2 , @xmath3 , @xmath4 ) .
our theory gives @xmath5 implying that continuous steering is the optimal for phototaxis of _
volvox_. in fact , _ volvox _ coordinates thousands of flagella to make steering motion , and even has an adaptation mechanism .
+ finally , we consider two limiting cases .
first , in the limit of @xmath6 , i.e. if steering accompanies no time cost , we have checked that @xmath7 monotonously decreases in @xmath8 $ ] so that the optimal angle is always @xmath9 .
this corresponds to a `` perfect '' steering case , where the particle swims straight to the target to minimize the travel distance .
the second case we want to consider is a `` passive '' reorientation : when @xmath10 , we now wait for the rotational diffusion to reorient it . in that case , the average reorientation duration @xmath11 is given by @xmath12 using this new equation for @xmath11 , @xmath7 becomes monotonously increasing in @xmath8 $ ] so that @xmath13 . |
a growing body of evidence suggests that long - duration grbs arise from the death of massive stars .
if true , the large - scale redshift distribution of grbs should trace the star - formation history of the universe @xcite . and , as explosive and transient events detectable to high redshift ,
grbs clearly offer a probe of the universe that is distinct from other lighthouses @xcite .
already , grbs have helped shed light on the nature of damped lyman @xmath2 absorbers ( e.g. , * ? ? ?
* ) and the nature of the faint end of the luminosity function of galaxies at moderate redshifts ( e.g. , * ? ? ?
* ) . as a redshift sample selected from an apparent isotropic population and from hosts spanning the breadth of the galaxy luminosity function
, grbs have the potential to provide a unique census of the redshift distribution of matter in the universe . not only can the large - scale distribution of grbs help constrain the star - formation rate ( sfr ) , but , in a manner complementary to pencil - beam surveys and magnitude - limited galaxy and quasar surveys , the small - scale distribution can help test hypotheses and observational suggestions about clustering and periodicity of sources in redshift space . given the penetrative powers of grbs through dust , several studies have attempted to compare the large - scale redshift distribution of grbs with the sfr obtained by other means @xcite .
such studies focused on using the prompt @xmath3-ray properties of batse bursts , calibrated with some measured grb redshifts , to determine the bursting rate .
with so few actual redshifts measured , each of which were measured under different observational constraints , the task of distinguishing different sfr scenarios from the sparsely sampled grb rate is considerably challenging . indeed we show herein , that when the observational biases in redshift determination are taken into account , several proposed forms for the universal sfr can not be distinguished .
investigation of the small - scale distribution of grbs has not been presented thus far .
of particular interest is whether the clustering properties in redshift space are significant in light of the observational biases and the intrinsic redshift distribution .
do grbs occur at preferred redshifts ?
there is enough historical , albeit controversial , evidence to warrant an investigation of the significance of special redshifts in the universe using grbs .
@xcite , for instance , found evidence for periodicity of clustering of redshifts on 128 @xmath4 mpc scales in two pencil - beam surveys . while excess power on these scales has not been definitively confirmed in higher - redshift studies ( e.g. , with the 2df redshift survey : @xcite ; although see @xcite and @xcite ) the @xcite result may be statistically significant in comparison with @xmath5-body simulations @xcite .
but , as @xcite emphasize , large clusters in a galaxy sample tend to accentuate the appearance of excess power at certain scales .
a grb sample does not suffer such a bias since bursts occur in regions of space that are gravitationally ( and causally ) disconnected .
the focus of this paper is on the small - scale ( clustering ) distribution of grb redshifts .
in addition we examine the large - scale distribution of grb redshifts , and note the difficulty in determining the global grb rate due to the small sample and strong observational biases . in
[ sec : sample ] we present the grb redshift sample for 26 bursts , corrected for the heliocentric motion through the local hubble flow and construct an observationally - motivated probability distribution for redshift discovery . in
[ sec : res ] , we compare the large - scale distribution of grb redshifts with that expected from the observing biases and an intrinsic rate distribution . after showing the insensitivity of the present sample to distinguishing various sfrs , and under the _ anzatz _ that grbs _ should _ trace the sfr , we fix a model for redshift discovery probability that adequately reproduces the universal sfrs . using this distribution
, we then test the significance of the number of observed pairs of grbs with small recession velocity differences .
we conclude with a discussion about the importance and usefulness of conducting precise redshift measurements for future grbs .
table [ tab : z ] presents a summary of the sky location and observed redshifts associated with 26 cosmological grbs , in order of increasing redshift .
the highest observed redshift system derived using absorption or emission features with the highest reported accuracy is given in column seven .
the line features used to measure the reported redshift are given in column six .
absorption - line redshifts , which place a strict lower limit to the actual grb redshift , are found when a spectrum of the afterglow is absorbed by metal - line systems along the line of sight .
emission - line redshifts are derived from spectroscopy of the galaxy associated with the grb position , either found promptly , superimposed with the spectrum of the transient afterglow ( e.g. , 980703 ; @xcite ) , or at later times , when the afterglow light has faded .
nineteen sources in the sample have measured emission - line redshifts , and 12 have absorption - line redshifts . the uncertainty in the redshift measurements are usually provided in the literature , but where none were provided we assumed an error of 10 percent in the least significant digit reported .
since we are interested in testing the observed rate and redshift distribution of grbs against different models for star - formation and small - scale clustering , we now discuss the observational biases in grb detection , localization , and finally , redshift determination .
if grbs arise in collimated emission , then the relativistic doppler beaming allows for detection of a grb only if the detector lies in the direction of the cone of emission . by studying the opening angles of those grbs which have been detected ,
it is reckoned that only about one in every five hundred grbs are detectable at ( i.e. , pointed toward ) earth due to beaming @xcite . of those bursts that are beamed toward earth , only those that reach a certain critical flux level
will be detected by the triggering algorithms of a given satellite .
the typical threshold for on - board detection in the batse catalog is a peak flux of a 0.3 photon @xmath6 s@xmath0 @xcite . in finding that the total energy release in grbs is approximately constant
, @xcite showed that bursts that are more highly collimated appear to be intrinsically brighter ( in flux / fluence per unit solid angle ) .
thus , for a scenario where grbs are jetted and release about the same amount of electromagnetic energy , brighter bursts tend to be _
more distant_. as a clear illustration of this , note that the peak flux of the highest redshift burst , grb000131 ( @xmath7 ) , was at the top 5% of the entire batse catalogue @xcite .
this somewhat counterintuitive trend implies that the brightest grbs can be detected to extremely high redshifts , certainly beyond the age of reionization ( @xmath810 ) .
more important for the present study is that malmquist bias should play little role in diminishing the observed burst rate at high - redshift : bursts significantly fainter ( by up to @xmath9 ) than grb000131 could have been detected at similar redshifts . using the prompt high - energy properties , irrespective of individual burst redshifts
, it has been shown that the brightness distribution of grbs is consistent with a broad range of star - formation rates and luminosity functions ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . using distance indicators calibrated from a sub - set of the bursts with known redshift , @xcite claimed that the grb rate increases monotonically at least out to @xmath10 . in our opinion , this suggestion points more to the pitfalls of extrapolating grb distance indicators beyond the redshift range in which they were calibrated , than revealing concrete properties of the high - redshift bursting rate density .
another approach to constrain the burst rate , which we attempt herein , is to use only those bursts with actual measured redshifts .
for those grbs that are triggered and rapidly localized using the prompt x - rays or @xmath3-rays , the observing community tries to determine a sub - arcsec position and ultimately associate a redshift via host or absorption - line spectroscopy .
this is accomplished by first identifying a transient afterglow .
three major biases are important for afterglow discovery .
first , grbs will be preferentially localized if they occur at a time and place in the sky where ground- or space - based telescopes can rapidly observe the prompt burst position with large enough fields of view .
second , only grb afterglows with brightnesses above a given threshold for the particular afterglow - discovery observations will be localized .
last , only grb afterglows that are not heavily extincted by dust obscuration will be first localized at optical wavelengths .
clearly , radio and x - ray afterglow discoveries do not suffer this third bias .
the existence of a population of dark bursts @xcite , those bursts without detectable optical afterglow emission , is clearly a result of all three biases .
( the current dark burst debate resolves around the relative importance of dust versus observational biases in the manifestation of dark bursts @xcite . ) since , for now , grbs with redshifts are preferentially those that were first localized at optical wavelengths ( all but 970828 , 980329 , and 000210 were first localized at optical wavelengths ) , it seems appropriate to suppose that a population of optically - localized grbs should trace a star - formation rate derived from other optically - selected samples ; @xcite have provided a parameterized version of a sfr based upon the un - dust corrected hubble deep field measurements ( model sf1 ) ; in this model the sfr drops beyond @xmath11 . in future grb redshift samples of bursts that are first localized at radio , x - ray or even infrared wavelengths ,
the expectation is that the observed rate will more closely trace the true ( high mass ) sfr in the universe .
for those grbs that are triggered and localized , most are followed up spectroscopically .
if an absorption or emission - line redshift is found , it is of interest to know the relationship between this and the true grb redshift .
a definitive measure of burst redshifts would need to connect the bursting location with some stationary gas local to the grb , resulting in a detection of transient features .
to date , no transient absorption or emission features at optical wavelengths features have been associated with a grb afterglow .
transient features , consistent with ( later - time ) optical spectroscopic redshift measurements , have been seen in a prompt burst spectrum @xcite and a handful of x - ray afterglow spectra ( e.g. , * ? ? ?
* ; * ? ? ?
at present , the uncertainties in these x - ray - derived redshifts are orders of magnitude larger than those found typically at optical wavelengths . perhaps more important
, redshifts derived purely from x - ray spectroscopy are contaminated from an unknown ( potentially relativistic ) outflow speed of the emitting or absorbing material ; thus , precise measurements of the systematic redshift of grb progenitors are best determined using the uv and optical features of host hii regions and/or intervening host clouds .
if a grb occurs in intergalactic space , with no afterglow absorption due to local gas , then the observed redshift would be systematically lower than the true redshift .
however , to date , the nature of the highest @xmath12 absorption lines , suggests the presence of gas columns and abundances not typically associated with tenuous outflow gas halos of galaxies . specifically , the equivalent widths of the highest redshift absorption systems in grb afterglows are significantly larger than in systems seen though quasar lines of sight ( e.g. , * ? ? ?
this is naturally understood in the context of the prevailing progenitor model , namely that long - duration bursts arise from the death of massive stars and , owing to the rapid evolution of massive stars , grb explosion locations should be in or near relatively dense regions of active star - formation .
thus , grbs should reside in or near the region that gives rise to the gas absorption .
consistent with this picture is the observation that well - localized grbs , a subset of which comprises the present sample , have been observationally connected to the location of the light of the putative host galaxies @xcite .
the probability of a spurious ( spatial ) association of an afterglow with its putative host has been shown to be small ( @xmath13 ) for most bursts .
in fact in a sample of 20 grbs with arcsecond localizations , statistically , at most a few spatial associations could have been spurious @xcite . therefore , from the location argument
, grb redshifts determined from emission line spectroscopy of the hosts are likely to be close to the true systematic velocities of the progenitor ; at most , we would expect a velocity offset on the order of a few hundred km s@xmath0 from the motion of the progenitor system about its host . likewise , unless the afterglow light pierces some high - velocity outflowing material , absorption - line redshifts should also provide a accurate measure of the true grb redshift .
perhaps most convincing that measured redshifts are closely related to the real systemic redshifts , is that of the five grbs with both emission and absorption line spectroscopy , the highest redshift absorption system is always found to be consistent with the emission - line redshift of the host .
one striking feature of the current redshift sample is the small apparent redshift differences , on the order of a thousand km s@xmath0 , between some bursts from different sky locations and apparently random trigger dates ( differences of 6 months to 5 years ) .
( unlike in pencil - beam surveys the proximity in redshift can have nothing to do with systems , such as galaxies in clusters , that are in causal contact . )
these small velocity differences are comparable to the velocity of the solar system through the local hubble frame , and so a correction for this systematic motion is warranted .
such motion was measured by the angular dependence of temperature variations in the cosmic microwave background ( cmb ) with the cobe satellite @xcite . this measured velocity ( @xmath14 km s@xmath0 ) and direction ( toward galactic coordinates @xmath15 deg , @xmath16 deg ; @xcite )
are due to vector sum of the heliocentric motion about the galactic center , the peculiar motion of the milky way in the local group and the local group in - fall .
we assume that all reported redshifts have been corrected to the heliocentric redshifts ; this correction is at most of order tens of km s@xmath0 and so this assumption is relatively unimportant . in order to remove this systematic bias from a measured redshift @xmath12
, we find the angle ( @xmath17 ) between the grb position and the heliocentric motion through the cmb ; we then find the projected heliocentric velocity toward the grb as @xmath18 .
if @xmath19 is the uncorrected apparent recession velocity of a distant source , then , from the definition of redshift and the formula for relativistic velocity addition , the corrected redshift in the local hubble frame ( lhf ) is found from : @xmath20 with the uncorrected apparent recession velocity as , @xmath21 the calculated values of @xmath22 are given in column eight of table [ tab : z ] .
the associated uncertainties are derived by error analysis of equation [ eq : eq1 ] , assuming that the errors on @xmath23 and @xmath17 are uncorrelated and noting that the fractional error on @xmath17 is significantly smaller than the fractional error on @xmath23 .
as seen in table [ tab : z ] , for all but the most accurately measured redshifts , this correction adds negligibly to the fractional error of the redshift .
table [ tab : z1 ] shows the absolute difference in redshift and recession velocity of each burst matched with the burst that is closest in apparent recession velocity .
figure [ fig : pairs ] shows this distribution of `` nearest neighbor '' apparent recession velocity differences versus redshift .
the smallest groupings occur around redshift of unity , where the density of known grb redshifts is highest .
four pairs are closer than @xmath241000 km s@xmath0 ( 0.0034@xmath25 ) in redshift space , and the largest velocity difference is between the two highest redshift bursts ( @xmath26 ) . interestingly , the velocity differences of the four closest burst pairs ( at redshifts @xmath270.692 , 0.844 , 0.961 , and @xmath28 ) are less than the velocity dispersion of a large cluster of galaxies and so , given the unknown peculiar velocity of the grb hosts and the progenitor system within the hosts , these close burst pairs are consistent with having occurred at the same cosmic time . even assuming no peculiar velocities of grb progenitors relative to their local hubble flow , eight of 26 grbs occurred within 30 @xmath1 myr of another grb .
this was determined by computing the look - back times between the four close pairs and assuming a cosmology with @xmath29 km s@xmath0 mpc@xmath0 , @xmath30 and @xmath31 = 0.3 .
this cosmology is assumed throughout the paper where needed .
how significant is this proximity in cosmic time ?
the bursts listed in table [ tab : z ] have been detected with recession velocities in the range @xmath32 ( grb011121 ) to @xmath33 = 0.94@xmath25 ( grb000131 ) . in an ensemble of @xmath34 bursts , assuming uniform probability of a grb occurring and being detected between @xmath35 and @xmath33 , the probability that given pair of successive bursts are not as close as a distance @xmath36 is @xmath37 .
therefore , the probability that at least one close pair exists in the ensemble , is , @xmath38 for @xmath39 km s@xmath0 and @xmath40 , the probability of at least one close pair with @xmath41 km s@xmath0 is 0.967 .
the probability of at least one close pair with at most the smallest velocity difference in the observed grb sample ( @xmath42 km s@xmath0 ) is 0.585 .
both these probabilities thus indicate that there is nothing particularly unusual about the occurrence of _ at least one _ close pair in the current sample . in general , the probability that exactly @xmath43 pairs
are close can be approximated as , @xmath44^k \frac{p^{n - 1 } \times n!}{(n - k)!\ , k!}. \label{eq : pk}\ ] ] for the present sample , there are @xmath45 close pairs . using this simplistic calculation
we therefore expect four pairs to occur by random chance in 18.6% of samples .
we have verified this approximation by a monte carlo simulation , selecting @xmath34 bursts uniformly over the range [ @xmath46 ; however , this approximation breaks down for large velocity differences as it does not account for relativistic velocity subtraction between successive bursts in velocity space .
the above probability estimate does not take into account some important observational and endemic biases .
first , the rate of grbs is not uniform in redshift or recession velocity space .
instead , we expect grbs to trace ( or at least approximate ) the star - formation rate @xcite , so that the peak of bursting activity should occur around @xmath47
. this will make close velocity pairs more likely at redshifts near unity .
second , the chance that a burst will be _ localized _ well enough to follow - up with spectrometers is not uniform in redshift .
instead , this chance depends , for optical localizations , sensitively on the native dust obscuration of the afterglow ( see [ sec : darkbias ] ) .
third , observing conditions and instrumentation play a strong role in determining whether the redshift of a grb can be detected at that redshift .
most important , emission lines and absorption lines falling outside the broadband spectral coverage hamper the ability to detect redshifts ( this may be an explanation for why no redshifts have been detected for 980519 , 980326 , and 980329 ) . even if a redshifted line falls within the spectral range , the presence of night sky lines make detection of redshifts at certain wavelengths more unlikely .
fourth , instrumental sensitivity varies across the spectral range and from instrument to instrument , night to night , and airmass to airmass . since we wish to know how the observed redshift sample compares with that expected given the above considerations we construct a toy probability model , @xmath48 , giving the differential probability that a burst at redshift @xmath12 could ultimately yield a redshift measurement .
we construct a global ( rather than individual ) distribution , neglecting the instrumental sensitivity differences from burst to burst and the non - negligible chance that a redshift will be incorrectly assigned even after the detection of one or more spectral features .
we use several parametrized versions of the star formation rate ( sfr ) from @xcite .
the overall shape of the distribution , @xmath49 , is then proportional to the number of bursts per unit redshift per unit observer time ; that is , @xmath49 is proportional to the co - moving volume element and the co - moving sfr ( @xmath50 ) versus redshift : @xmath51 following eqs .
[ 2][4 ] in @xcite . here
@xmath52 is the luminosity distance and @xmath53 .
the factor of @xmath54 in the denominator accounts for the time dilation of the co - moving grb rate . following the discussion in
[ sec : darkbias ] , we have nominally chosen the un - dust corrected sfr from @xcite ( sf1 ) .
we also explore the two other models for @xmath49 in @xcite . in our model
we do not take into account any biases related to the distance of the source in localization ( in [ sec : darkbias ] we argued that there should be little bias in triggering based on distance ) .
however , because of the observed anti - correlation between jet opening angles and prompt emission fluence @xcite , and the theoretical correlation between prompt burst emission and afterglow luminosity @xcite , aside from dust obscuration , the afterglow from _ triggered _ grbs that originate from higher redshift grbs should not be significantly dimmer at the same observer time ( see also @xcite ; @xcite ) .
nevertheless , if there is any dependence of triggering / localization on @xmath55 , we can try to account for such using a probability function related to luminosity distance , @xmath56 .
there is also a clear effect of distance upon redshift discovery : once a burst afterglow is localized and a spectrum is acquired , the detectability of emission features is diminished with increasing luminosity distance ( nominally as @xmath57 ) .
systematically , only grb hosts with higher rates of unobscured star formation will be detectable at higher redshifts for a given integration time .
for a given afterglow brightness , the distance bias does not exist for absorption - redshift grbs as long as the particular redshifted line is observable in the spectrum . to try to account for these elusive effects ,
we take the probability of redshift discovery due to distance [ @xmath56 ] as unity from @xmath59 to @xmath60 , decreasing with @xmath55 to some power @xmath61 .
nominally we take the values of @xmath62 and @xmath63 but allow these quantity to vary in our modeling ( see [ sec : sfr ] ) . using the observability of emission and absorption lines in the spectral range and in the presence of night skylines ,
we construct a relative probability of redshift detection , @xmath64 .
we take the spectrum range as 38009800 and assume a redshifted line is not observable if it falls within a wavelength that is half the instrumental resolution of a strong sky line listed in @xcite . here , we assume the instrumental resolution of 5 ( dispersion of @xmath242.5 / pixel ) . based upon table 1 , we assume that a redshift can be obtained unambiguously with the detection of at least one of four star - formation emission features : ly
@xmath2 @xmath65 , [ o ii ] @xmath66 , h@xmath67 , and [ o iii ] @xmath68 @xmath69 . also based on table 1 , for redshifts based upon absorption features only , we require that at least three of the following lines are detectable : fe ii @xmath702344.2 , fe ii @xmath70 2374.5 , fe ii @xmath70 2382.8 , fe ii @xmath70 2586.7 , fe ii @xmath70 2600 , mg ii @xmath68 2796.4 , 2803.5 , mg i @xmath70 2853.0 .
if no redshifted star formation line nor prominent absorption line is observable at a given redshift , @xmath71 , then we set @xmath72 .
even if a line is observable , it may be too weak in emission or too low in equivalent width to be detected .
therefore with less and less lines observable at a given @xmath12 , the value of @xmath64 should be diminished .
figure [ fig : model2 ] shows our toy model for @xmath64 , based upon the above considerations .
we stress that this is a toy model for ground - based optical spectroscopy .
different instruments will provide free - spectral ranges that differ from our nominal range ( 38009800 ) .
the resolution of each instrument also will also particularly affect the observability of faint , narrow emission / absorption features in the presence of sky lines . a moderate - resolution spectrograph ( @xmath73 )
should allow for feature detection closer in wavelength to a sky line ; even features which partially overlap a skyline could be detectable ( e.g. , * ? ? ?
the resultant relative probability distribution of detecting an optical redshift for a triggered grb is constructed as @xmath74 and is depicted in figure [ fig : model ] .
the relative probability distribution was calculated in bins of @xmath75 . as can be seen in the normalized cumulative distributions shown in figure [ fig : kss ] , the overall shape of the cumulative distributions are largely unaffected by the sky lines .
the most pronounced feature in @xmath48 is the drop in detection probability from @xmath76 , due to the inaccessibility of strong emission lines in the optical bandpass .
the relative probability for detection of bursts at the redshift of grb980425 ( @xmath77 ) is exceedingly small and thus serves to emphasize the distinction between grb980425 and the other long - duration grbs with known redshift ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
as such , we have not included grb980425 in our analysis .
as shown in figure [ fig : kss ] , the large - scale shape of the observed redshift distribution is adequately described by the model for @xmath78 , with the ks probability that the observed deviations are consistent with a random selection from the model of @xmath79 ( sf1 ) , @xmath80 ( sf2 ) , and @xmath81 ( sf3 ) .
models where no account for the observational bias of detecting emission lines from high - redshift grb hosts ( @xmath82 ) are clearly ruled out .
while there are small differences in the ks probability between various models for the star - formation rate , such models can not be statistically distinguished by the current sample of grbs with known redshifts .
however , given the low ks probabilities for @xmath82 , the data do support the notion that the observational biases of detecting emission lines from high - redshift grb hosts must be taken in account .
note that a fairly large range of @xmath61 values ( @xmath83 to @xmath84 for sf1 ) yields a ks consistency with the data .
how sensitive are these results to our toy model for @xmath64 and @xmath56 ? changing the value of @xmath60 we still get acceptable ( but lower ) ks statistics for @xmath78 : @xmath85 = 0.20 ( @xmath86 ; sf1 ) , 0.22 ( @xmath87 ; sf1 ) , 0.04 ( @xmath88 ; sf1 ) . with @xmath89 ,
the @xmath78 case is effectively equivalent to the @xmath82 case ( @xmath90 for @xmath91 ; sf1 ) . by removing the effect of the sky lines and limited free - spectral range in redshift determination , the ks statistics are still acceptable , albeit with lower values of the ks statistic : @xmath92 ( sf1 ) , 0.19 ( sf2 ) , 0.10 ( sf3 ) .
assuming that grbs _ do _ trace the global star - formation rate we fix the form of @xmath56 that gives a reasonable agreement with all three sf models .
from here , we can test the significance of the small - scale clustering properties . to do so
, we produce a monte carlo realization as sets of grb redshifts drawn from the probability for redshift detection @xmath48 for each sfr model .
we simulated 5000 iterations of sets of grb redshifts .
for each iteration , 26 bursts were selected uniformly from [ 0,1 ] and then mapped to redshift using the cumulative and normalized distribution of @xmath48 , interpolating between bins to increase the resolution .
we thought of no obvious existing statistic to compare the small - scale redshift structure of the observed distribution with the monte carlo set .
however , since one feature is the existence of such close redshift pairs we can ask how often such numbers of pairs are found in the distribution .
table [ tab : res ] summarizes the results of the comparison .
column two lists the number of observed grbs within an apparent recession velocity of @xmath93 of another grb .
columns three , four and five give the probability of such occurrences in the simulated distributions for the three different grb rates , @xmath49 . following the table
, the simulation predicts that at least two bursts ( one pair ) should occur by random chance in 53% of real world samples for the smallest observed velocity difference .
this is close to the probability obtained in our simplistic calculation ( 59% ) in [ sec : simple ] .
for @xmath94 values near 1000 km s@xmath0 , the probability drops to @xmath9519% before rising again at large velocity offsets .
therefore , despite the apparent close pairings in redshift space , we find no significant small - scale redshift clustering in the present sample .
note , however , this comparison only references 2-point correlations .
following column eight of table [ tab : z ] and column 6 of table [ tab : z1 ] , there are two groupings of three bursts within 2500 km s@xmath0 ( at @xmath96 and @xmath97 ) . using the monte carlo set , the probability of getting 2 groupings of three bursts within 2500 km s@xmath0 is 0.24 ( sf1 ) , 0.17 ( sf2 ) , and 0.13 ( sf3 ) .
again , the present sample shows no evidence for significant multi - burst clustering . with the advent of _ swift _ , it is not unreasonable to expect upwards of one hundred new grb redshifts within the next several years . as the density of redshifts increases , the number of close redshift pairs ( and triples ) will certainly increase .
commensurately , for an expanded sample to yield significant clustering results , the number of close pairs must increase . using the present models , as a function of bursts in a sample , we predict the required number of observed close pairs in order to be considered a significant excess above a random sample . figure [ fig : future ]
depicts the results . for the present sample , approximately 1 ( 2 ) more close pairs would be needed with velocity differences less than 1000 km s@xmath0 , for the clustering results to be considered significant at the 0.05 ( 0.01 ) level . from the figure
, we see that in a sample of 50 grb redshifts , there is a 5% random chance of getting 5 close pairs within about 250 km s@xmath0 .
if 6 or 7 more pairs are found , this would be considered a significant result .
we have demonstrated , via a ks test , that the large - scale distribution of observed long - duration grb redshifts is compatible with having been drawn from a reasonable model for the detection probability of burst redshifts .
this supports the suggestion , based upon @xmath3-ray properties , that the rate of grbs rises rapidly out to redshifts of order unity ( cf . * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
to our knowledge , this is the first investigation to claim this using only the observed grb redshifts and a model for the observational and intrinsic selection function for grb redshift discovery .
the data are relatively insensitive to various forms of the underlying star - formation for a given set of redshift selection functions @xmath64 and @xmath56 .
we have shown that by not considering the biases of ground - based optical spectroscopy in redshift determination ( i.e. , setting @xmath98 ) the large - scale redshift distributions are still consistent with the sfr expectations , although at a lower significance .
( without the presence of sky - lines , a derivation of the functional form of @xmath64 will be significantly more tractable with redshifts obtained from space - based spectroscopy , such as from _
swift_. ) the large - scale rate results are much more sensitive to @xmath56 , which is unfortunately the most difficult of the relevant probabilities to determine _ ab initio _ ; in general @xmath56 should be constructed on a case by case basis , including any biases of distance upon triggering , localization probabilities , and redshift determination .
there are certain psychological elements in the redshift determination ( e.g. , integrating longer until a redshift is found ) that make this component particularly difficult to model .
it is clear , since the large - scale ks probability drops precipitously for @xmath99 , that there must be some strong diminution of the detection rate of redshifts at larger distances . given these difficulties , since we can not test which star - formation rate grbs trace best , we use the notion that grbs _
should _ trace the universal sfr as a point of departure .
that is , we fix a functional form of @xmath48 that provides a reasonable agreement with the expectations for the grb rate on a large scale . under this assumption ( @xmath78 and @xmath100 ) , we test the significance of the apparent small - scale clustering . as can be seen in table [ tab : res ] , there does not appear to be any significant small - scale clustering of grbs in redshift space when compared with the monte carlo set ; however , the observed number of bursts paired with @xmath101 km s@xmath0 occurs in only one out of five iterations .
the small - scale comparison in table [ tab : res ] is rather conservative , however .
the only obvious _ a posteriori _ injection to the comparison is in choosing the velocity bins based upon the observed dataset .
we have not required close pairs in the comparison set to fall within a certain redshift range , say @xmath102 $ ] , nor have we required that the close pairs in the comparison sets be spaced by at least @xmath103 ( as observed ) .
any such restrictions would tend to increase the significance of the observed small - scale clustering .
however , without some _ a priori _ hypothesis that the particular redshifts and spacings in the observed datasets are of interest , further restrictions are unwarranted . in the context of cosmologies with oscillating
hubble `` constants '' ( see 1 ) some _ a priori _ preference for special redshifts may indeed exist .
for example , we might expect that the most massive clusters of galaxies would reside at redshifts where the hubble constant loiters at a local minimum .
one of the most , if not the most massive clusters known , with the highest observed x - ray temperature and high velocity dispersion of member galaxies ( @xmath104 km s@xmath0 ) , is @xcite . correcting the systemic redshift reported in @xcite to the local hubble frame ( following [ sec : correctz ] )
, the redshift of the cluster is @xmath105 .
this is just 177 @xmath106 125 km s@xmath0 from the redshift of grb970508 yet the cluster and the grb site are separated by @xmath107 on the sky .
further , the redshift of is 1456 and 1872 km s@xmath0 from grb990705 and grb000210 ( respectively ) , less than two times the velocity dispersion of the cluster itself .
using our monte carlo simulation with @xmath108 sf1 , the probability that three bursts out of 26 would fall within 2000 km s@xmath0 of @xmath109 by random chance is 0.035 . since the association with a massive cluster
was chosen _ a posteriori _ we do not claim this to be a significant result , but it should be of interest to test the association of new grb redshifts with other massive clusters at moderate to high redshifts .
the existence and expectation of several close groupings in redshift space holds some practical implications for observations of future grbs .
first , precise redshift determinations via moderate - resolution optical spectroscopy ( @xmath110 ) will continue to be important even though approximate redshifts , using grb distance - indicators ( such as the lag - luminosity or variability relations ) , may someday be used reliably to constrain the overall large - scale redshift distribution .
second , new future close redshift groupings will enable efficient detailed narrow - band imaging studies of multiple grb hosts , such as the study of grb000926 and grb000301c @xcite .
we end by noting the curious absence of detected grb redshifts in the redshift range between @xmath111 , where a relatively clean window for lyman @xmath2 emission detection exists at optical wavelengths ( see fig . [
fig : pairs ] ) .
one explanation for this is that redshift discovery via ly @xmath2 emission should be fruitless for about half of high redshift grbs , since the average equivalent width of ly @xmath2 for lyman break galaxies beyond @xmath112 is near zero ( k. adelberger , personal communication ) ; the absence of bursts in this range could simply be due to the low number of grbs followed - up with sufficiently small delay to detect lyman @xmath2 in absorption at optical wavelengths . in this respect , the uniformity and sheer rate of redshift discovery with _
, should be most enlightening . , s. g. _ et al .
_ 2001 , in gamma - ray bursts in the afterglow era , proceedings of the international workshop held in rome , cnr headquarters , 1720 october , 2000 , ed .
e. costa , f. frontera , & j. hjorth ( berlin heidelberg : springer ) , 218 , i. , vreeswijk , p. m. , kaper , l. , rol , e. , van den heuvel , e. p. j. , tanvir , n. , and ellison , s. 2002 , in lighthouses of the universe : the most luminous celestial objects and their use for cosmology proceedings of the mpa / eso , 197 the author thanks d. frail and p. van dokkum for helpful suggestions in improving the paper and gratefully acknowledges discussions with k. adelberger at several stages of this project . the anonymous referee is acknowledged and thanked for their insightful comments and suggestions .
the author is supported by a junior fellowship to the harvard society of fellows and by a generous research grant from the harvard - smithsonian center for astrophysics .
lcccccrrl 011121 & & 69.1 & 130.0 & em & & 0.362(1 ) & 0.363(1 ) & + 990712 & & 99.4 & -59.4 & em / abs & & 0.4331(2 ) & 0.4328(2 ) & + 010921 & & 145.8 & -302.0 & em & & 0.4509(4 ) & 0.4494(4 ) & + 020405 & & 45.6 & 255.5 & em & & 0.68986(4 ) & 0.69130(8 ) & + 970228 & & 94.1 & -26.3 & em & & 0.6950(3 ) & 0.6949(3 ) & + 991208 & & 88.4 & 10.2 & em & & 0.7055(5 ) & 0.7056(5 ) & + 970508 & & 92.3 & -14.7 & em / abs & & 0.8349(3 ) & 0.8348(3 ) & + 990705 & & 83.6 & 40.6 & em & & 0.8424(2 ) & 0.8426(2 ) & + 000210 & & 119.0 & -176.8 & em & & 0.8463(2 ) & 0.8452(2 ) & + 970828 & & 103.1 & -82.5 & em & & 0.9578(1 ) & 0.9573(1 ) & + 980703 & & 168.4 & -357.6 & em / abs & & 0.9662(2 ) & 0.9639(2 ) & + 991216 & & 92.2 & -14.0 & em / abs & & 1.02(2 ) & 1.02(1 ) & + 000911 & & 134.0 & -253.5 & em & & 1.0585(1 ) & 1.0568(1 ) & + 980613 & & 78.9 & 70.0 & em & & 1.0969(2 ) & 1.0974(2 ) & + 000418 & & 32.4 & 308.2 & em & & 1.1181(1 ) & 1.1203(1 ) & + 020813 & & 122.7 & -197.3 & abs & & 1.254(2 ) & 1.253(2 ) & + 990506 & & 22.1 & 338.2 & em & & 1.306576(42 ) & 1.309180(135 ) & + 010222 & & 70.4 & 122.4 & abs & & 1.4768(2 ) & 1.4778(2 ) & + 990123 & & 76.5 & 85.1 & abs & & 1.6004(8 ) & 1.6011(8 ) & + 990510 & & 75.4 & 91.9 & abs & & 1.6187(15 ) & 1.6195(15 ) & + 000301c & & 82.1 & 50.1 & em / abs & & 2.0335(3 ) & 2.0340(3 ) & + 000926 & & 94.1 & -26.2 & abs & & 2.0369(6 ) & 2.0366(7 ) & + 011211 & & 15.0 & 352.6 & abs & & 2.140(1 ) & 2.144(1 ) & + 021004 & & 158.4 & -339.4 & em & & 2.332(1 ) & 2.328(1 ) & + 971214 & & 72.6 & 109.1 & em & & 3.42(1 ) & 3.42(1 ) & + 000131 & & 75.2 & 93.2 & abs & & 4.511(2 ) & 4.513(2 ) & lcccrr grb 011121 & 0.363(1 ) & grb 990712 & 30.3 & 0.070(1 ) & 15053 @xmath106 223 + grb 990712 & 0.4328(2 ) & grb 010921 & 114.4 & 0.0166(4 ) & 3457 @xmath106 93 + grb 010921 & 0.4494(4 ) & grb 990712 & 114.4 & 0.0166(4 ) & 3457 @xmath106 93 + grb 020405 & 0.69130(8 ) & grb
970228 & 133.4 & 0.0036(3 ) & 628 @xmath106 55 + grb 970228 & 0.6949(3 ) & grb 020405 & 133.4 & 0.0036(3 ) & 628 @xmath106 55 + grb 991208 & 0.7056(5 ) & grb 970228 & 121.4 & 0.0107(5 ) & 1887 @xmath106 102 + grb 970508 & 0.8348(3 ) & grb 990705 & 152.1 & 0.0078(3 ) & 1278 @xmath106 59 + grb 990705 & 0.8426(2 ) & grb 000210 & 39.0 & 0.0026(2 ) & 416 @xmath106 47 + grb 000210 & 0.8452(2 ) & grb 990705 & 39.0 & 0.0026(2 ) & 416 @xmath106 47 + grb 970828 & 0.9573(1 ) & grb 980703 & 81.4 & 0.0066(2 ) & 1008 @xmath106 38 + grb 980703 & 0.9639(2 ) & grb 970828 & 81.4 & 0.0066(2 ) & 1008 @xmath106 38 + grb 991216 & 1.02(1 ) & grb 000911 & 42.3 & 0.04(1 ) & 5419 @xmath106 2967 + grb 000911 & 1.0568(1 ) & grb 991216 & 42.3 & 0.04(1 ) & 5419 @xmath106 2967 + grb 980613 & 1.0974(2 ) & grb 000418 & 54.6 & 0.0229(2 ) & 3253 @xmath106 35 + grb 000418 & 1.1203(1 ) & grb 980613 & 54.6 & 0.0229(2 ) & 3253 @xmath106 35 + grb 020813 & 1.253(2 ) & grb 990506 & 104.1 & 0.057(2 ) & 7446 @xmath106 266 + grb 990506 & 1.309180(135 ) & grb 020813 & 104.1 & 0.057(2 ) & 7446 @xmath106 266 + grb 010222 & 1.4778(2 ) & grb 990123 & 6.2 & 0.1233(8 ) & 14550 @xmath106 95 + grb 990123 & 1.6011(8 ) & grb 990510 & 126.2 & 0.0184(17 ) & 2109 @xmath106 195 + grb 990510 & 1.6195(15 ) & grb 990123 & 126.2 & 0.0184(17 ) & 2109 @xmath106 195 + grb 000301c &
2.0340(3 ) & grb 000926 & 23.8 & 0.0026(7 ) & 259 @xmath106 75 + grb 000926 & 2.0366(7 ) & grb 000301c & 23.8 & 0.0026(7 ) & 259 @xmath106 75 + grb 011211 & 2.144(1 ) & grb 000926 & 105.4 & 0.107(1 ) &
10383 @xmath106 119 + grb 021004 & 2.328(1 ) & grb 011211 & 163.0 & 0.185(1 ) & 17082 @xmath106 132 + grb 971214 & 3.42(1 ) & grb
000131 & 134.1 & 1.09(1 ) & 65197 @xmath106 654 + grb 000131 & 4.513(2 ) & grb 971214 & 134.1 & 1.09(1 ) & 65197 @xmath106 654 lccccc 259 & 2 & 0.530 & 0.471 & 0.438 & 0.408 + 416 & 4 & 0.303 & 0.247 & 0.215 & 0.252 + 628 & 6 & 0.206 & 0.145 & 0.120 & 0.191 + 1008 & 8 & 0.187 & 0.126 & 0.096 & 0.171 + 1278 & 9 & 0.192 & 0.120 & 0.088 & 0.160 + 1997 & 10 & 0.402 & 0.275 & 0.213 & 0.110 + 2109 & 12 & 0.225 & 0.133 & 0.093 & 0.099 + 3253 & 14 & 0.373 & 0.222 & 0.159 & 0.034 + 3457 & 16 & 0.199 & 0.092 & 0.060 & 0.026 + 5419 & 18 & 0.364 & 0.199 & 0.135 & 0.002 + 7446 & 20 & 0.401 & 0.230 & 0.165 & 6e-5 + | the 26 long - duration gamma - ray bursts ( grbs ) with known redshifts form a distinct cosmological set , selected differently than other cosmological probes such as quasars and galaxies . since the progenitors are now believed to be connected with active star - formation and since burst emission penetrates dust , one hope is that with a uniformly - selected sample , the large - scale redshift distribution of grbs can help constrain the star - formation history of the universe .
however , we show that strong observational biases in ground - based redshift discovery hamper a clean determination of the large - scale grb rate and hence the connection of grbs to the star formation history .
we then focus on the properties of the small - scale ( clustering ) distribution of grb redshifts . when corrected for heliocentric motion relative to the local hubble flow ,
the observed redshifts appear to show a propensity for clustering : eight of 26 grbs occurred within a recession velocity difference of 1000 km s@xmath0 of another grb .
that is , four pairs of grbs occurred within 30 @xmath1 myr in cosmic time , despite being causally separated on the sky .
we investigate the significance of this clustering using a simulation that accounts for at least some of the strong observational and intrinsic biases in redshift discovery .
comparison of the numbers of close redshift pairs expected from the simulation with that observed shows no significant small - scale clustering excess in the present sample ; however , the four close pairs occur only in about twenty percent of the simulated datasets ( the precise significance of the clustering is dependent upon the modeled biases )
. we conclude with some impetuses and suggestions for future precise grb redshift measurements . |
we thank m.v . berry for encouraging us to look for a special function that characterizes the crossing and e. berg , a. elgart and l. sadun for helpful suggestions .
we thank c. a. mead for useful comments .
this research was supported in part by the israel science foundation , the fund for promotion of research at the technion and the dfg .
10 m. born and r. oppenheimer , ann .
( leipzig ) * 84 * , 457 ( 1927 ) p. aventini and r. seiler , comm . math .
phys , * 41 * , 2 , 34 ( 1975 ) ; f. bornemann , _ homogenization in time of singularly perturbed mechanical systems _ , lecture notes in mathematics 1687 , springer ( 1998 ) ; j.m .
combes , p. duclos and r. seiler , in _ rigorous atomic and molecular physics _ , g. velo and a. wightman ed . , plenum ( 1981 ) g.a .
hagedorn , ann .
h. poincar * 47*,1 - 16,(1987 ) ; m. klein , a. martinez , r. seiler and x.p .
wang , comm .
144 * , 607 - 639 ( 1992 ) m.v .
berry , proc .
* a392 * 45 ( 1984 ) h. c. longuet - higgins , u. pik , m. h. l. pryce , and r. a. sack , proc .
london , ser .
a * 244 * , 1 ( 1958 ) . c. a. mead , rev .
mod . phys . * 64 * , 51 , ( 1992 ) c. a. mead , j. chem . phys . * 78 * , 807 ( 1983 ) ; t.c .
thompson , d.g .
trulhar and c.a .
mead , j. chem .
phys . * 82 * , 2392 ( 1985);t.c .
thompson and c.a .
mead , j. chem .
phys . * 82 * , 2409 ( 1985 ) . c. a. mead and d. g. truhlar , j. chem
. phys . * 70 * , 2284 ( 1979 ) r. jackiw , int . j. mod .
a3 * ( 1988 ) a. shapere and f. wilczek , _ geometric phases in physics _ , world sceintific ,
singapore ( 1989 ) .
strictly , on should distinguish the case of an integral spin from the case of half - integrla spin where @xmath18 is quaternionic real see f. dyson , j. math .
* 3 * 140 , ( 1964 ) j. von neumann and e. p. wigner , phys . z. * 30 * ( 1929 ) , 467 , l. d. landau and e. m. lifshitz , _ quantum mechnics _
perganom press , 3-rd ed .
j. l. powell and b. crasemann , _ quantum mechanics _ , addison - wesley ( 1961 ) .
e. p. wigner , _ group theory and its application to the quantum mechanics of atomic spectra _ , n. y. academic press ( 1959 ) .
this is strictly true for @xmath103 , the case @xmath103 requires a special treatment see e.g. w. e. boyce and r. c. diprima , _ elementary differential equations _ , wiley , 6-th ed . | we consider the born - oppenheimer problem near conical intersection in two dimensions . for
energies close to the crossing energy we describe the wave function near an isotropic crossing and show that it is related to generalized hypergeometric functions @xmath0 .
this function is to a conical intersection what the airy function is to a classical turning point . as an application we calculate the anomalous zeeman shift of vibrational levels near a crossing . *
introduction . * the born - oppenheimer ( bo ) problem @xcite is concerned with the analysis of schrdinger type operators where the small electron to nucleon mass ratio , plays the role of the semiclassical parameter . @xcite .
the theory identifies distinct energy scales : the electronic scale which , in atomic units , is of order one and the scale of nuclear vibrations which is of order @xmath1 in these units .
@xmath2 is the nucleus to electron mass ratio .
the identification of the electrons as the fast degrees of freedom is central to the theory .
the clean splitting between fast and slow degrees of freedom fails near eigenvalue crossing of the electronic hamiltonian where there is strong mixing between electronic and vibrational modes .
this lies at the boundary of the conventional bo theory .
since the coupling between different electronic energy surfaces becomes infinite near a crossing , the nuclear wave function does not reduce to a solution of a scalar ( second order ) schrdinger equation .
we describe the ( double surface ) nuclear wave function near an isotropic conical crossing , for energies close to the energy of the crossing .
the strong mixing of the electronic and nuclear degrees of freedom near crossing leads an anomalous zeeman effect . to describe the anomaly recall that the zeeman splitting in molecules
is reduced compared to the zeeman splitting of atoms .
it is convenient to parameterize the reduction by a parameter @xmath3 so that the zeeman splitting is of the form ( and order ) @xmath4 with @xmath5 the external magnetic field .
the low lying vibrational levels have large reduction , @xmath6 .
this is what one expects from nuclei whose magnetic moments are by factor @xmath2 smaller than the bohr magneton .
levels _ near the crossing energy _ can have a small reduction expressed by the fact that @xmath7 .
for the isotropic situation we calculate @xmath8 , so that the zeeman shift is anomalously large , by a factor of about 2000 , than the normal zeeman splitting of molecular levels . more precisely , we find that the zeeman splitting near isotropic crossing is @xmath9 the sign @xmath10 means equality in the limit @xmath11 .
@xmath5 is proportional to the magnetic field , with a coefficient dependent only on the electronic wave functions at the crossing .
@xmath12 , a half odd integer , is the azimuthal quantum number , @xmath13 is an electronic time scale , see eq .
( [ period ] ) below .
@xmath14 is a universal dimensionless factor which is determined by the nuclear wave function near the crossing , see eq .
( [ gyro ] ) below . as we shall see @xmath15 and numerical estimates of eq .
( [ gyro ] ) give @xmath16 @xmath14 is a molecular analog of the land g factor in atoms : so , while lang g factor describes the zeeman shift due to the mixing of spin and orbital degrees of freedom , @xmath14 does it for the nuclear and electronic ones
. one can formulate the bo problem in the following way @xcite : @xmath17 where @xmath18 , the electronic hamiltonian , depends parameterically on the nuclear coordinates @xmath19 .
when time reversal is not broken , @xmath18 is a real symmetric matrix @xcite .
the wigner von neumann crossing rule @xcite says that @xmath18 has generically a crossing point for two modes of vibrations , @xmath20 .
here we shall consider the simple scenario where @xmath18 is a @xmath21 matrix of and @xmath20 .
this means that we shall treat only the restriction of the electronic hamiltonian to the two dimensional subspace spanned by the two degenerate eigenstates at the crossing point .
we shall assume that @xmath18 has a single crossing point at @xmath22 , and set the crossing energy at @xmath23 .
we further assume that the crossing is conic , that @xmath18 is isotropic about the origin , and that the @xmath19 dependence of @xmath18 is smooth near the origin .
this is a common model @xcite .
we first recall why the standard bo theory fails near a crossing .
when @xmath18 is symmetric it can be diagonalized by an _ orthogonal _
transformation @xmath24 . in the @xmath21 case , and when @xmath18 is non - degenerate , @xmath24 is uniquely determined , up to an overall sign , by requiring @xmath25 . hence , away from crossing , @xmath26 is unitarily equivalent to @xmath27 where @xmath28 are the two eigenvalues of @xmath18 .
the vector potential @xmath29 is purely off - diagonal . for linear
crossing , @xmath30 as @xmath19 surrounds the origin @xcite .
this forces a @xmath31 singularity of the vector potential for small @xmath19 .
far from the origin , to leading order in @xmath32 , the two components of the wave function decouple and are given by @xmath33 where @xmath34 is a semiclassical solution of the schrdinger equation with potential @xmath35 in the cut plane with antiperiodic boundary conditions @xcite .
this decoupling holds provided @xmath36 @xcite .
we shall henceforth denote @xmath37 , and refer to the region @xmath38 as `` far from the crossing '' .
in contrast , the divergence of the off diagonal part of @xmath39 near the crossing prevents such a decoupling near the crossing . our aim is to analyze the bo theory , to leading order in @xmath40 , near the crossing of @xmath41 .
the reason one can still hope to say something useful near the crossing is that the asymptotic form of @xmath18 near the crossing , i.e. for @xmath42 , is universal @xcite @xmath43 where @xmath44 are the pauli matrices
. we shall refer to @xmath42 as `` close to the crossing '' . notice that our notions of close and far from the crossing have a nonempty intersection .
this enables us to match the solution close to the crossing with the standard , decoupled , bo solution , eq .
( [ bo ] ) .
* the zero energy wave function close to the crossing .
* we study the wave functions near the crossing for energies close to the crossing energy .
everything to be said from now on is true in the limit of @xmath11 , to leading order in negative powers of @xmath2 .
we shall assume that zero is an eigenvalue of ( [ hbo ] ) , since there is always an eigenvalue that is close enough to zero @xcite .
it turns out to be convenient first to unitary - transform ( [ can2 ] ) with @xmath45 .
this will replace @xmath46 in ( [ can2 ] ) by @xmath47 . in this representation ,
close to the crossing the zero energy wave function satisfies approximately the differential equation @xmath48 @xmath49 stands for a two component column matrix and @xmath50 is a scaled variable . the operator @xmath51 commutes with the operator on the l.h.s . of ( [ scaled2 ] ) .
it does not have the meaning of total angular momentum , since the pauli matrices do not represent spin .
we thus consider solutions of the differential equation ( [ scaled2 ] ) which are eigenfunctions of @xmath52 with an eigenvalue @xmath12 , namely : @xmath53 @xmath54 are the polar coordinates related to @xmath55 .
@xmath12 must be _ half odd integer _
, for the wave function to be single valued . separating variables , the radial equation obtained from ( [ scaled2 ] ) , in the @xmath12-th sector , takes the form : @xmath56 of the four linearly independent solutions of eq .
( [ scaled ] ) , basically due to boundary conditions , only one linear combination is fit to represent a wave function .
we denote it by @xmath57 and its components by @xmath58 and @xmath59 .
@xmath60 is to a crossing point what the airy function is to a classical turning point @xcite : it interpolates between the near region where the wave function is intrinsically a two component spinor , and the far region where the wave function is highly oscillatory and given by eq .
( [ bo ] ) .
@xmath60 has a closed expression in terms of the generalized hypergeometric functions of the kind @xmath0 .
it has the asymptotic ( for @xmath61 ) form @xmath62 solving ( [ scaled ] ) asymptotically near the origin gives@xcite @xmath63 solving ( [ scaled ] ) asymptotically at infinity we obtain @xmath64 the four dimensional family of solutions can be parameterized by either @xmath65 or @xmath66 . requiring the solution to be bounded at the origin and at infinity means ( for positive @xmath12 ) that @xmath67 .
imposing three homogeneous conditions on a four dimensional linear space leaves us with one dimensional subspace , i.e. a certain function times an arbitrary constant .
this is the celebrated @xmath68 .
while the reason we require @xmath69 to be bounded at the origin is obvious , the reason we require it to be bounded at infinity is a bit subtle , since ( [ scaled ] ) is meaningful only close to the crossing .
however , `` close to the crossing '' , @xmath70 , extends farther and farther in terms of @xmath71 as @xmath2 gets large .
@xmath72 at ( [ infinity ] ) should be exponentially small , and can set to zero to leading order .
the solutions which are regular at the origin can be obtained from the fourth order differential equations , one for each component @xmath73 , obtained from ( [ scaled ] ) .
these are related to the differential equation that defines the generalized hypergeometric functions @xmath0 .
the two linearly independent solutions that are regular at the origin are @xcite : @xmath74 where @xmath75 are generalized hypergeometric functions @xcite .
the linear combination @xmath76 is bounded at infinity provided @xcite : @xmath77 we have set the global coefficient in front of ( [ f ] ) such that ( [ cos ] ) will be true .
formulae ( [ hyper_solutions1]-[a ] ) are correct only for positive @xmath12-s . their negative counterparts can be obtained by interchanging the lower and upper components , as can be seen from ( [ scaled ] ) .
* the anomalous zeeman shift .
* let us now turn to the zeeman shift .
the magnetic field breaks time reversal symmetry , which means that it adds an the imaginary term to @xmath18 , in the representation where @xmath18 is real .
generically the magnetic field will remove the conical intersection of @xmath18 at @xmath22 and create a gap between the two energy sheets .
the gap will be proportional to the magnetic field in atomic units , and the coefficient will be in general of order one .
we can therefore introduce the magnetic field into our model by adding the term @xmath78 to @xmath18 at ( [ can2 ] ) .
there is no harm in taking @xmath5 to be independent of @xmath19 , since only the value of @xmath5 in the origin will be of importance to us .
@xmath5 is therefore a constant , proportional to the magnetic field in atomic units .
we shall not minimally couple the magnetic field directly to the vibrations for this turns out to be a weaker effect , of order @xmath79 , while the shift mediated by the electrons , in the rotationally invariant case , is @xmath80 , as we shall see .
let us consider the case where the rest of @xmath18 , namely the @xmath81 part in ( [ can2 ] ) , does not break the full rotational symmetry of its linear part , so that @xmath12 is still a good quantum number . in the rotational invariant case , the model has , for @xmath82 , a two - fold degeneracy : the states @xmath12 and @xmath83 are degenerate . the magnetic field @xmath5 breaks this degeneracy .
the splitting is twice the zeeman shift in the energy for the two states @xmath84 move in opposite directions .
equipped with approximants to the wave function near and far from the crossing , with a degenerate perturbation theory one can calculate , to leading order in @xmath32 , the zeeman splitting and obtain ( [ main2 ] ) .
we describe this calculation in details elsewhere @xcite . here
we would only like to sketch the derivation of the power in ( [ main2 ] ) .
far from the crossing , one can neglect the vector potential in ( [ shbo ] ) .
the wkb approximant to the radial part of @xmath85 is @xmath86 where we have employed the linearity of the energy surfaces near the crossing . it is a general property of the wkb approximation , that the normalization coefficient @xmath87 is independent of @xmath2 , to leading order in @xmath32 . from ( [ cos ] ) and ( [ psi_cl ] )
one sees that the interpolation of the bo radial wave function towards the crossing is @xmath88 @xmath5 removes the degeneracy of the electronic levels at @xmath22 .
the gap created there due to @xmath5 is equal to @xmath89 . intuitively , the zeeman shift of a vibrational level will be proportional to the amount of probability density in the vicinity of the crossing times @xmath5 . by
vicinity " we mean a neighbourhood of order @xmath90 of the origin , the area of which is of order @xmath91 .
the density associated with the wave function is large there and by ( [ interp ] ) is proportional to @xmath92 .
hence the weight is proportional to @xmath93 which gives the power of the zeeman splitting in eq .
( [ main2 ] ) .
the coefficient of proportion will include an integral over the components of @xmath94 , @xcite , which gives @xmath14 @xmath95 n gives the factor @xmath96 , where @xmath13 is @xmath97 the integration is carried out between the two turning points of @xmath98 , the negative energy sheet ( [ shbo ] ) .
eq .
( [ period ] ) is proportional to the time in takes a particle with a unit ( electronic ) mass to travel classically across the potential .
it is independent of the nuclear mass , and @xmath99 has the order of magnitude of electronic energies . from the invariance under @xmath100 of @xmath101 it follows that @xmath15 .
one motivation for this work was an attempt to gain some understanding of the different status of crossing in theory and experiment .
theory puts crossing and avoided crossing in distinct baskets : conic crossings come with fractional azimuthal quantum numbers while avoided crossings come with integral quantum numbers .
in contrast , measurements of molecular spectra normally can not tell a crossing from near avoided crossing . only with _ precision _
measurements @xcite and _ precise _ quantum mechanical calculations @xcite can one tell when molecular spectra favor an interpretation in terms of crossing and half integral quantum numbers or avoided crossing with integral quantum numbers .
zeeman splitting appears to be a useful tool to study crossing .
the anomaly of crossing is characterized by a fractional power @xmath3 of the molecular reduction of the zeeman splitting , @xmath102 .
a system of choice for studying crossing is molecular trimers @xcite . since trimers are not rotationally invariant eq .
( [ main2 ] ) does not apply and we can not conclude that near crossing @xmath8 for trimers . however , it is natural to expect that the qualitative features of our results carry over also to the non - isotropic case where crossing will manifest itself by an anomalously large zeeman splitting and a _ fractional _ @xmath3 .
it is an interesting challenge to calculate or measure the value of @xmath3 for ( other ) molecular crossings , and trimers in particular . |
while a certain , partially gauge fixed , light - cone formulation of higher dimensional extended objects has , for quite some time @xcite , been known to yield a _ polynomial _
hamiltonian ( density ) , with the resulting field equations easily checkable by comparing with the corresponding ( gauge - fixed ) lagrangian equation of motions , the final step , the elimination of the dynamical longitudinal field and its canonically conjugate momentum ( up to their overall integrals ) remained including most string - theory reviews somewhat unclear concerning the reduction of the poisson - structure ( despite of @xcite , in which the string case is treated , and @xcite , in which a hamiltonian formulation of the supermembrane is given , as well as @xcite ) .
as that reduction to ( almost ) purely transversal degrees of freedom recently also turned out to be related to generalisations of the witt - virasoro algebra , and a novel dynamical symmetry @xcite , it seems a good idea to give a detailed account of the hamiltonian reduction ( cp .
theorem [ reduction theorem ] ) corresponding to the elimination of the longitudinal field .
in this section , we recall the hamiltonian formulation of the extremal volume problem for relativistic extended objects in light - cone gauge . consider a @xmath0-dimensional submanifold @xmath1 in @xmath2-dimensional minkowski space with metric @xmath3 of signature @xmath4 .
requiring that the dirac - nambu - goto action ( volume functional ) @xmath5 @xmath6 is stationary under variations of the embedding functions @xmath7 , @xmath8 , gives the equations of motion @xmath9 where @xmath10 , collectively denoted by @xmath11 , are local coordinates on @xmath1 , and @xmath12 . for a hamiltonian formulation ,
one introduces canonical momenta are written @xmath13 . ]
@xmath14 satisfying the dynamical poisson brackets is suppressed here and in the following . ]
( all other brackets zero ) @xmath15 because of the general reparameterisation invariance of the action ( [ dirac - nambu - goto action ] ) , the definition ( [ canonical momenta ] ) leads to first class primary constraints ( see also @xcite ) @xmath16 with @xmath17 after the partial gauge fixing @xmath18 the constraint @xmath19 becomes second class , @xmath20 with the canonical momenta conjugate to @xmath21 defined as @xmath22 while the constraints @xmath23 , @xmath24 , remain first class .
following @xcite , one introduces a corresponding `` dirac bracket '' ( assuming @xmath25 ) @xmath26 where @xmath27 parametrises the @xmath28-dimensional extended object .
it follows that @xmath29 after replacing all poisson brackets with the dirac brackets ( [ dirac brackets of canonical variables ] ) , the constraints @xmath19 and @xmath30 can be set strongly to zero . with no risk of confusion
, we can drop the subscript @xmath31 in ( [ dirac brackets of canonical variables ] ) and from now on refer to @xmath32 as the dynamical poisson bracket ( thus replacing ( [ canonical poisson brackets ] ) ) . the action ( [ dirac - nambu - goto action ] )
can then be written @xmath33 with the corresponding hamiltonian ( see also @xcite ) @xmath34 : = \int\limits_\sigma{\frac{{\vec{p}\,}^2 + g}{-2\pi}\ , d^m \varphi},\ ] ] where @xmath35 and the euclidean vectors @xmath36 and @xmath37 having components @xmath38 and @xmath39 , @xmath40 , respectively .
the remaining first class constraints become @xmath41 which , by eliminating @xmath42 , locally give the integrability conditions @xmath43 since the hamiltonian ( [ hamiltonian ] ) does not depend on the phase - space field @xmath42 , its conjugate momentum is conserved , @xmath44 .
we can therefore write for the solution to this equation of motion @xmath45 with a discrete dynamical variable @xmath3 satisfying @xmath46 , and a non - dynamical positive density @xmath47 normalized such that @xmath48 furthermore , ( [ constraints ] ) allows us to solve for @xmath42 ( in terms of @xmath36 , @xmath37 and @xmath3 ) up to the zero mode @xmath49 the plan of this note is to perform a further partial hamiltonian reduction such that the canonical fields @xmath42 and @xmath50 can be discarded in favour of the discrete zero mode variables @xmath51 and @xmath3 , the latter satisfying @xmath52 and with the field variables @xmath36 and @xmath37 then constrained ( locally ) by ( cf .
( [ integrability criterion ] ) ) @xmath53
in this section , we perform a partial gauge fixing corresponding to ( [ definition of eta ] ) , and introduce the primary constraint that becomes second class as a result of this gauge fixing . in the following ,
the domain of integration is always @xmath27 .
introduce a metric @xmath54 on @xmath27 such that @xmath55 and denote the non - constant eigenfunctions , respectively eigenvalues , of the corresponding laplacian on @xmath27 by @xmath56 and @xmath57 respectively , i.e. @xmath58 with @xmath59 .
the eigenfunctions are normalized such that @xmath60 resulting in @xmath61 defining @xmath62 one has @xcite @xmath63 and @xmath64 introduce the constraint functions @xmath65 with @xmath66 and ( introduced in @xcite ) @xmath67 . using the primary constraints
( [ constraints ] ) in ( [ auxiliary constraint function ] ) , integrating by parts , and applying ( [ laplace of green s function ] ) gives @xmath68 which means that @xmath69 is a primary constraint .
indeed , one easily checks that @xmath70 with @xmath23 given in ( [ constraints ] ) .
we now impose @xmath71 as an additional gauge constraint . using this
, one can solve for @xmath50 in terms of the discrete variable @xmath72 ( cf .
( [ definition of eta ] ) ) on the constraint surface .
[ lemma poisson brackets of constraint functions ] the constraints ( [ constraint functions ] ) satisfy the dynamical poisson brackets @xmath73 @xmath74 and @xmath75 the following variational derivatives of ( [ auxiliary constraint function ] ) will be needed : @xmath76 inserting ( [ constraint functions ] ) on the left hand side of ( [ poisson brackets of constraint functions 1 - 1 ] ) yields @xmath77 where we have used that @xmath78 using the definition of the poisson bracket , the variational derivatives ( [ variations of auxiliary constraint function ] ) , and integrating by parts gives @xmath79 applying ( [ integrability criterion ] ) to the expression within parenthesis in the last line above , together with the use of equations ( [ laplace of green s function ] ) and ( [ variations of auxiliary constraint function ] ) , leads to @xmath80 which proves ( [ poisson brackets of constraint functions 1 - 1 ] ) .
the poisson bracket ( [ poisson brackets of constraint functions 2 - 2 ] ) is trivial since @xmath81 only involves the field @xmath82 .
finally , ( [ poisson brackets of constraint functions 1 - 2 ] ) is easily proven using ( [ canonical poisson brackets ] ) , ( [ normalization of rho ] ) , and by noting that @xmath83 since @xmath84 does not contain @xmath42 .
in this section , we define a dirac bracket that turns the second class constraints ( [ constraint functions ] ) effectively first class . to this end
, the canonical procedure would be to set @xmath85 and then define the dirac bracket by @xmath86 however , as it stands , this expression is not well defined since ( [ constraint matrix ] ) is not invertible . more specifically , by equation ( [ poisson brackets of constraint functions 1 - 2 ] ) in lemma [ lemma poisson brackets of constraint functions ] , one finds that @xmath87 defines a projection onto the space of zero - mean real - valued functions on @xmath27 : @xmath88(\varphi ) : = \int { \chi ( \varphi , \tilde \varphi ) f(\tilde \varphi ) d^m \tilde \varphi } = f(\varphi ) - \bar f\ ] ] with the mean of @xmath89 defined by @xmath90 it follows that @xmath91 is non - trivial , and ( [ constraint matrix ] ) is indeed non - invertible .
fortunately , it turns out that the bracket obtained by restricting ( [ chi ] ) to @xmath92 , i.e. @xmath93 , will be sufficient for our purposes .
[ definition dirac bracket ] define @xmath94 this bracket has the desired properties , as seen by the following : the bracket ( [ alternative dirac bracket ] ) satisfies @xmath95 furthermore , for arbitrary functions of @xmath36 , @xmath51 , @xmath37 , and @xmath50 ( in particular @xmath3 ) , it holds that @xmath96 @xmath97 @xmath98 and @xmath99 the brackets ( [ second class requirement ] ) follow immediately using lemma [ lemma poisson brackets of constraint functions ] and the projection property of @xmath100 , while relations ( [ antisymmetry ] ) and ( [ product law ] ) are trivial .
the jacobi identity is proven along the same lines as the original proof by dirac @xcite . to simplify the algebra , we first write ( [ alternative dirac bracket ] ) symbolically as @xmath101 with @xmath102 and @xmath103 .
let @xmath104 denote cyclic permutation of 1 , 2 , 3 , and summation of the result , such that @xmath105 then @xmath106 the first two lines are zero by the jacobi identity for the poisson bracket . renaming dummy summation indices and again using the jacobi identity , the last line can be rewritten as @xmath107 by lemma [ lemma poisson brackets of constraint functions ] , this simplifies to @xmath108 when summing over indices .
but this is zero since @xmath109 for functions @xmath110 of @xmath36 , @xmath51 , @xmath37 , and @xmath50 . to prove ( [ bracket between constraint and f ] ) , we first note that @xmath111 where , in the second equality , we have used that variations of @xmath84 ( [ variations of auxiliary constraint function ] ) depend explicitly on @xmath112 .
analogous to ( [ vanishing integral of green function ] ) , all these terms vanish when one integrates with respect to @xmath113 above .
it follows that for an arbitrary function @xmath110 of @xmath36 , @xmath51 , @xmath37 , and @xmath50 , @xmath114 the hamiltonian reduction of the degrees of freedom is completed by : [ reduction theorem ] the phase space variables @xmath36 , @xmath51 , @xmath37 , and @xmath3 satisfy @xmath115 with all other brackets zero , and the reduced hamiltonian is @xmath116 = \frac{1}{2\eta } \int\limits_\sigma{\frac{{\vec{p}\,}^2 + g}{\rho}\ , d^m \varphi}.\ ] ] we apply definition [ definition dirac bracket ] ( suppressing the arguments of the field variables whenever there is no risk of confusion ) . since @xmath117 ,
the dirac brackets between the @xmath36 and @xmath37 coincide with the corresponding dynamical poisson brackets ( @xmath118 , etc . ) .
furthermore , @xmath119 the result now follows by using ( [ canonical poisson brackets ] ) , ( [ zero mode bracket ] ) , and the fact that the dynamical poisson brackets @xmath120 all vanish .
this work was supported by the gran gustafsson foundation and the knut and alice wallenberg foundation ( grant kaw 2005.0098 ) .
for the convenience of the reader , some relations between our notation and another one often used in the literature are summarised in this appendix . taking the metric signature @xmath121 ,
we have followed here the convention that light - cone coordinates are defined by @xmath122 and the scalar product is given by @xmath123 on the other hand , it is also common in the literature to use reversed metric signature @xmath124 , with the convention @xmath125 and scalar product @xmath126 so that a translation between notations is given by @xmath127 after gauge fixing @xmath128 , the conjugate momenta become ( the expression for @xmath129 following from the constraint @xmath19 in ( [ all primary constraints ] ) ) @xmath130 and the light - cone hamiltonian @xmath131 the zero mode variables remaining after the reduction of the phase space are @xmath132 with @xmath133 .
the mass - squared of the theory is given by @xmath134 99 j. hoppe , quantum theory of a massless relativistic surface and a two - dimensional bound state problem , ph.d .
thesis , mit 1982 ; + http://dspace.mit.edu/handle/1721.1/15717 a. hanson , t. regge , c. teitelboim , constrained hamiltonian systems , accademia nazionale dei lincei ( 1976 ) e. bergshoeff , e. sezgin , y. tanii , hamiltonian formulation of the supermembrane , nucl .
b. * 298 * , 187 ( 1988 ) j. hoppe , membranes and matrix models , arxiv : hep - th/0206192 , ihes / p/02/47 j. hoppe , fundamental structures of m(brane ) theory , arxiv:1003.5189 j. de woul , j. hoppe , d. lundholm and m. sundin , a dynamical symmetry for supermembranes , arxiv:1004.0266 j. hoppe , quantum reconstruction algebras , + http://urn.kb.se/resolve?urn=urn:nbn:se:kth:diva-13591 + ( submitted to arxiv may 20 2010 ) j. hoppe , matrix models and lorentz invariance ; manuscript / lectures given at kth in march 2010 ; to appear p. a. m. dirac , lectures on quantum mechanics , belfer graduate school of science , yeshiva university , new york ( 1964 ) v. moncrief , can one adm quantize relativistic bosonic strings and membranes ?
. gravit . * 38 * , 561 ( 2006 ) j. goldstone , unpublished notes p. a. m. dirac , generalized hamiltonian dynamics , canad .
* 2 * , 129 ( 1950 ) | the elimination of the non - transversal field in the standard light - cone formulation of higher - dimensional extended objects is formulated as a hamiltonian reduction . |
the distributions of mesons at large transverse momenta in @xmath4 or @xmath5 collisions are usually calculated in the framework of perturbative qcd using collinear factorization ( see e.g.@xcite ) .
while the shape at transverse momenta larger than 2 - 4 gev can be relatively well explained , there are discrepancies at lower transverse momenta . in this analysis
, the calculations are performed using a new approach , based on the unintegrated parton distributions . in recent years
only gluon degrees of freedom are taken explicitly in this context @xcite . in the present analysis , in addition to the @xmath6 mechanism
, we include also @xmath7 and @xmath8 mechanisms and similar ones for antiquarks , in order to obtain a fully consistent description .
the new contributions @xmath7 and @xmath8 are comparable to the contribution of the @xmath6 diagram at midrapidities and are dominant in the fragmentation region .
the new mechanisms are responisible for @xmath9 asymmetry .
a purely gluonic mechanism would lead to identical production of positively and negatively charged hadrons .
the recent results of the brahms experiment @xcite show that the @xmath10 and @xmath11 ratios differ from unity .
this put into question the successful description of ref.@xcite . in the light of this experiment ,
it becomes obvious that the large rapidity regions have more complicated flavour structure . at lower energies these ratios
are known to differ from unity drastically @xcite .
many unintegrated gluon distributions in the literature are ad hoc parametrizations of different sets of experimental data rather than derived from qcd .
recently kwieciski and collaborators @xcite have shown how to solve the so - called ccfm equations by introducing unintegrated parton distributions in the space conjugated to the transverse momenta @xcite .
we present results for pion and kaon production based on the unintegrated parton ( gluon , quark , antiquark ) distributions .
the formulae for inclusive quark / antiquark distributions are similar to the formula for @xmath6 @xcite @xmath12 these seemingly 4-dimensional integrals can be written as 2-dimensional integrals after a suitable change of variables @xcite .
the formulae can be written in the equivalent way in terms of the parton distributions in the space conjugated to the transverse momentum @xcite .
there are a few sets of fragmentation functions available in the literature , details were described e.g. in @xcite , @xcite , @xcite . the inclusive distributions of hadrons ( pions , kaons , etc .
) can be obtained through a convolution of inclusive distributions of partons and flavour - dependent fragmentation functions .
one dimensional distributions of hadrons can be obtained through the integration over the other variable ( see @xcite ) .
in fig . [ fig : pions_pt ] we compare the model invariant cross sections for @xmath13 ( left panel ) and @xmath14 ( middle panel ) as a function of pion transverse momentum at w = 27.4 gev for different values of the parameter @xmath15 of our gaussian nonperturbative form factor ( for explanation see @xcite ) with the experimental data from ref.@xcite , @xcite . in principle , the result should not exceed experimental data especially in the perturbative regime of @xmath16 2 gev where the perturbative @xmath17 parton subprocesses are crucial .
this limits the value of the nonperturbative form factor to @xmath18 0.5 gev@xmath19 .
inclusion of diagrams @xmath20 ( q@xmath21 g @xmath2 q@xmath21 ) and @xmath22 ( g q@xmath21 @xmath2 q@xmath21 ) ( see @xcite ) in conjunction with the flavour dependent fragmentation functions leads to the @xmath9 asymmetry . in the right panel of fig.[fig : pions_pt ] we show the asymmetry as the function of pion transverse momentum .
the asymmetry is well described by this model , in contrast to individual distributions .
this seems to suggest the right relative contributions of diagram @xmath23 ( gg @xmath2 g ) , @xmath20 and @xmath22 .
the asymmetry depends only weakly on the value of the parameter @xmath15 of the gaussian nonperturbative form factor .
the phenix collaboration has measured invariant cross section as a function of the @xmath24 transverse momentum at w = 200 gev in a very narrow interval of pseudorapidity @xmath25 = 0.0 @xmath26 0.15 . in fig.[fig : phenix_ff ] we show full result ( diagrams @xmath23 , @xmath20 and @xmath22 @xcite,@xcite ) for different fragmentation functions @xcite . in fig.[chk1 ]
we present dependence of the invariant cross section for kaon production , calculated for w = 27.4 gev , on the value of the parameter @xmath15 of the gaussian nonperturbative form factor .
the data for @xmath27 @xmath28 0.5 gev are well described by the @xmath3-factorisation approach with the kwieciski updfs for the parameter @xmath15 @xmath29 0.5 gev@xmath19 .
this is the same value of the paramater as that obtained for pions in the same energy range ( see fig .
[ fig : pions_pt ] ) .
more examples can be found in @xcite and @xcite .
the formalism based on updfs , which fulfill so - called kwieciski evolution equations , provides a reasonable description of the experimental data , including sps as well as recent data of the phenix , brahms and star collaborations . a good agreement with experimental data
is obtained , especially at relatively small transverse momenta and large values of pseudorapidity .
the mechanisms , which invole quark / antiquark degrees of freedom are significant and lead to an asymmetry in the production of @xmath0 and @xmath1 mesons . | inclusive cross section for pion production in proton - proton collisions are calculated based on unintegrated parton distribution functions ( updfs ) .
in addition to purely gluonic terms the present approach includes also quark degrees of freedom .
phenomenological fragmentation functions from the literature are used .
the new mechanisms are responsible for @xmath0 - @xmath1 asymmetry .
in contrast to standard collinear approach , application of 2 @xmath2 1 @xmath3 - factorization approach can be extended towards much lower transverse momenta , both at mid and forward rapidity region .
the results of the calculation are compared with sps and rhic data . |
a central and so far experimentally unexplored element of the standard model ( sm ) of particle physics is the origin of electroweak symmetry breaking . indirect evidence from precision measurements at @xmath6 colliders
suggests the existence of a light higgs boson in the mass range of @xmath10 gev at the 95% confidence level with a statistical preference towards the lower end @xcite . _ a priori _ , more complicated higgs sectors are phenomenologically just as viable .
a well known example is provided by the general two doublet higgs model ( 2dhm ) @xcite .
a constrained version of the 2dhm is in fact a part of the minimal supersymmetric extension of the sm ( mssm ) , where spontaneous symmetry breaking is induced by two complex higgs doublets leading to five physical scalars .
the lightest of these is predicted to have a mass below the @xmath11 boson although radiative corrections can soften that limit up to about @xmath12 gev due to the large value of the top mass . in the mssm
there are only two parameters in the extended higgs sector , conventionally chosen to be the ratio of the vacuum expectation values of the up and down type higgs bosons , @xmath8 , and the mass @xmath13 of the predicted neutral pseudoscalar higgs . at tree level
all other parameters are then fixed .
the main physics objectives of a future linear @xmath6 collider depend crucially on what has been discovered at either lep2 , the tevatron or the lhc .
if no higgs bosons would be discovered at any of these machines the main objective will be to perform precision tests of anomalous vector boson couplings to look for evidence of new physics at higher energy scales .
if supersymmetry is discovered , through the production of new particles for example , then it is mandatory to investigate the precise structure of its manifestation in order to hopefully have access to physics at scales where the standard model couplings become equal and even gravity enters . even if ` only '
a sm higgs is discovered , there will still be many unresolved issues concerning the validity domain and origin of sm physics .
a common feature of all these possible scenarios is that a high degree of experimental and theoretical precision will be required at a linear collider to gain deeper insight into the structure of the physical laws of nature . in this context
the photon - photon ` compton - collider ' option , from backscattered laser light off highly energetic and polarized electron beams @xcite , is a valuable ingredient .
it can provide complementary information about certain physical parameters which enter in different reactions , compared to the @xmath6 mode , at comparable event rates . in the higgs sector
, the compton collider option offers a unique way to obtain a precise determination of the partial @xmath14 width .
this quantity is important in two respects .
first , it allows for the model independent determination of the total higgs width , given that the appropriate branching ratio @xmath15 has been determined ( at the lhc for instance ) .
secondly , it is an important indicator of new physics as heavy charged particles which obtain their mass through the higgs mechanism do _ not _ decouple in the loop . at the lhc , pseudoscalar masses above @xmath16 gev are not detectable for a wide range of @xmath8 and for intermediate values of @xmath8 this regime reaches down to @xmath17 gev @xcite . in the mssm
this means that even in this so - called decoupling limit the diphoton partial higgs width could reveal the mass of the heavy pseudoscalar .
although the mssm behaves rather like the sm for heavy @xmath13 , studies suggest that the partial @xmath18 width can still differ by up to ten per cent @xcite .
it is clear from these lines of reasoning that it is very important at this stage of the design of future linear colliders to know what the photon - photon option can contribute to the high precision measurements in the search for new physics .
the purpose of this paper is therefore to investigate the level of the statistical accuracy of the determination of the partial diphoton higgs width .
we focus on the processes @xmath19 with @xmath20 , and use all the available calculations relevant to these for both the intermediate mass higgs signal as well as the continuum background .
the monte carlo results are then combined with expected tesla machine and detector design parameters to arrive at reliable predictions for the expected event rates .
we begin by summarizing the status of the qcd corrections to the tree - level processes .
in this section we begin by reviewing the qcd corrections to ( continuum ) heavy quark production in polarized photon - photon collisions .
as is well known , there are two possible @xmath22 states , and at the born level one has : @xmath23 where @xmath24 denotes the quark velocity , @xmath25 the @xmath5 c.m .
collision energy , @xmath26 the electromagnetic coupling , @xmath27 the charge of quark @xmath28 and @xmath29 its pole mass .
the scattering angle of the produced ( anti)quark relative to the beam direction is denoted by @xmath30 .
[ eq : bj0 ] clearly displays the important feature that the @xmath3 cross section has a relative @xmath31 suppression compared to the @xmath32 cross section @xcite . since the higgs @xmath33 process only occurs for @xmath34 ,
it is this polarization that is crucial for the precision measurement of the higgs partial decay width .
however the overall @xmath31 factor in the @xmath3 cross section implies that , unlike for eq .
[ eq : bj2 ] , potentially large radiative corrections with logarithmic mass singularities are _ not _ forbidden by the kinoshita - lee - nauenberg theorem @xcite .
this will become important in the next section where we discuss large double logarithmic ( dl ) corrections to this cross section .
in contrast , the qcd corrections in the @xmath32 case are not expected to be large . for both helicity configurations the exact differential one - loop corrections
are known @xcite and are included in the calculations reported in this paper . for the @xmath32 helicity configuration , which
, if we assume a high level of efficiency for producing the @xmath34 state , is expected to be heavily suppressed , these are sufficient for our purposes .
for the zero helicity configuration the large logarithmic corrections _
are _ phenomenologically important , and so we discuss them in more detail in the next section .
the dominant background to higgs production below 140 gev is @xmath35 with @xmath36 .
while this background is suppressed by @xmath37 at the born level , as shown above , higher - order qcd radiative corrections in principle remove this suppression @xcite .
in addition , large virtual non - sudakov double @xmath38 logarithms ( dl ) are present which at one - loop can even lead to a negative cross section @xcite .
the physical nature of these novel dl - effects was elucidated in ref .
@xcite . there ,
the two - loop contribution to the non - sudakov form factor was calculated and it was shown that it allows for reasonable qualitative estimates .
in particular , positivity to the cross section was restored . in ref .
@xcite the first explicit three - loop results in the dl approximation were presented which revealed a factorization of non - sudakov and sudakov double logarithms for this process and led to the all - orders resummation in the form of a confluent hypergeometric function @xmath39 : @xmath40 \right ) \label{eq : vps}\end{aligned}\ ] ] where @xmath41 is the one - loop non - sudakov form factor , @xmath42 is taken as a fixed parameter , and @xmath43 is the soft - gluon upper energy limit . in ref .
@xcite it was pointed out that one needs to include at least _ four _ loops ( at the cross section level ) of the non - sudakov logarithms in order to achieve positivity and stability . at this level of approximation
there is an additional major source of uncertainty in the scale choice of the qcd coupling , two possible ` natural ' choices @xmath44 and @xmath45 yielding very different numerical results .
however in ref .
@xcite this uncertainty was largely removed by employing the renormalization group to introduce a _ running _ qcd coupling .
we briefly summarize the results in the next section . in the derivation of the leading logarithmic corrections in ref .
@xcite the familiar sudakov technique @xcite of decomposing loop momenta into components along external momenta , denoted by @xmath46 , and those perpendicular to them , denoted by @xmath47 was used . for massless fermions , the effective scale for sudakov double logarithms for the coupling at each loop is @xmath48 , as was shown e.g. in refs .
@xcite by direct comparison with explicit higher - order calculations . for massive fermions
the effective scale is also given by @xmath49 as the dominant double logarithmic phase space is given by @xmath50 @xcite ( on a formal dl level , setting @xmath51 yields the massless sudakov form factor ) .
we use ) exactly . the next - to - leading order terms ( @xmath52 )
are included correctly through two - loops . ]
@xmath53 where @xmath54 , @xmath55 and for qcd we have @xmath56 , @xmath57 and @xmath58 as usual .
up to two - loops the massless @xmath59-function is independent of the chosen renormalization scheme and is gauge invariant in minimally subtracted schemes to all orders @xcite .
these features also hold for the renormalization group improved form factors below . for the massive sudakov form factor
we use the on - shell condition @xmath60 , even though the running coupling will now depend on _ two _ integration variables . since in this work we are only able to include the hard _ one - gluon _ matrix elements ( the exact nnlo corrections are presently unknown ) , the higher - order terms will inevitably be energy cut ( @xmath61 ) dependent . however our two - jet definition ( see below ) automatically restricts the higher - order hard gluon radiation phase space such that it is reasonable to neglect more than one gluon emission .
the complete result for the renormalization group improved massive sudakov form factor is then given by @xcite : @xmath62 assuming only @xmath63 .
expanding in @xmath64 gives the dl - sudakov form factor in eq .
[ eq : vps ] together with subleading terms proportional to @xmath65 and subsubleading terms proportional to @xmath66 entering into eq .
[ eq : rpvrgex ] is not related to the two - loop @xmath59-function coefficient . ] .
we emphasize that the two - loop running coupling is included in eq .
[ eq : rpvrgex ] to all orders and that all collinear divergences are avoided by keeping all non - homogeneous fermion mass terms . in ref .
@xcite it was shown that eq .
[ eq : rpvrgex ] exponentiates in analogy to the soft exponential term in eq . [ eq : vps ] .
the reason for adopting the above soft gluon energy regulator rather than the more conventional @xmath67 invariant mass prescription ( see refs .
@xcite for instance ) is connected with the straightforward dl - phase space of the _ massive _ sudakov form factor .
the latter is thus convenient for the inclusion of the renormalization group effects as outlined above .
more details are given in ref .
@xcite .
we next turn to the virtual non - sudakov dl corrections and investigate the rg effects for these contributions . here
we use the scale @xmath68 directly as the graphs in fig .
[ fig : rg ] are on a dl level identical to the sudakov topology up to the last integration over the ( regulating ) fermion line .
this last integration , however , does not renormalize the coupling . for the non - sudakov topologies depicted in fig .
[ fig : rg ] we find , after an order - by - order integration over the appropriate running coupling for the complete virtual renormalization group improved non - sudakov form factor : @xmath69 and thus for the rg - improved virtual plus soft real cross section we have @xmath70 + where the rg - improved massive sudakov form factor is given in eq .
[ eq : rpvrgex ] .
the effect of the renormalization group improved virtual plus soft real bremsstrahlung cross section is depicted in fig .
[ fig : vps ] .
the rg - improved cross section obtained using eq .
[ eq : srg ] lies between the theoretically allowed upper and lower limits given by the double logarithmic form factor of eq .
[ eq : vps ] with @xmath42 evaluated at the bottom mass and the higgs mass scale . for a lower value of the energy cutoff @xmath61 the background is more suppressed but the higher order ( uncanceled ) @xmath61-dependence is stronger .
this latter technical problem can be reduced by identifying @xmath61 with the physical energy cutoff of the detector efficiencies .
this will be discussed in section [ sec : nr ] .
an intermediate mass higgs boson has a very narrow total decay width .
it is therefore appropriate to compare the total number of higgs signal events with the number of ( continuum ) background events integrated over a narrow energy window around the higgs mass .
the size of this window depends on the level of monochromaticity that can be achieved for the polarized photon beams . in general ,
the number of events for the ( signal ) process @xmath72 is given by @xmath73 where @xmath74 denotes the center of mass energy .
for @xmath75 we have the following breit - wigner cross section , e.g. refs .
@xcite : @xmath76 where the conversion factor @xmath77 fb gev@xmath78 . in the narrow width approximation
we then find for the expected number of events @xmath79 to quantify this , we take the design parameters of the proposed tesla linear collider @xcite , which correspond to an integrated peak @xmath80-luminosity of 15 fb@xmath81 for the low energy running of the compton collider .
the polarizations of the incident electron beams and the laser photons are chosen such that the product of the helicities @xmath82 .
this ensures high monochromaticity and polarization of the photon beams @xcite . within this scenario
a typical resolution of the higgs mass is about 10 gev , so that for comparison with the background process @xmath83 one can use @xcite : @xmath84 with @xmath850.5 fb@xmath81/gev .
the number of background events is then given by @xmath86 in other words , the number of signal events is proportional to @xmath87 while the number of continuum heavy quark production events is proportional to @xmath88 . in principle
it is possible to use the exact compton profile of the backscattered photons to obtain the full luminosity distributions . the number of expected events is then given as a convolution of the energy dependent luminosity and the cross sections .
our approach described above corresponds to an effective description of these convolutions , since these functions are not precisely known at present .
note that the functional forms currently used generally assume that only one scattering takes place for each photon , which may not be realistic .
once the exact luminosity functions are experimentally determined it is of course trivial to incorporate them into a monte carlo program containing the physics described in this paper .
we next summarize the radiative corrections entering into the calculation of the expected number of higgs events .
for the quantity @xmath14 there are three main standard model contributions , depicted in fig .
[ fig : h ] : the @xmath89 and @xmath90- and @xmath7-quark loops .
we include these at the one - loop level , since the radiative corrections are significant only for the @xmath7-quark @xcite .
the branching ratio @xmath91 is treated in the following way .
the first component consists of the partial @xmath92 width .
obviously we must use the same two - jet criterion for the signal as for the background . for our purposes a cone - type algorithm is most suitable , and so we use the sterman - weinberg two - jet definition @xcite depicted in fig .
[ fig : jet ] .
note that the signal cross section is corrected by the _
same _ resummed renormalization group improved form factor given in eq .
[ eq : rpvrgex ] , since this factor does not depend on the spin of the particle coupling to the final state quark anti - quark pair .
in addition we use the exact one - loop corrections from ref .
these revealed that the largest radiative corrections are well described by using the running quark mass evaluated at the higgs mass scale .
we therefore resum the leading running mass terms to all orders . for the real bremsstrahlung corrections we use our own @xmath93 matrix elements .
an important check is obtained by integrating over all phase space and reproducing the analytical results of ref .
@xcite .
the second quantity entering the branching ratio is the total higgs width @xmath94 .
here we use the known results summarized in ref .
@xcite , and include the partial higgs to @xmath95 , @xmath96 , @xmath97 , @xmath98 , @xmath99 and @xmath100 decay widths with all relevant radiative corrections .
@xcite numerical predictions were given for an ( infra - red safe ) two - jet @xmath101 cross section in @xmath5 collisions in the energy range @xmath102 gev . a modified sterman - weinberg cone definition , imposed on the final state partons was employed .
thus , at leading order ( i.e. @xmath103 ) all events obviously satisfy the two@xmath7jet requirement . to ensure that both jets lie in the central region of the detector . ]
this defines our ` leading order ' ( lo ) cross section . at next to leading order ( nlo ) we can have virtual or real gluon emission .
for the latter , an event is defined as two@xmath7jet like if the emitted gluon @xmath104 we further subdivided region i according to whether the gluon energy is greater or less than the infrared cutoff @xmath61 ( @xmath105 ) .
adding the virtual gluon corrections to this latter ( soft ) contribution , to give @xmath106 , and calling the remaining hard gluon contribution @xmath107 , we have @xmath108 in ref .
@xcite each part of this cross section was evaluated exactly to @xmath109 and in addition the resummed non - sudakov form factor was included in @xmath106 .
this was necessary to yield a positive cross section .
we use the rg - improved expressions for the resummed form factors .
thus @xmath110 where @xmath111 is given in eq .
[ eq : srg ] and @xmath112 is the exact one - loop result minus the one - loop leading - logarithm pieces which are resummed in @xmath113 , i.e. @xmath114 \ ; .
\nonumber \\ & & \end{aligned}\ ] ] by adding the second ( sudakov ) piece in the square brackets we remove ( at least up to terms @xmath115 ) the dependence on the gluon energy cutoff @xmath61 .
note also that the complete expression for the two - jet cross section ( with the remaining @xmath61 dependence displayed ) @xmath116 contains a mixture of exact @xmath109 and resummed pieces . for the former ,
we use @xmath117 as the scale for @xmath42 .
such that @xmath118 for @xmath119 gev , at both leading and next - to - leading order . ]
the resummed contributions are based on the scale choice @xmath120 in the loops , as already discussed . before computing and combining the various components of the two - jet cross section in eq .
[ eq:2jetbis ] we must address the issue of the dependence on the unphysical infra - red parameter @xmath61 . if we were to expand out the resummed rg - improved form factor @xmath121 in powers of @xmath45 , and retain only the @xmath122 term , we would find that the @xmath61 dependence exactly canceled that of @xmath123 .
however in the full resummed expression , there is nothing to cancel the explicit @xmath61 dependence at higher - orders .
the canceling terms would come from the as yet unknown higher - order contributions to @xmath107 .
faced with this dilemma , we have several choices .
we could , as in @xcite , neglect the higher - order terms in the sudakov form factor altogether , and include only the non - sudakov form factor which is of course independent of @xmath61 .
furthermore , as shown in @xcite with the choice @xmath124 , the combined contribution of virtual gluons and real gluons with @xmath125 to @xmath126 was dominated by the non - sudakov ` @xmath127 ' part .
this suggests that the most reasonable procedure for the resummed cross section is to take @xmath128 and to vary @xmath129 .
we stress that this is an _ approximation _ , since it corresponds to making an assumption about the contribution of real multi - gluon emission with energies @xmath130 . as our ` best guess ' rg - improved , resummed two - jet cross section , therefore
, we have @xmath131 at this point it is appropriate to comment on detector and accelerator related issues which were adopted in our analyses @xcite . since we are not using a full detector simulation we employ effective performance parameters which should be achievable at a future linear collider .
we will display results for realistic scenarios for both currently accepted and more optimistic cases .
in particular the double b - tagging efficiency will be assumed to be @xmath132 throughout and the main input parameters concern the probability of counting a @xmath96-pair as @xmath95 and the ratio of the photon - pairs in a @xmath3 to @xmath133 state .
we emphasize again that these dependences are in real machine environment given by functional forms which can easily be determined through test runs at a later stage .
for our purposes here the effective description is sufficient .
the results discussed in the next section contain all radiative corrections summarized above .
the goal is to optimize the jet - parameters of the sterman - weinberg two - jet definition in order to maximize the statistical significance of the intermediate mass higgs - boson signal .
we begin with a few generic remarks concerning the uncertainties in our predictions .
the signal process @xmath134 is well understood and nnl calculations are available .
the theoretical error is thus negligible @xcite .
there are two contributions to the background process @xmath1 which we neglect in this paper .
firstly , the so - called resolved photon contribution @xcite was found to be a small effect , e.g. @xcite , especially since we want to reconstruct the higgs mass from the final two - jet measurements and impose angular cuts in the forward region .
in addition the good charm suppression also helps to suppress the resolved photon effects as they give the largest contribution .
the second contribution we do not consider here results from the final state configuration where a soft quark is propagating down the beam pipe and the gluon and remaining quark form two hard back - to back - jets @xcite .
we neglect this contribution here due to the expected excellent double b - tagging efficiency and the strong restrictions on the allowed acollinearity discussed below . for the continuum heavy quark production cross section
an exact nlo calculation exists @xcite but large radiative corrections in the @xmath3 channel require the resummation of large non - sudakov dl s as expounded on above .
assuming that the largest part of the nlo and higher subleading logarithms is contained in the renormalization of the strong coupling parameter @xmath42 , the virtual corrections seem to be well under control .
the largest uncertainty we thus expect to be contained in the missing hard @xmath135 bremsstrahlung corrections for which no @xmath37 suppression - factor exists .
theoretically , these can be controlled by limiting the available phase - space through a narrow two - jet definition . on the other hand
this means that we would also lose ( signal ) events which is clearly undesirable . in light of these two effects
we think it prudent to find a balancing middle ground for our mc - results for resonant energies . ] . more details are given below . a second source of uncertainty
is contained in the higher order @xmath61-dependence as mentioned above .
our strategy of identifying @xmath136 is reasonable as long as the neglected @xmath137 terms ( which have no born cross section suppression ) are negligible and can be identified with the physical detector energy cuts . in this paper
we will thus study two different values for the energy cut : @xmath138 and @xmath139 .
the value of @xmath129 is related to the allowed acollinearity of the two jet alignments corresponding to acollinearities of @xmath140 and @xmath141 .
we emphasize that the requirement of a small jet acollinearity substantially suppresses the @xmath133 background component and could play an important role in improving the photon collider energy resolution @xcite as well as in the suppression of the background due to the resolved processes .
below we display results assuming in each case a ( realistic ) ratio of @xmath142 in parallel with the ( optimistic ) ratio of @xmath143 .
we start with fig .
[ fig : ne203 ] assuming a ( quite realistic ) probability of counting a @xmath144- as a @xmath145-pair of @xmath146 and the sterman - weinberg parameters @xmath147 and @xmath148 .
the figure shows signal and bg events separately for two values of the thrust angle @xmath30 cut , @xmath149 and @xmath150 . in both scenarios
it can be seen that the largest component to the bg events for @xmath142 originates from the @xmath133 @xmath151-quark contribution .
the second largest corrections stem from the @xmath7-quark for both @xmath3 and @xmath133 while the @xmath3 @xmath151-quark contribution is small .
the smaller thrust - cut two - jet definition eliminates more of the background events in relative terms .
however , it also reduces the total number of events .
[ fig : sbg203 ] demonstrates that the @xmath150 scenario yields roughly a 50 % higher ratio of signal to bg events .
the inverse statistical significance of the higgs - boson process , defined as @xmath152 , however , is somewhat higher for the @xmath153 choice as is demonstrated in fig .
[ fig : sa203 ] .
the difference between the realistic @xmath142 and optimistic @xmath143 photon - polarization cases is small .
for the one - year running analysis of an intermediate mass higgs with @xmath154gev the inverse statistical significance is below 3% , which can be viewed as the minimal statistical expectation .
it seems now possible to assume an even better detector performance .
the improvement comes from assuming a better single point resolution , thinner detector modules and moving the vertex detectors closer to the beam - line @xcite .
thus we can assume a ( still realistic ) 1% probability of counting a @xmath144- as a @xmath145-pair .
[ fig : ne201 ] , [ fig : sbg201 ] and [ fig : sa201 ] display the same observables for otherwise identical two - jet definitions and machine - parameters .
the charm - contribution is visibly reduced and the number of signal to background events roughly 30% larger . the statistical accuracy for the higgs signal ,
however , is only slightly enhanced . with these results in hand
we now keep @xmath153 fixed and furthermore assume the @xmath144 misidentification rate of 1% .
we vary the cone angle @xmath155 between narrow ( @xmath156 ) , medium ( @xmath157 ) and large ( @xmath158 ) cone sizes for both @xmath147 and @xmath159 .
the upper row of fig . [ fig : cs ] demonstrates that for the former choice of the energy cutoff parameter we achieve the highest statistical accuracy for the large @xmath160 scenario of around 2% .
we again emphasize , however , that in this case also the missing @xmath135 bremsstrahlung corrections could become important .
the largest effect is obtained by effectively suppressing the background radiative events with the smaller energy cutoff of @xmath159 outside the cone ( the inside is of course independent of @xmath129 ) . here the lower row of fig . [
fig : cs ] demonstrates that the statistical accuracy of the higgs boson with @xmath154 gev can be below the 2% level after collecting one year of data .
we should mention again that for this choice of @xmath129 we might have slightly enhanced the higher order ( uncanceled ) cutoff dependence .
the dependence on the photon - photon polarization degree is visible but not crucial . in summary
, it seems very reasonable to expect that at the compton collider option we can achieve a 2% statistical accuracy of an intermediate mass higgs boson signal after collecting data over one year of running .
in this paper we have studied the higgs signal and continuum background contributions to the process @xmath161 at a high - energy compton collider .
we have used all relevant qcd radiative corrections to both the signal and bg production available in the literature .
the monte carlo results using a variety of jet - parameter variations revealed that the intermediate mass higgs signal can be expected to be studied with a statistical uncertainty between @xmath146 in a very realistic and @xmath162 in an optimistic scenario after one year of collecting data . together with the expected uncertainty of 1% from the @xmath6 mode determination of br@xmath163 , and assuming four years of collecting data ,
we conclude that statistically a measurement of the partial width @xmath14 below the 2% precision level should be possible .
this level of accuracy could significantly enhance the kinematical reach of the mssm parameter space in the large pseudoscalar mass limit and thus open up a window for physics beyond the standard model . for the total higgs width
, the main uncertainty is given by the error in the branching ratio br@xmath164 , which at present is estimated at the 15 % level @xcite .
for higgs masses above 110 gev , the total higgs width could be determined more precisely through the higgs - strahlung process @xcite and its decay into @xmath98 @xcite .
this is only possible , however , if the supersymmetric lightest higgs boson coupling to vector bosons is universal ( i.e. the same for @xmath165 and @xmath166 ) and provided the optimistic luminosity assumptions can be reached . in summary , using conservative machine and detector design parameters ,
we conclude that the compton collider option at a future linear collider can considerably extend our ability to discriminate between the sm and mssm scenarios .
* acknowledgements * + we would like to thank m. battaglia , g. jikia , r. orava , m. piccolo and especially v.i .
telnov for useful discussions .
this work was supported in part by the eu fourth framework programme ` training and mobility of researchers ' , network ` quantum chromodynamics and the deep structure of elementary particles ' , contract fmrx - ct98 - 0194 ( dg 12 - miht ) . | the high precision determination of the partial width @xmath0 of an intermediate mass higgs boson is among the most important measurements at a future photon photon collider .
recently it was shown that large non - sudakov as well as sudakov double logarithmic corrections can be summed to all orders in the background process @xmath1 , @xmath2 , from an initially polarized @xmath3 state . in addition , running coupling corrections were included exactly to all orders by employing the renormalization group .
thus all necessary theoretical results for calculating the higgs signal and the non - higgs continuum background contributions to the process @xmath1 are now known .
we are therefore able to present for the first time precise predictions for the measurement of the partial width @xmath4 at the compton collider ( @xmath5 ) option at a future linear @xmath6 collider .
the interplay between signal and background is very sensitive to the experimental cuts and the ability of the detectors to identify @xmath7-quarks in the final state .
we investigate this in some detail using a monte carlo analysis , and conclude that a measurement with a 2 % statistical accuracy should be achievable .
this could have important consequences for the discovery of physics beyond the standard model , in particular for large masses of a pseudoscalar higgs boson as the decoupling limit is difficult and for a wide range of @xmath8 impossible to cover at the lhc proton - proton collider .
hep - ph/9907238 + dtp/99/70 + july 1999 + * higgs boson production at the compton collider * + + and * valery a. khoze@xmath9 * + \3 ) infn - lnf , p.o.box 13 , i-00044 , frascati ( roma ) , italy |
different characteristics of the distribution of the discrete spectrum for non - self - adjoint perturbations of model differential self - adjoint operators , e.g. , a laplacian on @xmath7 , a discrete laplacian on @xmath8 , etc . , were studied in a number of papers ( see frank laptev lieb
seiringer @xcite , borichev golinskii kupin @xcite , demuth hansmann katriel @xcite ) .
this paper focuses on complex perturbations of one dimensional schrdinger operators with infinite band spectrum and certain behavior of the lengths of its gaps ( the case of finite band schrdinger operators was studied in @xcite ) .
so , consider a real - valued measurable function @xmath9 on @xmath10 and denote by @xmath11 a maximal multiplication operator by @xmath9 .
the standing assumption is that the schrdinger operator @xmath12 is self - adjoint , @xmath13 , and its spectrum @xmath14 is infinite band , i.e. , @xmath15 , \qquad a_k\to + \infty.\ ] ] we say that the gaps are relatively bounded if @xmath16 is the length of @xmath17 gap in .
a typical example here is the hill operator with a periodic potential ( see ( * ? ? ?
* section xiii.16 ) ) .
it is well known ( see @xcite ) that @xmath18 as @xmath19 for potentials @xmath9 from @xmath20 on a period , so obviously holds for such potentials .
furthermore , consider the form sum @xmath21 where @xmath5 is a complex - valued potential .
if @xmath22 is a relatively compact perturbation of @xmath23 , that is , @xmath24 , and @xmath25 is a compact operator for @xmath26 , then , by the celebrated theorem of weyl ( see , e.g. , ( * ? ? ?
* section iv.5.6 ) ) , @xmath27 and @xmath28 ( disjoint union ) , the discrete spectrum @xmath29 of @xmath4 , i.e. , the set of isolated eigenvalues of finite algebraic multiplicity , can accumulate only on @xmath30 .
the main goal of the paper is to obtain certain quantitative bounds for the rate of this accumulation .
the assumption on the background @xmath9 looks as follows @xmath31 as for the perturbation @xmath5 the conditions will vary depending on the sign of @xmath6 .
precisely , for general @xmath5 s we assume that @xmath32 and for accretive perturbations with @xmath33 we put @xmath34 under assumptions
@xmath4 is a well - defined , closed and @xmath35-sectorial operator , and there is a number @xmath36 such that @xmath37 where @xmath38 is the numerical range of @xmath4 ( see , e.g. , ( * ? ? ?
* chapter vi ) ) .
moreover , @xmath4 appears to be a relatively compact ( even @xmath39 ) perturbation of @xmath23 , @xmath39 being the schatten
von neumann class of compact operators .
denote by @xmath40 the euclidian distance from a point @xmath41 to a set @xmath42 .
[ t1 ] let @xmath23 be the schrdinger operator with @xmath9 satisfying @xmath43 .
assume that @xmath23 is the infinite band operator with the spectrum @xmath44 , \ \
0\le a_1<b_1<a_2<b_2<\ldots , \ \ a_n\to+\infty,\ ] ] and the lengths of gaps are relatively bounded @xmath45 . then for the perturbation @xmath4 @xmath46 with @xmath5 @xmath47 and for each @xmath48 @xmath49 the following lieb thirring type inequality @xmath50 holds , where a positive constant @xmath51 depends on @xmath52 and @xmath53 . if we take @xmath54 , bound can be simplified .
indeed , now @xmath55 , @xmath56 and so @xmath57 there is an elementary way to specify @xmath58 and eliminate it from the final expression .
the price we pay is an additional factor in the right hand side .
[ t2 ] under assumptions @xmath43 , @xmath47 @xmath59 where a positive constant @xmath51 depends on @xmath52 and @xmath53 . denote @xmath60 , @xmath61 .
[ t3 ] let @xmath33 . under assumptions
@xmath43 , @xmath62 the following lieb thirring type inequality holds for each @xmath63 @xmath64 the only reason we restricted ourselves with the case of one dimensional schrdinger operator @xmath23 as a background is that the class of multidimensional schrdinger operators with spectra is not well understood .
our technique works for any dimension @xmath65 , and the corresponding problem will be elaborated on elsewhere .
the main analytic tool in the proof of theorem [ t1 ] is the following distortion lemma for linear fractional transformations of the form @xmath66 the argument here is quite elementary ( though , rather lengthy ) . and @xmath67 with map @xmath68.,width=529 ] [ l1 ]
let @xmath69 , \quad 0\le a_1<b_1<a_2<b_2<\ldots , \quad a_n\to+\infty,\ ] ] and let @xmath70 be its image under the linear fractional transformation @xmath71 @xmath72 , \ \
{ \beta}_k({\omega})=\frac1{b_k-{\omega}}\ , , \ \ { \alpha}_k({\omega})=\frac1{a_k-{\omega}}\,.\ ] ] then for @xmath73 the following bounds hold : for @xmath74 or @xmath75 @xmath76 for @xmath77 , @xmath78 @xmath79 moreover , if @xmath80 and the gaps are relatively bounded @xmath45 , then the unique bound is valid @xmath81 with no loss of generality we can assume that @xmath82 .
we begin with the case @xmath83 and put @xmath84 . if @xmath85 and @xmath86 , then @xmath87 and so @xmath88 similarly , if @xmath89 , then @xmath90 and @xmath91)=|{\text{\rm im\,}}{\lambda}|=\frac{|y|}{|z|^2}\,.\ ] ] since now @xmath92 , we have @xmath93)}{\operatorname{\rm d}(z , i_z)}=\frac1{|z|^2}>\frac1{|z|(|z|+a_1)}\,.\ ] ] consider now the case when @xmath94 .
fix @xmath95 in @xmath17 s gap , @xmath96 ( we put @xmath97 and treat @xmath98 as a number zero gap ) .
then @xmath99 define two sets of positive numbers @xmath100 by equalities @xmath101 or , equivalently , @xmath102 we also put @xmath103 , so @xmath104 while the point @xmath105 traverses the line @xmath106 , @xmath107 , its image @xmath108 describes a circle with diameter @xmath109 $ ] .
we discern the following two cases . *
case 1*. assume that @xmath110 lies over the `` gaps for @xmath110 '' .
for each @xmath111 there are two options for @xmath110 : interior gaps @xmath112 and the rightmost gap @xmath113 for gaps we have @xmath114 define an auxiliary function @xmath115 on the right half - line @xmath116 clearly , @xmath115 is monotone increasing on @xmath117 and decreasing on @xmath118 with the minimum @xmath119 . hence and give @xmath120 since by @xmath121 , we see that @xmath122 for gaps let first @xmath123 .
then as above in @xmath124 but it is not clear now which term prevails .
if @xmath125 then @xmath126 and @xmath127 otherwise @xmath128 implies @xmath129 next , for @xmath130 one has @xmath131 , and in case @xmath132 and so @xmath133 finally , in the case of `` gaps for @xmath110 '' we come to the bound @xmath134 a modified version of will be convenient in the sequel . for @xmath123 in view of @xmath135 we have @xmath136 and so for @xmath78 @xmath137 @xmath138 is the length of @xmath17 s gap .
similarly , for @xmath130 one has from @xmath139 * case 2*. assume that @xmath110 lies over the `` bands for @xmath110 '' @xmath140\longleftrightarrow u_j\le |y|\le v_{j } , \qquad j = k+1,k+2,\ldots.\ ] ] now @xmath141 @xmath142 so that @xmath143 for @xmath123 ( interior gap for @xmath105 ) inequality can be simplified in view of @xmath144 @xmath145 let now @xmath130 , i.e. , @xmath146 . in our case @xmath147 and @xmath148 if @xmath149
then @xmath150 and so @xmath151 otherwise , @xmath152 implies @xmath153 it follows now from with @xmath130 that @xmath154 we can summarize the results obtained above in the following two bounds from below . a combination of , , and gives @xmath155 a combination of and provides @xmath156 since @xmath157 , the latter can be written as @xmath158 to work out the general case @xmath159 and prove and , it remains only to shift the variable and apply the results just obtained to the shifted sequence of bands @xmath160.\ ] ] the final statement follows from a simple observation @xmath161 the proof is complete .
the key ingredient of the proof of our main statements is the following result of hansmann ( * ? ? ?
* theorem 1 ) .
let @xmath162 be a bounded self - adjoint operator on the hilbert space , and let @xmath163 be a bounded operator with @xmath164 , @xmath165 .
then @xmath166 @xmath167 is an explicit ( in a sense ) constant , which depends only on @xmath52 .
we set @xmath168 @xmath58 is defined above , and @xmath169 in view of and .
we go back to with @xmath187 so by the spectral mapping theorem , in the notation of lemma [ l1 ] we have @xmath188 we apply lemma [ l1 ] in the form to obtain @xmath189 and follows .
the proof is complete . @xmath190
the proof of theorem [ t1 ] shows that bound essentially depends on the parameter @xmath58 . roughly speaking
, it comes from a bound from below of @xmath191 , and so it seems to be rather important to estimate this quantity in terms of @xmath9 and @xmath5 only .
_ proof of theorem [ t2 ] .
_ put @xmath192 @xmath193 is defined in .
we wish to show that @xmath194 .
indeed , by , @xmath195 so @xmath196 is invertible and @xmath197 .
an application of the identity @xmath198 completes the proof of our claim .
it is clear by the definition of @xmath202 that if @xmath203 , @xmath204 , then @xmath205 .
take @xmath206 so that @xmath207 it is easy to check that @xmath208 , so , in particular , @xmath209 and hence @xmath210 once again , says @xmath211 for @xmath165 and , using lemma [ l1 ] , we come to @xmath212 by the choice of @xmath213 , we have @xmath214 for @xmath215 , and so @xmath216 and @xmath217 next , @xmath218 and hence @xmath219 the proof is complete . @xmath190 the lower bound for the difference of resolvents is the same as above in theorem [ t1 ] .
it is a consequence of the result of hansmann and lemma [ l1 ] @xmath222 as for the upper bound , we follow the line of reasoning from ( * ? ? ?
* proof of theorem 3.2 ) , where such bound was proved in the case @xmath223 . as a matter of fact
, the argument goes through under assumption as well . at any rate , we have @xmath224 since @xmath225 for @xmath226 and @xmath227
, we come to the following inequality @xmath228 as in ( * ? ? ?
* proof of theorem 3.3 ) , we multiply through by @xmath229 , @xmath230 , to obtain @xmath231 and then integrate the latter inequality with respect to @xmath232 .
the proof is complete .
@xmath190 _ acknowledgments .
_ the authors are grateful to s. denisov , f. gesztesy and m. hansmann for valuable discussions on the subject of the article .
the paper was prepared during the visit of the first author , funded by the idex programm of university of bordeaux , to the institute of mathematics of bordeaux ( imb umr5251 ) .
he would like to thank this institution for the hospitality .
l. golinskii , s. kupin , on discrete spectrum of a complex perturbations of finite band schrdinger operators , in : `` recent trends in analysis : proceedings of the conference in honor of nikolai nikolski : bordeaux , 2011 '' , theta foundation , bucarest , 2013 , 113121 . | let @xmath0 be an infinite band schrdinger operator on @xmath1 with a real - valued potential @xmath2 .
we study its complex perturbation @xmath3 , defined in the form sense , and obtain the lieb thirring type inequalities for the rate of convergence of the discrete spectrum of @xmath4 to the joint essential spectrum .
the assumptions on @xmath5 vary depending on the sign of @xmath6 . _
dedicated to yu .
m. berezanskii + on occasion of his 90-th birthday _ |
let @xmath0 be a monic polynomial of degree @xmath1 , with complex coefficients .
suppose that @xmath0 has a monic factor @xmath2 , so that @xmath7 where @xmath8 is also a monic polynomial .
define the uniform ( sup ) norm on a compact set @xmath5 in the complex plane @xmath9 by @xmath10 we study the inequalities of the following form @xmath11 where the main problem is to find the best ( the smallest ) constant @xmath6 , such that ( [ 1.2 ] ) is valid for _ any _ monic polynomial @xmath0 and _ any _ monic factor @xmath2 . in the case
@xmath12 , where @xmath13 the inequality ( [ 1.2 ] ) was considered in a series of papers by mignotte @xcite , granville @xcite and glesser @xcite , who obtained a number of improvements on the upper bound for @xmath14 .
d. w. boyd @xcite made the final step here , by proving that @xmath15 with @xmath16 the constant @xmath17 is asymptotically sharp , as @xmath18 and it can also be expressed in a different way , using mahler s measure .
this problem is of importance in designing algorithms for factoring polynomials with integer coefficients over integers .
we refer to @xcite and @xcite for more information on the connection with symbolic computations . a further development related to ( [ 1.2 ] ) for @xmath19,\ a>0,$ ]
was suggested by p. b. borwein in @xcite ( see theorems 2 and 5 there or see section 5.3 in @xcite ) . in particular
, borwein proved that if @xmath20 then @xmath21 where the bound is attained for a monic chebyshev polynomial of degree @xmath1 on @xmath22 $ ] and a factor @xmath23 .
he also showed that , for @xmath24 $ ] , the constant in the above inequality satisfies @xmath25 - 1 } \prod_{k = 1}^{[2n/3 ] } \left ( 1 + \cos \frac{2k -1}{2n } \pi \right)\right)^{1/n } \\ & = & \exp \left ( \int_0^{2/3 } \log \left(2 + 2\cos{\pi x}\right ) dx \right ) = 1.9081 \ldots,\end{aligned}\ ] ] which hints that @xmath26 } = \exp \left ( \int_0^{2/3 } \log \left(2 + 2\cos{\pi x}\right ) dx \right ) = 1.9081 \ldots.\ ] ] we find the asymptotically best constant @xmath6 in ( [ 1.2 ] ) for a rather arbitrary compact set @xmath5 .
the general result is then applied to the cases of a disk and a line segment , so that we recover ( [ 1.3])-([1.4 ] ) and confirm ( [ 1.6 ] ) .
our solution of the above problem is based on certain ideas from the logarithmic potential theory ( cf .
@xcite or @xcite ) .
let @xmath27 be the _ logarithmic capacity _ of a compact set @xmath28 .
for @xmath5 with @xmath29 , denote the _ equilibrium measure _ of @xmath5 ( in the sense of the logarithmic potential theory ) by @xmath30 .
we remark that @xmath30 is a positive unit borel measure supported on @xmath5 , @xmath31 ( see @xcite ) .
[ thm1 ] let @xmath28 be a compact set , @xmath32 .
then the best constant @xmath6 in ( [ 1.2 ] ) is given by latexmath:[\[\label{2.1 } c_e = \frac{{\displaystyle}\max_{u \in \partial e } \exp\left ( \int_{|z - u| \ge 1 } \log @xmath5 is regular then @xmath34 the above notion of regularity is to be understood in the sense of the exterior dirichlet problem ( cf .
@xcite ) . note that the condition @xmath35 is usually satisfied for all applications , as it only fails for very _ thin
_ sets ( see @xcite ) , e.g. , finite sets in the plane .
but if @xmath5 consists of finitely many points then the inequality ( [ 1.2 ] ) can not be true for a polynomial @xmath0 with zeros at every point of @xmath5 and for its linear factors @xmath2 .
on the other hand , theorem [ thm1 ] is applicable to any compact set with a connected component consisting of more than one point ( cf .
@xcite ) .
one can readily see from ( [ 1.2 ] ) or ( [ 2.1 ] ) that the best constant @xmath6 is invariant under the rigid motions of the set @xmath5 in the plane .
therefore we consider applications of theorem [ thm1 ] to the family of disks @xmath36 , which are centered at the origin , and to the family of segments @xmath22,\ a>0.$ ] [ cor1 ] let @xmath37 be a disk of radius @xmath38 .
then the best constant @xmath39 , for @xmath40 , is given by @xmath41 note that ( [ 1.3])-([1.4 ] ) immediately follow from ( [ 2.3 ] ) for @xmath42 the graph of @xmath39 , as a function of @xmath38 , is in figure [ fig1 ] .
[ cor2 ] if @xmath19,\ a>0 $ ] , then @xmath43 } = \left \ { \begin{array}{l } { \displaystyle}\frac{2}{a } , \quad 0 < a \le 1/2 , \\ \\
{ \displaystyle}\frac{2}{a } \exp \left({\displaystyle}\int_{1-a}^a \frac{\log(t+a)}{\pi \sqrt{a^2-t^2}}\ , dt \right ) , \quad a>1/2 . \end{array } \right.\ ] ] observe that ( [ 2.4 ] ) , with @xmath44 , implies ( [ 1.6 ] ) by the change of variable @xmath45 we include the graph of @xmath46}$ ] , as a function of @xmath47 , in figure [ fig2 ] .
we now state two general consequences of theorem [ thm1 ] .
they explain some interesting features of @xmath6 , which the reader may have noticed in corollaries [ cor1 ] and [ cor2 ] .
let @xmath48 be the euclidean diameter of @xmath5 .
[ cor3 ] suppose that @xmath49 if @xmath50 then @xmath51 it is well known that cap@xmath52 and cap@xmath53)=a/2 $ ] ( see @xcite ) , which clarifies the first lines of ( [ 2.3 ] ) and ( [ 2.4 ] ) by ( [ 2.5 ] ) .
the next corollary shows how the constant @xmath6 behaves under dilations of the set @xmath5 .
let @xmath54 be the dilation of @xmath5 with a factor @xmath55 [ cor4 ] if @xmath5 is regular then @xmath56 thus figures [ fig1 ] and [ fig2 ] clearly illustrate ( [ 2.6 ] ) .
we remark that one can deduce inequalities of the type ( [ 1.2 ] ) , for various @xmath57 norms , from theorem [ thm1 ] , by using relations between @xmath57 and @xmath58 norms of polynomials on @xmath5 ( see , e.g. , @xcite ) .
the proof of this result is based on the ideas of @xcite and @xcite . for @xmath59 , consider a function @xmath60 one can immediately see that @xmath61 is a subharmonic function in @xmath62 , which has the following integral representation ( see @xcite ) : @xmath63 where @xmath64 is the normalized angular measure on @xmath65 .
we use the well known bernstein - walsh lemma about the growth of a polynomial outside of the set @xmath5 ( see @xcite , for example ) : + let @xmath28 be a compact set , @xmath70 , with the unbounded component of @xmath71 denoted by @xmath72 .
then , for any polynomial @xmath0 of degree @xmath1 , we have @xmath73 is the green function of @xmath72 , with pole at @xmath74 .
the following representation for @xmath75 is found in theorem iii.37 of @xcite ) .
@xmath76 it follows from ( [ 3.1])-([3.4 ] ) and fubini s theorem that @xmath77 using the definition of @xmath78 , we obtain from the above estimate that @xmath79 hence @xmath80 ) , we consider the @xmath1-th fekete points @xmath81 for the set @xmath5 ( cf .
let @xmath82 be the fekete polynomial of degree @xmath1 .
define the normalized counting measures on the fekete points by @xmath83 it is known that ( see theorems 5.5.4 and 5.5.2 in @xcite ) @xmath84 furthermore , we have the following weak * convergence of counting measures ( cf .
@xcite ) : @xmath85 let @xmath66 be a point , where the maximum on the right hand side of ( [ 3.5 ] ) is attained . define the factor @xmath86 for @xmath87 , with zeros being the @xmath1-th fekete points satisfying @xmath88
. then we have by ( [ 3.7 ] ) that @xmath89 combining the above inequality with ( [ 3.6 ] ) and the definition of @xmath6 , we obtain that @xmath90 this shows that ( [ 2.1 ] ) holds true .
moreover , if @xmath66 is a regular point for @xmath72 , then we obtain by theorem iii.36 of @xcite ) and ( [ 3.4 ] ) that @xmath91 hence @xmath92 which implies ( [ 2.2 ] ) by ( [ 2.1 ] ) .
it is well known @xcite that cap@xmath93 and @xmath94 where @xmath95 is the angular measure on @xmath96 . if @xmath97 $ ] then the numerator of ( [ 2.1 ] ) is equal to 1 , so that @xmath98 assume that @xmath99 we set @xmath100 and let @xmath101 be a point where the maximum in ( [ 2.1 ] ) is attained .
on writing @xmath102 we obtain that @xmath103 by the change of variable @xmath104 recall that cap@xmath53)=a/2 $ ] ( see @xcite ) and @xmath105}(t)=\frac{dt}{\pi\sqrt{a^2-t^2 } } , \quad t \in [ -a , a].\ ] ] it follows from ( [ 2.1 ] ) that @xmath106 } = \frac{2}{a}\ , \exp \left({\displaystyle}\max_{u \in [ -a , a ] } \int_{[-a , a]\setminus(u-1,u+1 ) } \frac{\log|t - u| } { \pi \sqrt{a^2-t^2}}\ , dt \right).\ ] ] if @xmath107 $ ] then the integral in ( [ 3.8 ] ) obviously vanishes , so that @xmath46 } = 2/a$ ] . for @xmath108 ,
let @xmath109\setminus(u-1,u+1 ) } \frac{\log|t - u| } { \pi \sqrt{a^2-t^2}}\ , dt.\ ] ] one can easily see from ( [ 3.9 ] ) that @xmath110,\ ] ] and @xmath111.\ ] ] however , if @xmath112 then @xmath113 it is not difficult to verify directly that @xmath114 which implies that @xmath115 for @xmath116 hence @xmath117 collecting all facts , we obtain that the maximum for @xmath118 on @xmath22 $ ] is attained at the endpoints @xmath119 and @xmath120 , and it is equal to @xmath121 } f(u ) = \int_{1-a}^a \frac{\log(t+a)}{\pi \sqrt{a^2-t^2}}\ , dt.\ ] ] thus ( [ 2.3 ] ) follows from ( [ 3.8 ] ) and the above equation .
observe that @xmath123 for any @xmath124 , so that @xmath125 . since @xmath5 is regular , we use the representation for @xmath6 in ( [ 2.2 ] ) .
let @xmath126 be the dilation mapping
. then @xmath127 and @xmath128 .
this gives that @xmath129 where @xmath130 .
using the absolute continuity of the integral , we have that @xmath131 which implies ( [ 2.6 ] ) .
p. b. borwein , _ exact inequalities for the norms of factors of polynomials _ , can
. j. math .
* 46 * ( 1994 ) , 687 - 698 .
p. borwein and t. erdlyi , polynomials and polynomial inequalities , springer - verlag , new york , 1995 .
d. w. boyd , _ two sharp inequalities for the norm of a factor of a polynomial _ , mathematika * 39 * ( 1992 ) , 341 - 349 .
d. w. boyd , _ sharp inequalities for the product of polynomials _ , bull .
london math .
* 26 * ( 1994 ) , 449 - 454 .
d. w. boyd , _ large factors of small polynomials _ , contemp
* 166 * ( 1994 ) , 301 - 308 .
p. glesser , _ nouvelle majoration de la norme des facteurs dun polynme _ , c. r. math .
canada * 12 * ( 1990 ) , 224 - 228 .
a. granville , _ bounding the coefficients of a divisor of a given polynomial _ , monatsh
* 109 * ( 1990 ) , 271 - 277 .
s. landau , _ factoring polynomials quickly _ , notices amer . math .
* 34 * ( 1987 ) , 3 - 8 .
m. mignotte , _ some useful bounds _ , in computer algebra , symbolic and algebraic computation " ( b. buchberger et al .
, eds . ) , pp . 259 - 263 , springer - verlag , new york , 1982 .
i. e. pritsker , _ products of polynomials in uniform norms _ , to appear in trans .
i. e. pritsker , _ comparing norms of polynomials in one and several variables _ , j. math .
* 216 * ( 1997 ) , 685 - 695 .
t. ransford , potential theory in the complex plane , cambridge university press , cambridge , 1995 .
m. tsuji , potential theory in modern function theory , chelsea publ .
co. , new york , 1975 . | let @xmath0 be a monic polynomial of degree @xmath1 , with complex coefficients , and let @xmath2 be its monic factor .
we prove an asymptotically sharp inequality of the form @xmath3 , where @xmath4 denotes the sup norm on a compact set @xmath5 in the plane .
the best constant @xmath6 in this inequality is found by potential theoretic methods .
we also consider applications of the general result to the cases of a disk and a segment . |
the radial component of the position of a distant object is inferred from its cosmological redshift , induced by the expansion of the universe ; the light observed from a distant galaxy appears to us at longer wavelengths than in the rest frame of that galaxy .
the most accurate determination of the exact redshift , @xmath4 , comes from directly observing the spectrum of an extragalactic source and measuring a consistent multiplicative shift , relative to the rest frame , of various emission ( or absorption ) features .
the rest - frame wavelengths of these emission lines are known to a high degree of accuracy which can be conferred onto the measured spectroscopic redshifts , @xmath5 .
however , the demand on telescope time to obtain spectra for every source in deep , wide surveys is prohibitively high , and only relatively small area spectroscopic campaigns can reach faint magnitudes ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , or at the other extreme , relatively bright magnitudes over larger areas ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
this forces us towards the use of photometric observations to infer the redshift by other means . rather than individual spectra
, the emission from a distant galaxy is observed in several broad filters , facilitating the characterization of the spectral energy distribution ( sed ) of fainter sources , at the expense of fine spectral resolution .
photometric redshift methods largely fall into two categories , based on either sed template fitting or machine learning .
template fitting software such as hyperz ; @xcite , zebra ; @xcite , eazy ; @xcite and le phare @xcite rely on a library of sed templates for a variety of different galaxy types , which ( given the transmission curves for the photometric filters being used ) can be redshifted to fit the photometry .
this method can be refined in various ways , often with the use of simulated seds rather than only those observed at low redshift , composite seds , and through calibration using any available spectroscopic redshifts .
machine learning methods such as artificial neural networks ( e.g. annz ; * ? ? ?
* ; * ? ? ?
* ) , nearest - neighbour ( nn ) @xcite , genetic algorithms ( e.g. * ? ? ?
* ) , self - organized maps @xcite and random forest @xcite , to name but a few , rely on a significant fraction of sources in a photometric catalogue having spectroscopic redshifts .
these ` true ' redshifts are used to train the algorithm .
in addition to providing a point estimate , machine learning methods can provide the degree of uncertainty in their prediction @xcite .
both methods have their strengths and weaknesses , with the best performance often depending on the available data and the intended science goals .
as such , future surveys may well depend on contributions from both in tandem , but there has been extensive work on comparing the current state of the art in public software using a variety of techniques @xcite .
artificial neural networks motivate the most commonly used machine learning software @xcite , however gaussian processes ( e.g. * ? ? ?
* ) have not yet become well established in this area , despite comparison by @xcite suggesting that they may outperform the popular annz code , using the rms error as a metric . in this paper
, we introduce a novel sparse kernel regression model that greatly reduces the number of basis ( kernel ) functions required to model the data considered in this paper .
this is achieved by allowing each kernel to have its own hyper - parameters , governing its shape .
this is in contrast to the standard kernel - based models in which a set of global hyper - parameters are optimized ( such as is typical in gaussian process ( gp ) methods ) .
the complexity cost of such a kernel - based regression model is @xmath6 , where @xmath7 is the number of basis functions .
this cubic time complexity arise from the cost of inverting an @xmath7 by @xmath7 covariance matrix . in a standard gaussian process model @xcite ,
seen as a kernel regression algorithm , we may regard the basis functions , as located at the @xmath7 points in the training sample .
this renders such an approach unusable for many large training data applications where scalability is a major concern .
much of the work done to make gps more scalable is either to ( a ) make the inverse computation faster or ( b ) use a smaller representative set from the training sample to reduce the rank and ease the computation of the covariance matrix .
examples of ( a ) include methods such as structuring the covariance matrix such that it is much easier to invert , using toeplitz @xcite or kronecker decomposition @xcite , or inverse approximation as an optimization problem @xcite .
to reduce the number of representative points ( b ) , an @xmath8 subset of the training sample can be selected which maximizes the accuracy or the numerical stability of the inversion @xcite .
alternatively , one may search for `` inducing '' points not necessarily present in the training sample , and not necessarily even lying within the data range , to use as the basis set such that the probability of the data being generated from the model is maximized @xcite .
approaches such as relevance vector machines ( rvm ; * ? ? ?
* ) and support vector machines ( svm ; * ? ? ?
* ) are basis - function models .
however , unlike sparse gps , they do not learn the basis functions locations but rather apply shrinkage to a set of kernels in the form of weight - decay on the linear weights that couple the kernels , located at training data points , to the regression . in this paper
, we propose a non - stationary sparse gaussian model to target photometric redshift estimation .
the key difference between the proposed approach and other basis function models , is that our model does not use shrinkage ( automatic relevance determination ) external to the kernel , but instead has a length - scale parameter in each kernel .
this allows for parts of the input - output regression mapping to have different characteristic length - scales .
we can see this as allowing for shrinkage and reducing the need for more basis functions , as well as allowing for non - stationary mappings .
a regular gp , sparse gp or rvm does not do this , and we demonstrate that this is advantageous to photometric redshift estimation . furthermore , the model is presented within a framework with components that address other challenges in photometric redshift estimation such as incorporating a weighting scheme as an integral part of the process to remove , or introduce , any systematic bias , and a prior mean function to enhance the extrapolation performance of the model .
the results are demonstrated on photometric redshift estimation for a simulated _ euclid_-like survey @xcite and on observational data from the 12th data release of the sloan digital sky survey ( sdss ) @xcite . in particular
, we use the weighting scheme to remove any distribution bias and introduce a linear bias to directly target the mission s requirement .
the paper is organised as follows , a brief introduction to gaussian processes for regression is presented in section [ sec - gaussian - process ] followed by an introduction to sparse gps in section [ sec - sparse - gaussian - processes ] .
the proposed approach is described in section [ sec - proposed - approach ] followed by an application to photometric redshift estimation in section [ sec - application ] , where the details of the mock dataset are described .
the experiments and results are discussed in section [ sec - experiments ] on the simulated survey , and in section [ sec - experiments - sdss ] we demonstrate the performance of the proposed model and compare it to annz on the sdss 12th data release .
finally , we summarize and conclude in section [ sec - summary ] .
in many modelling problems , we have little prior knowledge of the explicit functional form of the function that maps our observables onto the variable of interest .
imposing , albeit sensible , parametric models , such as polynomials , makes a tacit bias . for this reason ,
much of modern function modelling is performed using _ non - parametric _ techniques . for regression ,
the most widely used approach is that of _ gaussian processes _ @xcite .
a gaussian process is a supervised non - linear regression algorithm that makes few explicit _ parametric _ assumptions about the nature of the function fit . for this reason ,
gaussian processes are seen as lying within the class of bayesian non - parametric models .
the underlying assumption in a gp is that , given a set of input @xmath9 and a set of target outputs @xmath10 , where @xmath7 is the number of samples in the dataset and @xmath11 is the dimensionality of the input , the observed target @xmath12 is generated by a function @xmath13 of the input @xmath14 plus additive noise @xmath15 : @xmath16 the noise @xmath17 is taken to be normally distributed with a mean of zero and variance @xmath18 , or @xmath19 . to simplify the notation
, it is assumed that @xmath20 ( this can readily be achieved without loss of generality , via a linear whitening process ) and univariate , although the derivation can be readily extended to multivariable problems .
the conditional probability of the observed variable given the function is hence distributed as follows : @xmath21 a gp then proceeds by applying a _
bayesian _ treatment to the problem to infer a probability distribution over the space of possible functions @xmath13 given the data : @xmath22 this requires us to define a prior , @xmath23 , over the function space . the function is normally distributed with a mean of zero , to match the mean of the normalized variable @xmath24 , with a covariance _ function _ @xmath25 , i.e. @xmath26 . the covariance function captures prior knowledge about the relationships between the observables .
most widely used covariance functions assume that there is local similarity in the data , such that nearby inputs are mapped to similar outputs .
the covariance @xmath25 can therefore be modelled as a function of the input @xmath27 , @xmath28 , where each element @xmath29 and @xmath30 is the covariance function . for @xmath25 to be a valid covariance it has to be symmetric and positive semi - definite matrix ; arbitrary functions for @xmath30 can not guarantee these constraints . a class of functions that guarantees these structural constraints are referred to as _ mercer kernels
a commonly used kernel function , which is the focus of this work , is the squared exponential kernel , defined as follows : @xmath31 where @xmath32 and @xmath33 are referred to as the height ( output , or variance ) and characteristic length ( input ) scale respectively , which correspond to tunable _ hyper - parameters _ of the model .
the similarity between two input vectors , under the squared exponential kernel , is a non - linear function of the euclidean distance between them .
we note that this choice of kernel function guarantees continuity and smoothness in the function and all its derivatives . for a more extensive discussion of covariances ,
the reader is referred to @xcite or @xcite . with the likelihood @xmath34 and prior @xmath23 ,
the marginal likelihood @xmath35 can be computed as follows : @xmath36 by multiplying the likelihood and the prior and completing the square over @xmath13 , we can express the integration as a normal distribution independent of @xmath13 multiplied by a another normal distribution over @xmath13 .
the distribution independent of @xmath13 can then be taken out of the integral , and the integration of the second normal distribution with respect to @xmath13 will be equal to one .
the resulting distribution of the marginal likelihood is distributed as follows @xcite : @xmath37 the marginal likelihood of the full data set can hence be computed as follows : @xmath38 the aim of a gp , is to maximize the probability of observing the target @xmath24 given the input @xmath27 , eq . .
note that the only free parameters to optimize in the marginal likelihood are the parameters of the kernel and the noise variance , collectively referred to as the _ hyper - parameters _ of the model .
it is more convenient however to maximize the log of the marginal likelihood , eq .
, since the log function is a monotonically increasing function , maximizing the log of a function is equivalent to maximizing the original function .
the log likelihood is given as : @xmath39 we search for the optimal set of hyper - parameters using a gradient - based optimization , hence we require the derivatives of the log marginal likelihood with respect to each hyper - parameter . in this paper ,
the l - bfgs algorithm was used to optimize the objective which uses a quasi - newton method to compute the search direction in each step by approximating the inverse of the hessian matrix from the history of gradients in previous steps @xcite .
it is worth mentioning that non - parametric models require the optimization of few hyper - parameters that do not grow with the size of the data and are less prone to overfitting .
the distinction between parameters and hyper - parameters of a model is that the former directly influence the input - output mapping , for example the linear coupling weights in a basis function model , whereas the latter affect properties of distributions in the probabilistic model , for example the widths of kernels .
although this distinction is somewhat semantic , we keep to this nomenclature as it is standard in the statistical machine learning literature .
once the hyper - parameters have been inferred , the conditional distribution of future predictions @xmath40 for test cases @xmath41 given the training sample can be inferred from the joint distribution of @xmath40 and the observed targets @xmath24 .
if we assume that the joint distribution is a multivariate gaussian , then the joint probability is distributed as follows : @xmath42 where we introduce the shorthand notations @xmath43 , @xmath44 , @xmath45 and @xmath46 .
the conditional distribution @xmath47 is therefore distributed as follows : @xmath48 if we assume a non - zero prior mean @xmath49 over the function , @xmath50 , and an un - normalized @xmath24 with mean @xmath51 , the mean of the posterior distribution will be equal to : @xmath52 the main drawback of gps is the @xmath6 computational cost required to invert the @xmath53 matrix @xmath54 .
the _ sparse gaussian process _ allows us to reduce this computational cost and is detailed in the following section .
gaussian processes are often described as non - parametric regression models due to the lack of an explicit parametric form .
indeed gp regression can also be viewed as a functional mapping @xmath55 parameterized by the data and the kernel function , followed by linear regression via optimization of the following objective : @xmath56 where @xmath57 are the set of coefficients for the linear regression model that maps the transformed features @xmath25 to the desired output @xmath24 .
the feature transformation @xmath25 evaluate how `` similar '' a datum is to every point in the training sample , where the similarity measure is defined by the kernel function . if two points have a high kernel response via eq .
, this will result in very correlated features , adding extra computational cost for very little or no added information . selecting a subset of the training sample that maximizes
the preserved information is a research question addressed in @xcite , whereas in @xcite the basis functions are treated as a search problem rather than a selection problem and their locations are treated as hyper - parameters which are optimized .
these approaches result in a transformation @xmath58 , in which @xmath59 is the number of basis functions used .
the transformation matrix @xmath25 will therefore be a rectangular @xmath7 by @xmath60 matrix and the solution for @xmath57 in eq .
is calculated via standard linear algebra as : @xmath61 even though these models improve upon the computational cost of a standard gp , very little is done to compensate for the reduction in modelling power caused by the `` loss '' of basis functions .
the selection method is always bounded by the full gp s accuracy , on the _ training _ sample , since the basis set is a subset of the full gp basis set . on the other hand ,
the sparse gp s ability to place the basis set freely across the input space does go some way to compensate for this reduction , as the kernels can be optimized to describe the distribution of the data . in other words , instead of training a gp model with all data points as basis functions , or restricting it to a subset of the training sample which require some cost to select them , a set of inducing points is used in which their locations are treated as hyper - parameters of the model to be optimized . in both a full and a low rank approximation gp , a global set of hyper - parameters is used for all basis functions , therefore limiting the algorithm s local modelling capability . moreover
, the objective in eq . minimizes the sum of squared errors , therefore for any non - uniformly distributed output , the optimization routine will bias the model towards the mean of the output distribution and will seek to fit preferentially the region of space where there are more data .
hence , the model might allow for very poor predictions for few points in poorly represented regions , e.g. the high redshift range , in order to produce good predictions for well represented regions .
therefore , the error distribution as a function of redshift is not uniform unless the training sample is well balanced , producing a model that is sensitive to how the target output is distributed . in the next section ,
a method is proposed which addresses the above issues by parametrizing each basis function with bespoke hyper - parameters which account for variable density and/or patterns across the input space .
this is particularly pertinent to determining photometric redshifts , where complete spectroscopic information may be restricted or biased to certain redshifts or galaxy types , depending on the target selection for spectroscopy of the training sample .
this allows the algorithm to learn more complex models with fewer basis functions .
in addition , a weighting mechanism to remove any distribution bias from the model is directly incorporated into the objective .
in this paper , we extend the sparse gp approach by modelling each basis ( kernel ) function with its own set of hyper - parameters . the kernel function in eq .
is hence redefined as follows : @xmath62 where @xmath63 are the set of basis coordinates and @xmath64 is the corresponding length scale for basis @xmath65 . the multivariate input is denoted as @xmath66 . throughout the rest of the paper , @xmath67 denotes the @xmath68-th row of matrix @xmath27 , or @xmath14 for short , whereas @xmath69 denotes the @xmath65-th column and @xmath70 refers to the element at row @xmath68 and column @xmath65 in matrix @xmath27 , and similarly for other matrices .
note that the hyper - parameter @xmath32 has been dropped , as it interferes with the regularization objective .
this can be seen from the final prediction equation @xmath71 , the weights are always multiplied by their associated @xmath32 . therefore , the optimization process will always compensate for decreasing @xmath72 by increasing @xmath73 .
dropping the height variance ensures that the kernel functions do not grow beyond control and delegates learning the linear coefficients and regularization to the weights in @xmath57 .
the derivatives with respect to each length scale and position are provided in equations eq . and
respectively : @xmath74 the symbol @xmath75 denotes the hadamard product , i.e. element - wise matrix multiplication and @xmath76 denotes a column vector of length @xmath7 with all elements set to 1 . finding the set of hyper - parameters that optimizes the solution , is in effect finding the set of radial basis functions defined by their positions @xmath77 and radii @xmath78 that jointly describe the patterns across the input space . by parametrizing them differently ,
the model is more capable to accommodate different regions of the space more specifically .
a global variance model assumes that the relationship between the input and output is global or equal across the input space , whereas a variable variance model , or non - stationary gp , makes no assumptions and learns the variable variances for each basis function which reduces the need for more basis functions to model the data .
the kernel in eq . can be further extended to , not only model each basis function with its own radius @xmath64 , but also model each one with its own covariance @xmath79 .
this enables the basis function to have any arbitrary shaped ellipses giving it more flexibility .
the kernel in eq . can be extended as follows : @xmath80 furthermore , to make the optimization process faster and simpler , we define the additional variables : @xmath81 where @xmath82 is a local affine transformation matrix for basis function @xmath65 and @xmath83 is the application of the local transformation to the data . optimizing with respect to @xmath84 directly ensures that the covariance matrix is positive definite .
this makes it faster from a computational perspective as the kernel functions for all the points with respect to a particular basis can be computed more efficiently as follows : @xmath85 the exponent in eq .
basically computes the sum of squares in each row of @xmath83 .
this allows for a more efficient computation of the kernel functions for all the points in a single matrix operation .
the derivatives with respect to each @xmath84 and @xmath86 are shown in eq . and
eq . respectively .
@xmath87 setting up the problem in this manner allows the setting of matrix @xmath84 to be of any size @xmath11 by @xmath88 , where @xmath89 which can be considered as a low rank approximation to @xmath90 without affecting the gradient calculations .
in addition , the inverse of the covariance can be set to @xmath91 in the low rank approximation case to ensure that the final covariance can model a diagonal covariance .
this is referred to as _ factor analysis distance _ @xcite but previously used to model a global covariance as opposed to variable covariances as is the case here . in the absence of observations , all bayesian models ,
gaussian processes included , rely on their priors to provide function estimation .
for the case of gaussian processes this requires us to consider the prior over the function , especially the prior mean .
for example , the first term in the mean prediction in eq . ,
@xmath49 , is our prior mean in which we learn the deviation from using a gp .
similarly , we may consider a mean _ function _ that is itself a simple linear regression from the independent to dependent variable .
the parameters of this function are then inferred and the gp infers non - linear deviations . in the absence of data ,
e.g. in extrapolative regions , the gp will fall back to the linear regression prediction @xcite .
we can incorporate this directly into the optimization objective instead of having it as a separate preprocessing step by redefining @xmath25 as a concatenation of the linear and non - linear features , or setting @xmath92 $ ] and @xmath93 $ ] , where @xmath94 is the linear regression s coefficients and @xmath95 is the bias .
the prediction can then be formulated as follows : @xmath96 furthermore , the regularization matrix , @xmath97 , in eq .
can be modified so that it penalises the learning of high coefficients for the non - linear terms , @xmath57 , but small or no cost for learning high linear terms , @xmath94 and @xmath95 , by setting the corresponding elements in the diagonal of @xmath98 to 0 instead of @xmath18 , or the last @xmath99 elements .
therefore , as @xmath18 goes to infinity , the model will approach a simple linear regression model instead of fallen back to zero . thus far in the discussion , we make the tacit assumption that the objective of the inference process is to minimize the sum of squared errors between the model and target function values .
although this is a suitable objective for many applications , it is intrinsically biased by uneven distributions of training data , sacrificing accuracy in less represented regions of the space . ideally we would like to train a model with a balanced data distribution to avoid such bias .
this however , is a luxury that we often do not have .
for example , the lack of strong emission lines that are detectable with visible - wavelength spectrographs in the `` redshift - desert '' at @xmath100 means that this redshift range is often under - represented in spectroscopic samples .
a common technique is to either over - sample or under - sample the data to achieve balance @xcite . in under - sampling
, samples are removed from highly represented regions to achieve balance , over - sampling on the other hand duplicates under represented samples .
both approaches come with a cost ; in the former good data are wasted and in the latter more computation is introduced due to the data size increase . in this paper , we perform cost - sensitive learning , which increases the intrinsic error function in under - represented regions . in regression tasks ,
such as we consider here , the output can be either discretized and treated as classes for the purpose of cost assignment , or a specific bias is used such as @xmath101 . to mimic a balanced data set in our setup
, the galaxies were grouped by their spectroscopic redshift using non - overlapping bins of width 0.1 .
the weights are then assigned as follows for balanced training : @xmath102 where @xmath103 is the error cost for sample @xmath68 , @xmath104 is the frequency of samples in bin number number @xmath68 , @xmath105 is the number of bins and @xmath106 is the set of samples in set number @xmath95 .
eq . assigns a weight to each training point which is the maximum bin frequency over the frequency of the bin in which the source belongs .
this ensures that the error cost of source @xmath68 is inversely proportional to its spectroscopic redshift frequency in the training sample .
the normalized weights are assigned as follows : @xmath107 after the weights have been assigned , they can be incorporated directly into the objective as follows : @xmath108 the difference between the objectives in eq . and
is the introduction of the diagonal matrix @xmath109 , where each element @xmath110 is the corresponding cost @xmath103 for sample @xmath68 . the first term in eq .
is a matrix operation form for a weighted sum of squares @xmath111 , where the solution can be found analytically as follows : @xmath112 the only modification to the gradient calculation is to set the matrix @xmath113 . in standard sum of squared errors , @xmath114 or the identity matrix .
it is worth emphasising that this component of the framework does not attempt to weight the training sample in order to match the distribution of the test sample , or matching the spectroscopic distribution to the photometric distribution as proposed in @xcite , @xcite and applied to photometric redshift in @xcite , but rather gives the user of the framework the ability to control the cost per sample to serve different science goals depending on the application . in this paper ,
the weighting scheme was used for two different purposes , the first was to virtually balance the data to mimic training on a uniform distribution , and the second was to directly target the weighted error of @xmath101 .
in this section , we specifically target the photometric bands and depths planned for _ euclid_. _ euclid _ aims to provide imaging data in a broad @xmath115 band and the more standard near - infrared @xmath116 , @xmath117 and @xmath118 bands , while ground - based ancillary data are expected in the optical @xmath119 , @xmath120 , @xmath68 and @xmath4 bands @xcite .
we use a mock dataset from @xcite , consisting of the @xmath119 , @xmath120 , @xmath68 , @xmath4 , @xmath115 , @xmath116 , @xmath117 and @xmath118 magnitudes ( to 10@xmath32 depths of 24.6 , 24.2 , 24.4 , 23.8 , 25.0 for the former , and 5@xmath32 depth of 24.0 for each of the latter three near - infrared filters ) for 185,253 simulated sources .
we remove all sources with @xmath121 to simulate the target magnitudes set for _ euclid_. in addition , we remove any sources with missing measurements in any of their bands prior to training ( only 15 sources ) .
no additional limits on any of the bands were used , however in section [ subsec - prior - mean ] we do explicitly impose limits on the riz band to test the extrapolation performance of the models .
the distribution of the spectroscopic redshift is provided in figure [ fig - zspec - histogram ] .
for all experiments on the simulated data , we ignore the uncertainties on the photometry in each band and train only on the magnitudes since in the simulated data set , unlike real datasets , the log of the associated errors are linearly correlated with the magnitudes , especially in the targeted range , therefore adding no information .
however , they were fed as input to annz to satisfy the input format of the code .
1 1 1 we preprocess the data using principle component analysis ( pca ; * ? ? ?
* ) to de - correlate the features prior to learning , but retain all features with no dimensionality reduction .
de - correlation accelerates the convergence rate of the optimization routine especially when using a logistic - type kernel machines such as neural networks @xcite . to understand this ,
consider a simple linear regression example where we would like to solve for @xmath57 in @xmath122 , the solution for this is @xmath123 .
note that if @xmath124 is de - correlated @xmath125 , therefore learning @xmath126 depends only on the @xmath68-th column of @xmath124 and it is independent from learning @xmath127 , where @xmath128 . in an optimization approach , the convergence rate is a function of the condition number of the @xmath129 matrix , which is minimized in the case of de - correlated data .
this represents a quadratic error surface which helps accelerate the search .
this is particularly important in the application addressed in this paper because the magnitudes measured in each filter are strongly correlated with each other .
five algorithms are considered to model the data ; artificial neural networks ( annz ; * ? ? ?
* ) , a gp with low rank approximation ( stablegp ; * ? ? ?
* ) , a sparse gp with global length scale ( gp - gl ) , a gp with variable length scale ( gp - vl ) and a gp with variable covariances ( gp - vc ) . for annz , a single layer network is used , and to satisfy the input format for the code , the data were not de - correlated and the uncertainties on photometry for each band were used as part of the training input . for stablegp , we use the sr - vp method proposed in @xcite . in subsequent tests
, the variable @xmath60 refers to the number of hidden units in annz , the rank in stablegp , and the number of basis functions in gp - gl , gp - vl and gp - vc .
the time complexities for each algorithm are shown in table [ table - time - complexity ] .
the data were split at random into 80 per cent for training , 10 per cent for validation and 10 per cent for testing .
we note that we investigate the accuracy for various training sample sizes in section [ sec - sizetraining ] .
all models were trained using the entire redshift range available , but we only report the performance on the redshift range of @xmath2 to target the parameter space set out in @xcite .
we train each model for 500 iterations in each run and the validation sample was used for model selection and parameter tuning , but all the results here are reported on the test sample , which is not used in any way during the training process .
table [ table - metrics ] shows the metrics used to report the performance of each algorithm .
+ annz ( @xmath130-layers ) & @xmath131 + stablegp & @xmath132 + gp - gl & @xmath133 + gp - vl & @xmath133 + gp - vc & @xmath134 + [ table - time - complexity ] [ cols="<,<,<",options="header " , ] [ table - final - results ] 0.35 0.35 0.35 0.35 0.35 0.35 0.35 0.35
in this paper a sparse gaussian process framework is presented and applied to photometric redshift estimation .
the framework is able to out perform annz , sparse gp parametrized by a set of global hyper - parameters and low rank approximation gp .
the performance increase is attributed to the handling of distribution bias via a weighting scheme integrated as part of the optimization objective , parametrizing each basis function with bespoke covariances , and integrating the learning of the prior mean function to enhance the extrapolation performance of the model .
the methods were applied to a simulated dataset and sdss dr12 where the proposed approach consistently outperforms the other models on the important metrics ( @xmath135 and @xmath136 ) .
we find that the model scales linearly in time with respect to the size of the data , and has a better generalization performance compared to the other methods even when presented with a limited training set .
results show that with only 30 per cent of the data , the model was able to reach accuracy close to that of using the full training sample . even when data were selectively removed based on @xmath115 magnitudes , the model was able to show the best recovery performance compared to the other models .
the cost - sensitive learning component of the framework regularizes the predictions to limit the effect caused by the biased distribution of the output and allows for direct optimization of the survey objective ( e.g. @xmath137 ) .
again , the algorithm consistently outperforms other approaches , including annz and stablegp , in all reported experiments .
we also investigate how the size of the training sample and the basis set affects the accuracy of the photometric redshift prediction .
we show that for the simulated set of galaxies , based on the work of @xcite , we are able to obtain a photometric redshift accuracy of @xmath138 and @xmath139 using 1600 basis functions which is a factor of seven improvement over the standard annz implementation . we find that gp - vc out - performed annz on the real data from sdss - dr12 , with an improvement in accuracy of @xmath140per cent , even when restricted to have the same number of free parameters . in future work
we will test the algorithm on a range of real data , and pursue investigations of how the algorithm performs over different redshift regimes and for different galaxy types .
iaa acknowledges the support of king abdulaziz city for science and technology .
mjj and snl acknowledge support from the uk space agency .
the authors would like to thank the reviewers for their valuable comments . | accurate photometric redshifts are a lynchpin for many future experiments to pin down the cosmological model and for studies of galaxy evolution . in this study ,
a novel sparse regression framework for photometric redshift estimation is presented .
synthetic dataset simulating the _ euclid _ survey and real data from sdss dr12 are used to train and test the proposed models .
we show that approaches which include careful data preparation and model design offer a significant improvement in comparison with several competing machine learning algorithms .
standard implementations of most regression algorithms use the minimization of the sum of squared errors as the objective function . for redshift inference , this induces a bias in the posterior mean of the output distribution , which can be problematic . in this paper
we directly minimize the target metric @xmath0 and address the bias problem via a distribution - based weighting scheme , incorporated as part of the optimization objective .
the results are compared with other machine learning algorithms in the field such as artificial neural networks ( ann ) , gaussian processes ( gps ) and sparse gps .
the proposed framework reaches a mean absolute @xmath1 , over the redshift range of @xmath2 on the simulated data , and @xmath3 over the entire redshift range on the sdss dr12 survey , outperforming the standard annz used in the literature .
we also investigate how the relative size of the training sample affects the photometric redshift accuracy .
we find that a training sample of > 30 per cent of total sample size , provides little additional constraint on the photometric redshifts , and note that our gp formalism strongly outperforms annz in the sparse data regime for the simulated data set .
[ firstpage ] methods : data analysis galaxies : distances and redshifts |
it is known @xcite that every vertex algebra @xmath1 is canonically filtered and therefore it can be considered as a quantization of its associated graded poisson vertex algebra @xmath2 .
the _ zhu s @xmath3-algebra _ @xmath4 of @xmath1 @xcite is a generating subalgebra of the differential algebra @xmath2 and has the structure of a poisson algebra .
its maximal spectrum @xmath5 is called the _ associated variety _ of @xmath1 ( @xcite ) .
the associated variety @xmath6 is a fundamental invariant of @xmath1 that captures some important properties of the vertex algebra @xmath1 itself ( see e.g. @xcite . as a poisson variety ,
the associated variety of a vertex algebra is a finite disjoint union of smooth analytic poisson manifolds , and it is stratified by its symplectic leaves .
the case where the associated variety has finitely many symplectic leaves is particularly interesting .
this happens for instance when @xmath1 is an admissible affine vertex algebra ( @xcite ) , when @xmath1 is the simple affine vertex algebra associated with a simple lie algebra that belongs to the the deligne exceptional series @xcite at level @xmath7 ( @xcite ) with @xmath8 the dual coxeter number , or when @xmath1 is the ( generalized ) drinfeld - sokolov reduction ( @xcite ) of the latter affine vertex algebras provided that it is nonzero ( @xcite ) .
this is also expected to happen for the vertex algebras obtained from four dimensional @xmath9 superconformal field theories ( @xcite ) , where the associated variety is expected to coincide with the spectrum of the chiral ring of the higgs branch of the four dimensional theory .
of course , it also happens when the associated variety of @xmath1 is a point , that is , when @xmath1 is _ lisse _ ( or @xmath3-cofinite ) . in our previous paper @xcite
we conjectured that , under reasonable assumptions on the vertex algebra @xmath1 , the associated variety @xmath10 is irreducible if it has only finitely many symplectic leaves ( see conjecture [ conj : equidim ] ) .
one of the aims of this paper is to verify this irreducibility conjecture for the known cases where @xmath10 has finitely many symplectic leaves .
it turns out this is a subtle problem for the associated varieties of ( conjecturally simple ) @xmath0-algebras , and some deep results on the geometry of nilpotent orbits are needed for the verifications .
more precisely , the associated variety in question has the form @xmath11 where @xmath12 is a nilpotent orbit in a simple lie algebra @xmath13 and @xmath14 is _ the _ slodowy slice at a nilpotent element @xmath15 of @xmath13 . the variety @xmath16 is called a _ nilpotent slodowy slice _ ( @xcite ) .
it is not irreducible in general . on the other hand ,
not every nilpotent slodowy slice appears as the associated variety of some @xmath0-algebra .
we investigate in detail the irreducibility of the variety @xmath16 that appears as the associated variety of a @xmath0-algebra and confirm the irreducibility for all the known cases .
another aim of this paper is to provide new examples of vertex algebras whose associated variety has finitely many symplectic leaves .
we do this by proving a conjecture stated in our previous article ( * ? ? ?
* conjecture 2 ) ( theorem [ th : amconj ] ) and also by showing that the associated variety of the simple affine vertex algebra associated with @xmath17 at level @xmath18 is the short nilpotent orbit closure ( theorem [ th : short ] ) .
theorems [ th : amconj ] and [ th : short ] give new examples of non - admissible affine vertex algebras whose associated variety is contained in the nilpotent cone .
we thank very much daniel juteau for his explanations about branchings and nilpotent slodowy slices .
we thank drazen adamovi for inviting us to the university of zagreb in june 2016 , and the cirm luminy for its hospitality during our stay as `` research in pairs " in august , 2016 .
the first named author is supported by jsps kakenhi grant numbers 25287004 and 26610006 .
the second named author is supported by the anr project geolie grant number anr-15-ce40 - 0012 .
let @xmath1 be a vertex algebra , and let @xmath19,\quad a\mapsto a(z)=\sum_{n\in { \mathbb{z}}}a_{(n)}z^{-n-1},\]]be the state - field correspondence .
the zhu s @xmath3-algebra is by definition the quotient space @xmath20 , where @xmath21 , equipped with the poisson algebra structure given by @xmath22 for @xmath23 with @xmath24 .
the associated variety @xmath10 of @xmath1 is the reduced scheme of the spectrum of @xmath4 , @xmath25 . in (
* ? ? ? * conjecutre 1 ( 2 ) ) , we stated the following conjecture .
[ conj : equidim ] let @xmath26 be a simple , finitely strongly generated ( i.e. , @xmath4 is finitely generated ) , positively graded conformal vertex operator algebra such that @xmath27 .
assume that @xmath10 has finitely many symplectic leaves .
then @xmath10 is irreducible . in the following we list the known examples of vertex algebras whose associated variety has finitely many symplectic leaves .
recall @xcite that a vertex algebra is _ lisse _ ( or _ @xmath3-cofinite _ ) if @xmath28 .
under the assumptions on @xmath1 of conjecture [ conj : equidim ] , @xmath29 and @xmath30 , so @xmath10 is conic , that is , admits a @xmath31-action that is contracting to a point .
therefore , under the assumptions on @xmath1 of conjecture [ conj : equidim ] , @xmath1 is lisse if and only if @xmath32 . hence , if so , @xmath10 is a trivial symplectic variety and is irreducible .
let @xmath13 be a complex simple lie algebra , and @xmath33 the affine kac - moody algebra associated with @xmath13 : @xmath34{\mathop{\oplus}}{\mathbb{c}}k { \mathop{\oplus}}{\mathbb{c}}d,\end{aligned}\ ] ] where the commutation relations are given by @xmath35=[x , y]{{\otimes}}t^{m+n}+m(x|y)\delta_{m+n,0}k,\quad [ k,{\widehat{{\mathfrak}{g}}}]=0,\quad [ d , x{{\otimes}}t^m]=mx{{\otimes}}t^{m}\end{aligned}\ ] ] for @xmath36 , @xmath37 . here
@xmath38killing form of @xmath13 . for @xmath39 ,
set @xmath40{\mathop{\oplus}}{\mathbb{c}}k{\mathop{\oplus}}{\mathbb{c}}d)}{\mathbb{c}}_k,\end{aligned}\ ] ] where @xmath41 is the one - dimensional representation of @xmath42{\mathop{\oplus}}{\mathbb{c}}k{\mathop{\oplus}}{\mathbb{c}}d$ ] on which @xmath43 acts as multiplication by @xmath44 and @xmath42{\mathop{\oplus}}{\mathbb{c}}d$ ] acts trivially .
as it is well - known @xmath45 is naturally a vertex algebra , called the _
universal affine vertex algebra _ associated with @xmath13 at level @xmath44 .
the unique simple ( graded ) quotient @xmath45 is called the _ simple affine vertex algebra _ associated with @xmath13 at level @xmath44 .
we have @xmath46 , and therefore the associated variety @xmath47 is a poisson subscheme of @xmath48 which is @xmath49-invariant and conic , where @xmath49 is the adjoint group of @xmath50 .
thus , identifying @xmath50 with @xmath51 through @xmath52 , the symplectic leaves of @xmath47 are exactly the adjoint orbits of @xmath49 in @xmath47 , and @xmath47 has finitely many symplectic leaves if and only if @xmath47 is contained in the nilpotent cone @xmath53 of @xmath13 .
we list in table [ tab : known - cases ] the known pairs @xmath54 where @xmath55 .
we do not claim that the list is exhaustive .
.known pairs ( @xmath56 ) for which @xmath57 [ cols="^,^,^,^ " , ] the above verifications show that , except for the non - normal nilpotent orbit @xmath58 of dimension 8 in @xmath59 that appears as the associated variety of simple affine vertex algebras of type @xmath59 at admissible levels with denominator @xmath60 @xcite , all the other nilpotent orbits appearing in table [ tab : known - cases ] are actually normal .
in this section we prove the following assertion . [
th : amconj ] let @xmath61 with @xmath62 even , @xmath63 , and let @xmath64 .
then @xmath65 let @xmath61 with @xmath62 even , @xmath66 .
let @xmath67 be the root system of @xmath50 and take @xmath68 for the set of positive roots . by theorem 1.3 of @xcite
we know that @xmath69 . hence it is sufficient to show that @xmath70 . by theorem
[ th : w - algebra - variety ] ( 1 ) , this is equivalent to that @xmath71 let @xmath72 , and let @xmath73 be an @xmath74-triple in @xmath13 .
the weighted dynkin diagram of @xmath15 is the following .
@xmath75 thus , we may assume that @xmath76 , where @xmath77 is the @xmath78-th fundamental coweight of @xmath13 .
we have @xmath79 where @xmath80=2jx\}$ ] .
set @xmath81 , so that @xmath82 .
set @xmath83 and put @xmath84}=\{v\in v_k({{\mathfrak}{g}})\mid d_h v = d v\}$ ] .
since @xmath85 for any positive real root @xmath86 of @xmath33 by , we have @xmath87}.\end{aligned}\ ] ] the operator @xmath88 extends to the grading operator of @xmath89 as it commutes with the differential of the complex associated with the drinfeld - sokolov reduction ( @xcite ) .
thus , @xmath90},\end{aligned}\ ] ] where @xmath91}=\{c\in h_{ds , f}^0(v_k({{\mathfrak}{g}}))\mid d_h c = d c\}$ ] .
now , let @xmath92 be the sualgebra of @xmath33 generated by @xmath93 , @xmath94 , @xmath95 , @xmath96 , @xmath97 , @xmath98 .
then @xmath92 is isomorphic to @xmath99 and @xmath92 acts on the each homogeneous subspace @xmath84}$ ] as its elements commute with @xmath88 .
we have @xmath100 where @xmath101 is the subalgebra of @xmath13 corresponding to the roots @xmath102 with @xmath103 , and @xmath104 is the subalgebra corresponding to the roots @xmath105 .
let @xmath106 be the parabolic subalgebra of @xmath92 whose nilradical @xmath107 is @xmath108 .
the levi subalgebra @xmath109 is isomorphic to @xmath110 the @xmath92-module @xmath111}$ ] is a highest weight representation of @xmath92 with highest weight @xmath112 under the isomorphism @xmath113 .
this weight verifies the conditions of the jantzen s simplicity criterion @xcite ( see also ( * ? ? ?
* section 9.13 ) ) .
thus from jantzen s simplicity criterion one finds that @xmath114}\cong u({\mathfrak}{a}){{\otimes}}_{u({\mathfrak}{p})}{\mathbb{c}}|0{{\rangle}}.\end{aligned}\ ] ] the nilpotent element @xmath115 belongs to @xmath116 under the isomorphism @xmath113 , and is a very even nilpotent element that corresponds to the partition @xmath117 .
the weighted dynkin diagram of @xmath118 as nilpotent element of @xmath119 is given by @xmath120 from the definition of the drinfeld - sokolov reduction we find that @xmath121}\cong h_0({\mathfrak}{m}_- , v_k({{\mathfrak}{g}})_{[0]}{{\otimes}}{\mathbb{c}}_{\chi}),\ ] ] where @xmath122 is the opposite algebra of @xmath107 spanned by @xmath123 , @xmath124 , so that @xmath125 , and @xmath126 is the one - dimensional representation of @xmath122 defined by the character @xmath127 .
since @xmath111}$ ] is a free @xmath128-module by , it follows that @xmath129}{{\otimes}}{\mathbb{c}}_{\chi})\cong h_0({\mathfrak}{m}_- , v_k({{\mathfrak}{g}})_{[0]})\cong { \mathbb{c}}$ ] , and therefore @xmath130}$ ] is nonzero .
thus , @xmath131 is nonzero , and this completes the proof .
let @xmath132 . by theorem
[ th : w - algebra - variety ] and theorem [ th : amconj ] , @xmath133 , for @xmath62 even , is lisse .
the central charge of @xmath134 is given by @xmath135 in particular the central charge of @xmath133 is zero .
note that the central charge of @xmath136 is zero if @xmath136 is trivial .
[ conj : lisse - typed ] let @xmath62 be even , @xmath63 , and @xmath132 .
1 . @xmath137 .
2 . @xmath134 is lisse for any integer @xmath44 such that @xmath138 .
parts ( 1 ) and ( 2 ) of conjecture [ conj : lisse - typed ] are true for @xmath139 by @xcite . also , part ( 2 ) of the conjecture is true for @xmath64 as mentioned just above .
in this section , @xmath141 with @xmath142 .
let @xmath143 be the root system of @xmath50 and take @xmath144 for the set of positive roots
. denote by @xmath145 the chevalley generators of @xmath50 in the bourbaki numbering , and fix the root vectors @xmath146 , @xmath147 so that @xmath148 is a chevalley basis satisfying the conditions of ( * ? ? ?
* chapter iv , definition 6 ) .
denote by @xmath149 the corresponding simple roots of @xmath50 .
let @xmath150 be the corresponding triangular decomposition . for @xmath147 , denote by @xmath151 $ ] the corresponding coroot
. denote by @xmath152 the fundamental weights and by @xmath153 the corresponding fundamental co - weights .
note that @xmath77 and @xmath154 are proportional , with @xmath155 the isomorphism induced from @xmath52 . in particular @xmath156 .
there is a unique short nilpotent orbit @xmath157 in @xmath13 , which is the nilpotent orbit associated with the @xmath74-triple @xmath158 , where @xmath159 is the highest short root @xmath160 and where @xmath161 .
so , @xmath162 where @xmath163=2jx\}$ ] . the nilpotent orbit @xmath157 is labeled by the partition @xmath164 , and its weighted dynkin diagram is given by @xmath165 we have @xmath166 in particular , @xmath167 is simple . as a module over @xmath168
, @xmath169 decomposes as @xmath170 let @xmath174 be the singular vector with respect to the adjoint action of @xmath50 as in ( * ? ? ?
* theorem 4.2 ) which generates the irreducible representation @xmath175 of @xmath50 with highest weight @xmath176 in @xmath177 , where @xmath177 is the component of @xmath178 of degree @xmath60 and @xmath179 is the highest root of the simple lie algebra generated by the roots @xmath180 .
namely , @xmath181 with @xmath182 , @xmath183 , @xmath184 , @xmath185 then @xmath186 is a singular vector of @xmath187 if and only if @xmath172 ( * ? ?
* theorem 4.2 ) where @xmath188 is the natural embedding of @xmath50-modules from @xmath189 to @xmath190 ( see ( * ? ? ?
* lemma 4.1 ) ) .
we observe that that @xmath194 is the unique minimal @xmath195-degeneration of @xmath196 ( see definitions [ def : degeneration ] and [ def : deg - min ] ) .
therefore , it is enough to show that @xmath191 do not contain @xmath197 . to proceed ,
we apply ( * ? ? ?
* lemma 3.3 ) and we argue as in the proofs of lemma 7.6 and lemma 9.2 of @xcite .
since the verification are similar we omit the details here .
it is easy to verify the statement for @xmath201 and @xmath139 where @xmath175 has highest weight @xmath202 . for @xmath203
, @xmath175 has highest weight @xmath204 .
hence @xmath205 and it is easy to show that the zero weight space of @xmath175 has dimension @xmath200 .
[ lem : b - gen ] we can describe a set of generators of @xmath207 by induction on @xmath208 as follows . set @xmath209 assume that @xmath210 and that generators @xmath211 of @xmath212 have been constructed for any @xmath213 .
set : @xmath214 where @xmath215 is a homogeneous polynomial of degree 2 in the variables @xmath216 with no term in @xmath217 .
then @xmath218 generate @xmath207 .
first of all , we have the following relations whose verifications are left to the reader : @xmath221 \equiv h_{\delta_1+\theta_2 } h_{\beta_1+\theta_2 } , \qquad [ f_{\delta_2+\theta_2 } , [ f_{\beta_2+\theta_2 } , w_2 ] ] \equiv h_{\delta_2+\theta_2 } h_{\beta_2+\theta_2 } , \\ & [ f_{\theta_2 } , [ f_{\theta } , w_2 ] ] \equiv h_{\theta } h_{\theta_2 } , \qquad [ f_{\alpha_4},[f_{\alpha_3+\alpha_4 + 2(\alpha_5 + \cdots+\alpha_{{r } } ) } , [ f_{\theta},w_2 ] ] ] \equiv h_{\theta } h_{\alpha_4 } , \\ & [ f_{\alpha_3+\alpha_4},[f_{\alpha_4 + 2(\alpha_5 + \cdots+\alpha_{{r } } ) } , [ f_{\theta},w_2 ] ] ] \equiv h_{\theta } h_{\alpha_3+\alpha_4 } , \end{aligned}\ ] ] where @xmath222 means that the equality is modulo @xmath223 .
moreover , if @xmath203 then for any @xmath224 : @xmath225 ] & \equiv h_{\theta } h_{\alpha_4+\cdots+\alpha_j } .
& \end{aligned}\ ] ] from this , we deduce that the elements @xmath226 are in @xmath207 .
moreover , if @xmath203 , then the element @xmath227 is in @xmath207 for any @xmath224 . for @xmath203 , @xmath228 and for @xmath229
, @xmath230 then , since the elements @xmath231 , @xmath232 for @xmath224 , and @xmath233 are in @xmath207 , the elements @xmath234 also belong to @xmath207 for @xmath235 and @xmath210 . in conclusion ,
the elements @xmath236 are in @xmath207 .
@xmath237 @xmath229 .
let @xmath246 be the subalgebra of @xmath50 of type @xmath247 generated by @xmath248 and assume the statement true for @xmath246 .
we denote by @xmath249 the elements of @xmath250 corresponding to @xmath246 , and by @xmath251 , @xmath252 , the roots corresponding to @xmath253 , @xmath252 , for @xmath246 .
a direct computation shows that @xmath254 = e_{\theta ' } e_{\theta'_2}-\sum_{k=1}^2 e_{\beta'_j+\theta'_2 } e_{\delta'_j+\theta'_2 } = w'_2.\end{aligned}\ ] ] it follows from ( [ eq : ind1 ] ) that the elements @xmath255 , @xmath256 and @xmath257 are in @xmath258 .
moreover , by ( [ eq : elem ] ) , since @xmath259 , @xmath260 , @xmath261 and @xmath262 are in @xmath258 , the elements @xmath263 , @xmath264 and @xmath265 are in @xmath258 too .
hence the statement is true for @xmath229 .
@xmath237 assume the statement true for any rank @xmath266 .
let @xmath246 be the subalgebra of @xmath50 of type @xmath267 generated by @xmath268 and assume the statement true for @xmath246 .
we denote by @xmath249 the elements of @xmath250 corresponding to @xmath174 for @xmath246 , and by @xmath251 , @xmath252 , the roots corresponding to @xmath253 , @xmath252 , for @xmath246 .
a direct computation shows that @xmath269 = e_{\theta ' } e_{\theta'_2}-\sum_{k=1}^2 e_{\beta'_j+\theta'_2 } e_{\delta'_j+\theta'_2 } = w'_2.\end{aligned}\ ] ] from ( [ eq : ind ] ) , we deduce that the elements @xmath270 , for @xmath271 , viewed as polynomials in the variables @xmath216 , are in @xmath207 .
in addition , by ( [ eq : elem ] ) , @xmath272 , @xmath273 , @xmath274 and @xmath275 , @xmath276 , are in @xmath207 .
we first deduce that @xmath277 is in @xmath278 next , for @xmath279 , we have @xmath280 whence @xmath281 by the induction hypothesis .
finally , we deduce that @xmath282 is in @xmath207 since @xmath281 for any @xmath283 .
we have proven the expected statement for @xmath208 .
[ lem : b - zero ] let @xmath284 be a nonzero semisimple element of @xmath285 which belongs to @xmath191 .
then either @xmath286 , or @xmath287 for some @xmath288 , or @xmath289 .
in particular , @xmath290 if and only if @xmath291 let @xmath294 that we write as @xmath295 , @xmath296 .
it suffices to prove that if @xmath290 , then @xmath284 is the union of the sets as described in the lemma .
indeed , it is easy to verify that , conversely , these sets all lie in @xmath191 .
we prove the statement by induction on @xmath208 .
assume that @xmath290 .
by lemma [ lem : b - gen ] , we get @xmath297 so , if @xmath298 , then either @xmath299 , or @xmath300 , and if @xmath301 , then @xmath302 and @xmath303 , whence the statement .
assume @xmath210 and the statement true for any rank @xmath304 .
we have to solve the systems of equations @xmath305 for @xmath306 .
let @xmath246 be as in the proof of lemma [ lem : b - gen ] .
if @xmath302 then by induction applied to @xmath246 , we get the statement .
otherwise , using lemma [ lem : b - gen ] and the equation @xmath307 , for @xmath308 , we deduce that @xmath309 .
therefore , from @xmath310 we deduce that either @xmath299 or @xmath300 , whence the statement .
denote by @xmath311 the standard levi subalgebra generated by @xmath268 .
then @xmath312 and its center is generated by the semisimple element @xmath313 .
let @xmath314 be the dixmier sheet associated with @xmath311 ( cf .
it is the the unique sheet containing the nilpotent orbit @xmath315 of @xmath316 .
in fact , @xmath317 where @xmath318 is the induced nilpotent orbit of @xmath13 from @xmath319 in @xmath311 , and @xmath315 can not be induced in another way , see e.g. ( * ? ? ? * appendix ) .
we have ( cf .
e.g. @xcite ) @xmath320 let @xmath322 be the quotient of @xmath323 by the submodule generated by @xmath186 .
then @xmath324 is the zero locus of the @xmath325-submodule of @xmath326 generated by the image of @xmath174 in @xmath327 , that is , @xmath191 . by lemma
[ lem2:b - zero ] , lemma [ lem : b - zero ] and ( * ? ? ?
* lemma 2.1 ( 2 ) ) , we obtain that @xmath328 .
this completes the proof as @xmath45 is a quotient of @xmath322 .
consider the subalgebra @xmath92 of @xmath33 generated by @xmath93 , @xmath94 , @xmath95 , @xmath96 , @xmath331 , @xmath98 , which is isomorphic to @xmath13 .
the vacuums vector @xmath332 generates the irreducible representation of @xmath333 with highest weight @xmath334 .
then as in the proof of theorem [ th : amconj ] , we use jantzen s simplicity criterion @xcite ( see also ( * ? ? ?
* section 9.13 ) ) to find that @xmath335 where @xmath336 is the parabolic subalgebra of @xmath92 generated by @xmath93 , @xmath94 , @xmath95 , @xmath96 .
it follows from ( * ? ? ?
* theorem 6.3 ) that @xmath329 is nonzero and almost irreducible .
as @xmath329 is cyclic , the almost irreduciblity is the same as the simplicity .
finally , the last assertion follows from theorem [ th : w - algebra - variety ] ( 2 ) .
let @xmath343 denote the universal virasoro vertex algebra at central charge @xmath344 , and @xmath345 the unique simple quotient of @xmath343 .
we know that the following conditions are equivalent ( @xcite , see also @xcite ) : let @xmath322 be as in the proof of proposition [ pro : sheet ] .
then @xmath353 by theorem [ th : w - algebra - variety ] .
since @xmath354 is a semisimple element that is conjugate to @xmath355 , we have ( see ( * ? ? ?
* remark 3.5 ) ) , @xmath356 this implies that @xmath357 is the polynomial ring generated by the image of the conformal vector @xmath358 of @xmath359 .
hence , @xmath359 is strongly generated by @xmath358 .
as @xmath45 is a quotient of @xmath322 , the exactness of @xmath360 ( theorem [ th : w - algebra - variety ] ( 1 ) ) implies that @xmath361 is also strongly generated by its conformal vector .
since it is simple by proposition [ pro : nonzero - and - simple ] , @xmath361 must be isomorphic to @xmath362 . by proposition [ pro : sheet ] and proposition [ pro : nonzero - and - simple ] , @xmath47 is a @xmath49-invariant subvariety of @xmath338 containing @xmath339 .
thus @xmath363 or @xmath364 by lemma [ lem : or ] .
but if @xmath363 then @xmath365 is one - dimensional and @xmath366 can not be lisse , which contradicts theorem [ th : reduction - is - simple - virasoro ] , whence @xmath364 .
y. namikawa , birational geometry of symplectic resolutions of nilpotent orbits , moduli spaces and arithmetic geometry , 75116 . , 45 , math . soc .
japan , tokyo , 2006 preprint http://arxiv.org/pdf/math/0408274v1.pdf . | we investigate the irreducibility of the nilpotent slodowy slices that appear as the associated variety of @xmath0-algebras .
furthermore , we provide new examples of vertex algebras whose associated variety has finitely many symplectic leaves . _
dedicated to the 60th birthday of professor efim zelmanov _ |
in the weak coupling picture of the pre - thermal evolution of heavy - ion collisions , the post collision debris that end up in the mid - rapidity region undergo several stages that are characterised by widely different physics and degrees of freedom . on the one hand , according to the saturation paradigm @xcite , at very early times @xmath3 , where @xmath4 is the typical energy scale right after the collision , the energy is deposited in strong color fields .
these strong fields admit a description in terms of classical yang - mills theory to leading order in t hooft coupling @xmath5 .
indeed , there have been several interesting studies of classical yang - mills fields under longitudinal expansion in the past years @xcite . on the other hand ,
once the system has reached a local thermal equilibrium , or at least approximately isotropized @xcite , the matter in the mid - rapidity region ( at sufficiently low @xmath6 ) is described by relativistic fluid dynamics @xcite .
there has been a sizeable and very successful program of numerical simulations of relativistic hydrodynamics .
it is well known that classical yang - mills theory can not reach thermalization due to the rayleigh - jeans catastrophe nor can it isotropize when exposed to rapid longitudinal expansion as noticed by @xcite .
instead classical evolution drives the system further away from equilibrium , making the system less occupied ( i.e. weaker fields ) but more anisotropic , which has also been observed in the simulations of @xcite .
therefore , there is a missing link between the physics of saturated gluon fields and fluid dynamics that needs to be bridged in order to fully carry the predictions of saturation physics to the hydrodynamic regime .
this gap can be filled with the systematically improvable framework of effective kinetic theory ( ekt ) of @xcite .
the ekt faithfully describes to leading order in @xmath7 systems where the typical occupancies of gluons are not nonperturbative @xmath8 and have momentum significantly larger than the in - medium screening scale @xmath9 . at weak coupling
, these conditions are certainly fulfilled in thermal equilibrium and therefore we can describe the system all the way to the equilibrium .
while these conditions are not fulfilled at the very earliest times , both ekt and classical yang - mills give an equally valid leading order description for a wide range of large but perturbative occupancies @xmath10 @xcite .
therefore a possible strategy to simulate the system through all time scales is to start the simulation with a classical yang - mills simulation ( for @xmath11 and @xmath12 ) and subsequently pass the system to ekt at some arbitrary time @xmath13 .
then , once ekt has brought the system sufficiently close to the thermal equilibrium , both hydrodynamics and ekt should give equivalent descriptions and the system can be passed to a hydrodynamical simulation at some arbitrary time @xmath14 .
the proof of principle of such a procedure was shown in a series of papers @xcite in an isotropic setting in the absence of expansion . here , we present first results of numerical simulations of the ekt in a boost invariant 2 + 1d setting and demonstrate the connection with both classical yang - mills simulations and viscous hydrodynamics .
we make a first attempt to model the early stages of heavy - ion collision starting from the over - occupied region all the way to the thermal equilibrium and find that after time @xmath15 the time evolution of energy density is described by viscous hydrodynamics to a better than @xmath16 accuracy for a realistic coupling @xmath17 corresponding to @xmath18 .
the current result is , however , incomplete in two ways .
firstly , there have been significant steps to describe the melting of the strong color - fields in classical statistical field theory @xcite , but currently we do not have a reliable initial state from a 3 + 1d simulation to enter into our ekt simulation .
therefore we initialize our system at @xmath19 with the energy density given by 2 + 1d simulation @xcite and vary the parameters of the initial condition to quantify the ignorance of the very early time dynamics . secondly
, certain non - perturbative chromo - weibel instabilities @xcite arising from anisotropic screening play a parametrically leading order role during the whole non - equilibrium evolution @xcite .
however , numerical simulations of classical fields @xcite have not been able to see them beyond the very early times , suggesting that even if parametrically leading order , their effect is numerically small for values of @xmath20 that are phenomenologically interesting . therefore ,
as a formally correct treatment of anisotropic screening is rather complicated @xcite and not a fully solved problem , we in this first work will treat the screening in an isotropic way .
therefore , our results are correct to leading order in @xmath20 for systems that are close to isotropy ( late times ) and for large anisotropies our results have leading logarithmic accuracy apart from the instabilities .
the ekt of @xcite is defined though the effective boltzmann equation for the color and spin averaged distribution function of gluons , and to leading order in @xmath20 it contains effective @xmath21 scattering and @xmath22 splitting terms .
@xmath23 + \mathcal{c}_{2\leftrightarrow 2}[f_\p ] + \mathcal{c}_{\rm exp } [ f_\p].\end{aligned}\ ] ] we restrict ourselves to azimuthally symmetric distributions but allow anisotropy in the @xmath24-direction so that it is enough to specify @xmath25 with @xmath26 .
the effect of longitudinal expansion is encapsulated in @xmath27(\p ) = - \frac{p_z}{\tau}\frac{\partial } { \partial p_z } f(\p)$ ] @xcite .
the @xmath21 effective scattering term reads @xmath28(\tilde{\p})=
\frac{(2\pi)^3 } { 4\pi \tilde p^2 } \frac{1}{8 \nu}\int d\gamma_{ps}|\mathcal{m}|^2 \nonumber \\ & \times \left(f_\p \ , f_\k \ , g_\pp \ , g_\kp - f_\pp \ ,
f_\kp \ , g_\p \ , g_\k \right)\label{c22 } \\ & \times \left [ \delta(\tilde{\p}-\p)+\delta(\tilde{\p}-\k)-\delta(\tilde{\p}-\pp)-\delta(\tilde{\p}-\kp)\right ] , \nonumber\end{aligned}\ ] ] where @xmath29 and @xmath30 .
@xmath31 is the integral measure over the phase space of @xmath21 processes , singling out the @xmath32-direction @xmath33 in these coordinates the angles of incoming and outgoing momenta read @xmath34 this relation holds for @xmath35 , @xmath36 , and @xmath37 with @xmath38 , @xmath39 , @xmath40 and @xmath41 .
the effective matrix element for most kinematics is the ordinary vacuum tree - level element @xmath42 for small @xmath43 or @xmath44 the tree - level element has an ir divergence that is regulated by the physics of screening . in the formulation of @xcite ,
the screening is treated by replacing the naive tree - level matrix element by the retarded hard thermal loop ( htl ) resummed expression in the soft kinematic region .
the resummed propagator in the anisotropic case , however , contains a non - integrable singularity , signalling the presence of an instability in the system .
therefore , in the anisotropic system this procedure is not fully satisfactory . here
, in this work we will restrict ourselves to isotropic screening and use the prescription derived in @xcite , replacing in the denominator ( similarly for @xmath45 ) @xmath46 with @xmath47 , which is leading order accurate in the case of isotropic distribution .
the effective @xmath48 collinear splitting term reads @xmath49(\tilde{\p } ) = \frac{(2\pi)^3 } { 4\pi \tilde p^2 } \frac{1}{\nu}\int_0^{\infty}dp\int_0^{p/2}dk ' \left [ 4\pi \gamma(p ; p ' , k')\right]\nonumber \\ & \times \left ( f_{x_p , p } g_{x_p , p ' } g_{x_p , k ' } - g_{x_p , p } f_{x_p , p ' } f_{x_p , k ' } \right ) \nonumber \\ & \times \left [ \delta(\tilde{p}-p)-\delta(\tilde{p}-p')-\delta(\tilde{p}-k')\right ] , \label{c12 } \end{aligned}\ ] ] where @xmath50 , and the effective splitting rate reads @xmath51 again , consistently assuming only isotropic screening , the function @xmath52 is given by the solution to the linear integral equation @xmath53\nonumber.\end{aligned}\ ] ] @xmath54 and @xmath55 .
we solve this equation using an efficient numerical method introduced in @xcite .
for the numerical solution of the ekt , we discretize @xmath56 on a 2d grid and monte carlo estimate the integrals of eqs .
( [ gamma_ps],[c12 ] ) .
we use a logarithmically spaced grid for both @xmath57 and @xmath58 with at least 250 grid points in angular direction and at least 100 in the @xmath57-direction .
we have varied the number of grid points to verify that the results are insensitive to the number of grid points .
our algorithm , the discrete-@xmath57 method of @xcite , conserves energy exactly and has exactly correct particle number violation for @xmath48 term .
( solid lines ) and @xmath59 ( dahsed lines ) and varying coupling @xmath20 in a plane of mean occupancy ( weighted by the energy of particles ) and anisotropy .
the @xmath60 line corresponds to classical field approximation .
the violet dots refer to the times in fig . [ pz_scaling ] .
the simulations at finite coupling reach thermal equilibrium located at points indicated by the black crosses.,scaledwidth=45.0% ]
we shall now apply ekt to simulate the prethermal evolution of the expanding fireball created in a heavy - ion collision . in saturation framework , the initial condition is typically described in terms of `` gluon liberation coefficient '' @xmath61 and mean transverse momentum @xmath62 @xcite .
the gluon liberation coefficient is proportional to the total gluon multiplicity per unit rapidity @xmath63 after the classical fields have decohered and can be described in terms of quasi - particles .
lappi @xcite finds in jimwlk evolved mv model values relevant for heavy - ion collisions relevant for lhc of roughly @xmath64 and @xmath65 extracted at time @xmath66 from a 2d classical yang - mills simulation . by construction the distribution
then has @xmath67 .
but it is has been noted @xcite that certain plasma instabilities will broaden the distribution in @xmath68 in a time scale @xmath69 .
therefore , as a rough estimate of the initial condition we instead take somewhat arbitrarily our initial condition at the time @xmath70 to be @xmath71 choosing @xmath72 such that comoving energy density @xmath73 is fixed .
we then vary @xmath74 to quantify our ignorance of the initial nonperturbative dynamics .
figure [ trajectories ] displays a set of trajectories from simulations with varying @xmath20 and @xmath75 on a plane of mean occupancy ( weighted by the energy of particles ) and anisotropy measured by the ratio of the transverse and longitudinal pressures @xmath76 .
the line with @xmath60 corresponds to the classical field limit @xmath77 with fixed @xmath7 , which is obtained in ekt by including only the highest power of @xmath78 s in eqs.([c22],[c12 ] ) .
the classical field theory can not thermalize and indeed instead it flows to a stationary scaling solution . by performing classical yang - mills simulations berges et .
al have established that the scaling solution can described by a scaling form of the distribution function @xcite , @xmath79 where @xmath80 is approximately constant as a function of time .
this behavior is demonstrated in figure [ pz_scaling ] where we plot a section of rescaled distribution function @xmath80 measured at various times as a function for @xmath81 at fixed @xmath82 following berges et al .. our results corroborate that such a scaling solution exists at late times within the classical approximation and we observe that the scaling regime is reached after a time @xmath83 . moving on to the finite but small coupling @xmath84 , we see qualitative agreement with the parametric picture of bottom - up thermalization of @xcite : there are three distinct stages of evolution visible . in the first stage
the classical evolution drives the system more anisotrpic and less occupied .
once the occupancies reach @xmath85 , there is a qualitative change in the dynamics of the system as bose enhancement is lost .
this has the effect that ansiotropy freezes but the system still continues to get more dilute . only in the last stage which is characterized by a radiational break up of the particles at the scale @xmath4 ,
does the trajectory turn back and reach thermal equilibrium , denoted by the black crosses in the figure [ trajectories ] . for larger values of coupling @xmath86 , these features become however less pronounced and the system takes more straight trajectory towards equilibrium .
it should be noted that for these values of @xmath20 the assumption of @xmath87 is not satisfied and large nlo corrections are to be expected . at @xmath88 in classical approximation at vastly different times .
the good overlap of the curves indicates that system has reached the classical scaling solution of eq .
( [ eq : scaling ] ) .
in contrast , @xmath89 has not yet reached the scaling solution . [ pz_scaling ] , scaledwidth=35.0% ]
we expect that late time evolution should be described by relativistic hydrodynamics . under flow with translational
invariance along transverse directions and boost invariance , the hydrodynamical relations read to second order in gradients @xcite @xmath90 with longitudinal and transverse pressure @xmath91 and @xmath92 .
first order hydrodynamics corresponds to setting @xmath93 in eq .
( [ 1st ] ) . at weak coupling ,
the transport coefficients @xmath94 , @xmath95 and @xmath96 are known @xcite leaving zero free parameters to fit . besides a time when the hydrodynamics is initialized .
we initialize the 1st order hydrodynamics at the latest time we have in our simulation and integrate eq . ( [ 1st ] ) backwards in time . for 2nd order hydrodynamics integrating backwards
is highly unstable and we initialize the energy density at some arbitrary earlier time and integrate forwards in time . , energy density @xmath97 , and transverse pressure @xmath98 from a simulation with @xmath99 and @xmath100 ( top panel ) and @xmath101 ( bottom panel ) .
the components of @xmath102 have been scaled by @xmath103 so that the ideal hydrodynamics corresponds to horizontal lines .
the scale on x - axis with [ fm / c ] corresponds to @xmath104gev@xmath105/fm .
the simulations with @xmath106 are also displayed with thin dotted lines.,title="fig:",scaledwidth=40.0% ] + , energy density @xmath97 , and transverse pressure @xmath98 from a simulation with @xmath99 and @xmath100 ( top panel ) and @xmath101 ( bottom panel ) .
the components of @xmath102 have been scaled by @xmath103 so that the ideal hydrodynamics corresponds to horizontal lines .
the scale on x - axis with [ fm / c ] corresponds to @xmath104gev@xmath105/fm .
the simulations with @xmath106 are also displayed with thin dotted lines.,title="fig:",scaledwidth=40.0% ] in figure [ hydro ] we examine the validity of the hydrodynamical expansion at small @xmath100 and at realistic intermediate @xmath107 ( @xmath18 ) values of coupling . in both cases we see that the evolution of the components of the energy momentum tensor asymptotes to their hydrodynamical values . in case of @xmath100 , the hydrodynamical behaviour is reached only at a rather late time @xmath108 .
we have checked that including 2nd order terms before this time does not make the convergence significantly better ; before this time the evolution differs qualitatively from the hydrodynamical prediction . however , rather remarkably , for @xmath107 even 1st order hydrodynamics gives a very good description of the data all the way to very early times @xmath109 ( corresponding to @xmath1101fm / c for @xmath104gev ) where the ratio of the pressures is still as large as @xmath111 .
in addition , including the second order terms significantly improves the convergence .
indeed , we find that initializing the 2nd order hydrodynamics at @xmath112 leads to only 10% relative error in the energy density at late times .
the dashed lines correspond to @xmath113 whereas the solid ones have @xmath59 .
[ hydro_diff ] , scaledwidth=40.0% ] the parametric estimate of baier et al .
@xcite for the time when the hydrodynamic behaviour sets is @xmath114 .
this arises from equating the age of the system @xmath115 with the time scale @xmath116 it takes to affect appreciably the scale @xmath4 in a thermal bath whose temperature depends on this time @xmath117 according to conservation of comoving energy density . in @xcite
it was assumed that the rate for affecting the scale @xmath4 is lpm suppressed @xcite , giving @xmath118 .
a selfconsistent solution of these equations gives the aforementioned estimate of @xcite .
arnold and lenaghan @xcite noted that as the scattering rate in thermal plasma is @xmath119 , there can be no process that would make the estimate faster than @xmath120 .
we have examined the validity of these scaling estimates by plotting the difference of the energy density obtained from the simulation and the 1st order hydrodynamic estimate and find that both of these estimates describe the data poorly . however ,
if we assume that the approach is governed by the hydrodynamical relaxation time @xmath121 we get an estimate @xmath122 .
figure [ hydro_diff ] displays the deviation of the hydrodynamics as a function of rescaled time .
in particular for intermediate couplings @xmath123 the overlap of the different curves indicates correct scaling .
note that this estimate is parametrically the same as the estimate of arnold and lenaghan as parametrically @xmath124
. however , there are large corrections beyond the parametric estimate in @xmath125 due to which it is important to use the full numerical result instead of the simpler parametric estimate .
we however believe ( in the absence of plasma instabilities ) that the correct scaling at sufficiently small @xmath20 is that of @xcite .
this estimate is however based on a large scale separation of @xmath126 .
numerically this ratio is not very large even in our simulation with small @xmath127 , where the ratio is @xmath128 ( at @xmath129 ) as can be inferred from figure [ hydro ] .
therefore we believe that the scaling predicted by @xcite sets in only at significantly smaller couplings . in conclusion , we have shown how a far from equilibrium overoccupied configuration of gluons reaches hydrodynamical flow under longitudinal expansion in a weak coupling setting that is systematically improvable .
it has been demonstrated by chesler and yaffe within ads / cft framework that at large values of t hooft coupling , hydrodynamics is surprisingly robust even in the presence of large anisotropy @xcite .
the main deliverable of this paper is to show that this robustness is present also in the weak coupling picture extrapolated to intermediate couplings relevant for heavy - ion collisions .
the authors thank stefan floerchinger , liam keegan , tuomas lappi , guilherme milhano , guy moore , and urs wiedemann for useful discussions .
. is supported by the european research council grant hotlhc erc-2011-stg-279579 .
e. iancu , a. leonidov and l. mclerran , hep - ph/0202270 .
e. iancu and r. venugopalan , in * hwa , r.c .
quark gluon plasma * 249 - 3363 [ hep - ph/0303204 ] .
f. gelis , t. lappi and r. venugopalan , int .
j. mod .
e * 16 * ( 2007 ) 2595 [ arxiv:0708.0047 [ hep - ph ] ] .
f. gelis , e. iancu , j. jalilian - marian and r. venugopalan , ann . rev .
nucl . part .
* 60 * ( 2010 ) 463 [ arxiv:1002.0333 [ hep - ph ] ] .
t. epelbaum and f. gelis , phys . rev
. lett .
* 111 * ( 2013 ) 232301 [ arxiv:1307.2214 [ hep - ph ] ] . t. epelbaum and f. gelis , phys .
d * 88 * ( 2013 ) 085015 [ arxiv:1307.1765 ] .
j. berges , k. boguslavski , s. schlichting and r. venugopalan , phys .
d * 89 * ( 2014 ) 11 , 114007 [ arxiv:1311.3005 [ hep - ph ] ] .
j. berges , k. boguslavski , s. schlichting and r. venugopalan , phys .
d * 89 * ( 2014 ) 7 , 074011 [ arxiv:1303.5650 [ hep - ph ] ] .
j. berges , b. schenke , s. schlichting and r. venugopalan , nucl .
a * 931 * ( 2014 ) 348 [ arxiv:1409.1638 [ hep - ph ] ] .
d. teaney , j. lauret and e. v. shuryak , phys .
* 86 * ( 2001 ) 4783 ; p. huovinen , p. f. kolb , u. w. heinz , p. v. ruuskanen and s. a. voloshin , phys .
b * 503 * ( 2001 ) 58 ; p. f. kolb , u. w. heinz , p. huovinen , k. j. eskola and k. tuominen , nucl .
a * 696 * ( 2001 ) 197 ; t. hirano and k. tsuda , phys .
c * 66 * ( 2002 ) 054905 ; p. f. kolb and r. rapp , phys .
c * 67 * ( 2003 ) 044903 .
r. baier , p. romatschke and u. a. wiedemann , phys .
c * 73 * ( 2006 ) 064903 [ hep - ph/0602249 ] .
p. romatschke and u. romatschke , phys .
lett . * 99 * , 172301 ( 2007 ) ; m. luzum and p. romatschke , phys .
c * 78 * , 034915 ( 2008 ) [ erratum - ibid .
c * 79 * , 039903 ( 2009 ) ] ; k. dusling and d. teaney , phys .
c * 77 * , 034905 ( 2008 ) ; h. song and u. w. heinz , phys .
c * 77 * , 064901 ( 2008 ) .
r. baier , a. h. mueller , d. schiff and d. t. son , phys .
b * 502 * ( 2001 ) 51 [ hep - ph/0009237 ] .
p. b. arnold , g. d. moore and l. g. yaffe , jhep * 0301 * ( 2003 ) 030 [ hep - ph/0209353 ] .
a. h. mueller and d. t. son , phys .
b * 582 * ( 2014 ) 279 [ hep - ph/0212198 ] ; v. mathieu , a. h. mueller and d. n. triantafyllopoulos , eur .
j. c * 74 * ( 2014 ) 2873 [ arxiv:1403.1184 [ hep - ph ] ] .
m. c. abraao york , a. kurkela , e. lu and g. d. moore , phys .
d * 89 * ( 2014 ) 7 , 074036 [ arxiv:1401.3751 [ hep - ph ] ] .
a. kurkela and g. d. moore , phys .
d * 86 * ( 2012 ) 056008 [ arxiv:1207.1663 [ hep - ph ] ] . a. kurkela and e. lu , phys .
* 113 * ( 2014 ) 18 , 182301 [ arxiv:1405.6318 [ hep - ph ] ] .
s. mrowczynski , phys .
b * 214 * ( 1988 ) 587 [ phys . lett .
b * 656 * ( 2007 ) 273 ] ; s. mrowczynski , phys .
b * 314 * ( 1993 ) 118 ; s. mrowczynski and m. h. thoma , phys .
d * 62 * ( 2000 ) 036011 [ hep - ph/0001164 ] .
a. kurkela and g. d. moore , jhep * 1111 * ( 2011 ) 120 [ arxiv:1108.4684 [ hep - ph ] ] .
a. kurkela and g. d. moore , jhep * 1112 * ( 2011 ) 044 [ arxiv:1107.5050 [ hep - ph ] ] .
romatschke , phys .
c * 75 * ( 2007 ) 014901 [ hep - ph/0607327 ] .
p. romatschke and m. strickland , phys .
d * 68 * ( 2003 ) 036004 [ hep - ph/0304092 ] ; p. romatschke and m. strickland , phys .
d * 70 * ( 2004 ) 116006 [ hep - ph/0406188 ] .
b. arnold , g. d. moore and l. g. yaffe , phys .
d * 72 * ( 2005 ) 054003 [ hep - ph/0505212 ] .
d. bodeker and k. rummukainen , jhep * 0707 * ( 2007 ) 022 [ arxiv:0705.0180 [ hep - ph ] ] . a. h. mueller , phys .
b * 475 * ( 2000 ) 220 [ hep - ph/9909388 ] .
j. ghiglieri and g. d. moore , jhep * 1412 * ( 2014 ) 029 [ arxiv:1410.4203 [ hep - ph ] ] .
p. romatschke and r. venugopalan , phys .
* 96 * ( 2006 ) 062302 [ hep - ph/0510121 ] ; p. romatschke and r. venugopalan , eur .
phys . j. a * 29 * ( 2006 ) 71 [ hep - ph/0510292 ] ; p. romatschke and r. venugopalan , phys . rev .
d * 74 * ( 2006 ) 045011 [ hep - ph/0605045 ] .
h. iida , t. kunihiro , b. mueller , a. ohnishi , a. schaefer and t. t. takahashi , phys .
d * 88 * ( 2013 ) 094006 [ arxiv:1304.1807 [ hep - ph ] ] . a. h. mueller , nucl .
b * 572 * ( 2000 ) 227 [ hep - ph/9906322 ] .
y. v. kovchegov , nucl .
a * 692 * ( 2001 ) 557 [ hep - ph/0011252 ] .
r. baier , p. romatschke , d. t. son , a. o. starinets and m. a. stephanov , jhep * 0804 * ( 2008 ) 100 [ arxiv:0712.2451 [ hep - th ] ] .
b. arnold , g. d. moore and l. g. yaffe , jhep * 0305 * ( 2003 ) 051 [ hep - ph/0302165 ] .
m. a. york and g. d. moore , phys .
d * 79 * ( 2009 ) 054011 [ arxiv:0811.0729 [ hep - ph ] ] .
l. d. landau and i. pomeranchuk , dokl .
nauk ser .
* 92 * ( 1953 ) 535 .
a. b. migdal , phys .
* 103 * ( 1956 ) 1811 .
r. baier , y. l. dokshitzer , a. h. mueller , s. peigne and d. schiff , nucl . phys .
b * 484 * ( 1997 ) 265 [ hep - ph/9608322 ] .
p. b. arnold and j. lenaghan , phys .
d * 70 * ( 2004 ) 114007 [ hep - ph/0408052 ] .
p. m. chesler and l. g. yaffe , phys .
d * 82 * ( 2010 ) 026006 [ arxiv:0906.4426 [ hep - th ] ] ; p. m. chesler and l. g. yaffe , phys . rev .
* 106 * ( 2011 ) 021601 [ arxiv:1011.3562 [ hep - th ] ] ; p. m. chesler and l. g. yaffe , arxiv:1501.04644 [ hep - th ] . | we numerically solve 2 + 1d effective kinetic theory of weak coupling qcd under longitudinal expansion relevant for early stages of heavy - ion collisions .
we find agreement with viscous hydrodynamics and classical yang - mills simulations in the regimes where they are applicable . by choosing initial conditions that are motivated by color - glass - condensate framework
we find that for @xmath0gev and @xmath1 , the system is approximately described by viscous hydrodynamics well before @xmath2 fm / c . |
for a link @xmath1 in @xmath0 , a compact , unoriented ( possibly non - orientable or disconnected ) surface in @xmath0 with boundary @xmath1 is called a _ spanning surface _ of @xmath1 .
a spanning surface is called a _ seifert surface _ if it is orientable .
two spanning surfaces are _
@xmath2-equivalent _ if they are related by a finite sequence of the following two moves and ambient isotopies : + ( 1 ) attaching / deleting a tube as illustrated in figure [ equi ] ( 1 ) , + ( 2 ) attaching / deleting a half twisted band locally as illustrated in figure [ equi ] ( 2 ) .
+ in particular , two seifert surfaces are _
@xmath3-equivalent _ if they are related by a finite sequence of ambient isotopies and attaching / deleting a tube that preserves orientability of surfaces .
it is well known that all seifert surfaces of a link are _
@xcite , @xcite .
this result is generalized to unoriented spanning surfaces @xcite .
( gordon and litherland @xcite)[main ] all spanning surfaces of a link are @xmath2-equivalent .
in @xcite , bar - natan , fulman and kauffman gave an elementary proof of the result that all seifert surfaces of a link are @xmath3-equivalent .
here we will give an elementary proof of theorem [ main ] .
this note is inspired by the article @xcite .
for a diagram @xmath4 in the @xmath5-sphere @xmath6 of a link @xmath1 , we color the regions of @xmath7 with black and white like a checkerboard .
then we obtain two spanning surfaces of @xmath1 that correspond to the black regions and white regions of @xmath7 .
we call these two spanning surfaces the _ checkerboard surfaces _ for @xmath4 . from now on , we will not care the color of these surfaces and describe both surfaces by shading .
the local changes , which are naturally induced by reidemeister moves i , ii and iii , for checkerboard surfaces as illustrated in figure [ r - move ] are called @xmath8-moves of type i , ii and iii respectivery .
-moves ] [ lemma : btow ] all checkerboard surfaces for all diagrams of a link are related by a finite sequences of @xmath8-moves and ambient isotopies .
it is obvious that for two link diagrams @xmath4 and @xmath9 of a given link , a checkerboad surface for @xmath9 is related to one of two checkerboad surfaces for @xmath4 . hence it is enough to show that the two checkerboard surfaces for @xmath4 are related by a finite sequences of @xmath8-moves and ambient isotopies .
figure [ btow ] shows that the two checkerboard surfaces for the diagram @xmath4 are related by a finite sequence of @xmath8-moves and ambient isotopy .
is a simple closed curve in @xmath6 that separates @xmath4 into one trivial arc and the other . ]
[ lemma : rbys ] each @xmath8-move for checkerboard surfaces can be realized by @xmath2-equivalence .
we start with a easy but important claim mentioned in @xcite . while they consider only orientable surfaces , the claim holds for non - orientable surfaces
. _ claim . _ the two local changes for spanning surfaces of a link as illustrated in figure [ observation ] can be realized by attaching / deleting a tube . in the right - side surface . ]
one of @xmath8-moves of type i ( resp .
ii ) is related by ambient isotopy and the other @xmath8-move of type i ( resp .
ii ) is realized by attaching / deleting a half twisted band ( resp .
a tube as the local change of figure [ observation ] ( 2 ) ) . for @xmath8-move of type iii ,
figure [ deform ] completes the proof .
here we use attaching / deleting a half twisted band and the local change of figure [ observation ] ( 1 ) .
-move of type iii given by @xmath2-equivalence ] by combining proposition [ lemma : btow ] and lemma [ lemma : rbys ] , we have the following proposition .
[ checckerboad ] all checkerboard surfaces for all diagrams of a link are @xmath2-equivalent .
the following proposition , together with proposition [ checckerboad ] , completes the proof of theorem [ main ] .
any spanning surface of a link is @xmath2-equivalent to a checkerboard surface for a diagram of the link .
a spaning surface @xmath10 can be assumed as a surface which is one disk with several bands attached since @xmath10 is a compact surface with boundary .
for example , see figure [ disk - band ] .
moreover , we may assume that @xmath10 has a diagram whose singular parts appear in only bands as half twists of bands or crossings between bands as illustrated in figure [ band ] .
applying local changes of figure [ observation ] ( 1 ) to all crossings between bands , we have a checkerboard surface .
the author is partially supported by a grant - in - aid for scientific research ( c ) ( @xmath1123540074 ) of the japan society for the promotion of science . | gordon and litherland showed that all compact , unoriented , possibly non - orientable surfaces in @xmath0 bounded by a link are realted by attaching / deleting tubes and half twisted bands . in this note
we give an elementary proof for this result . |
the hot phase ( @xmath0 k ) of the interstellar medium ( ism ) is generated primarily by the mechanical power imparted by stellar winds from massive stars and supernovae ( sne ) .
the clustered nature of massive star formation necessarily dictates that some stellar winds and core - collapse sne will occur in low - density cavities excavated by other winds or sne , thus creating superbubbles .
theoretical work by @xcite demonstrated that superbubbles could be considered more powerful versions of stellar wind - blown bubbles , which already had a well - developed theoretical model @xcite .
the most detailed of these papers was that of @xcite , which has become the de - facto standard model explaining the structure and physics of wind - blown bubbles and superbubbles .
superbubbles play important roles in generating and maintaining the multi - phase ism in star - forming galaxies ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the influence of superbubbles can extend beyond the thin gaseous disk , as some superbubbles vent their hot , metal - enriched plasma into the galactic halo , creating galactic fountains @xcite in normal spiral galaxies like the milky way and feeding galactic superwinds in starburst galaxies @xcite . as the hot , collisionally ionized plasma in superbubbles emits primarily by x - ray line emission and thermal bremsstrahlung
, x - ray observations provide a crucial window on these energetic phenomena .
however both x - ray and optical observational studies indicate that our current understanding of superbubble physics and energetics is incomplete . the earliest imaging x - ray observations of star - forming complexes in the magellanic clouds with the _ einstein observatory _ revealed that some superbubbles were indeed detected as x - ray sources @xcite . surprisingly , the detected superbubbles typically had x - ray luminosities an order of magnitude greater than predicted from the standard weaver et al . model ( we refer to these as x - ray - bright superbubbles ) .
these x - ray - bright superbubbles demonstrate that the processes responsible for x - ray emission in superbubbles are not yet fully understood .
not all superbubbles are over - luminous , however .
other superbubbles in the large magellanic cloud ( lmc ) are as faint or fainter than expected based on the weaver et al .
model ( @xcite ; referred to as x - ray - dim superbubbles ) .
subsequent observations using the _ rntgensatellit _ ( _ rosat _ ) and _ asca _ x - ray observatories ( see e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) have confirmed these results .
solutions to the x - ray - bright superbubble problem fall into two categories ; ( 1 ) replacing the standard weaver et al .
model at some or all epochs of superbubble growth , and ( 2 ) supplementing and extending the standard weaver et al . model
. a popular and elegant example of the first approach was pioneered by @xcite , who proposed that the impact of the blast waves from any sne that occurred near the superbubble shell can shock - heat the shell to x - ray - emitting temperatures . using 1-d models employing the kompaneets thin shell approximation @xcite they demonstrated that this shock - heating leads to a large , but _ temporary _ , increase in the total x - ray luminosity from the superbubble .
evidence exists for supernova remnant ( snr ) shocks within superbubbles , and snr shell impacts are also the most plausible cause of the unusually high shell expansion velocities seen in some superbubbles @xcite .
nevertheless , their precise contribution to the x - ray luminosity is uncertain .
the most - promising example of a solution of the second kind is that of @xcite . the standard weaver et al .
wind - blown bubble model takes no account of metal enrichment once sne occur ; the metal abundances in the hot plasma are assumed to be the same value at all times . for a collisionally ionized plasma in the temperature range @xmath1
@xmath2 k the x - ray emissivity is roughly proportional to the metal abundance of the plasma , as line and recombination dominate over pure thermal bremsstrahlung . @xcite point out that if the plasma is enriched in abundance by a factor of 10 or more by sn activity , then a similar enhancement in the soft x - ray luminosity should be expected . in this hypothesis
the structure of the hot x - ray - emitting superbubble interior remains that given by the weaver et al .
model at all times , but its chemical enrichment must be taken into account self - consistently .
this hypothesis remains to be tested as yet there is no observational evidence for high metal enrichment in x - ray - bright superbubbles . constraining the extent to which these , or any other , mechanisms contribute to the total x - ray emission of superbubbles requires higher quality x - ray data than was available with the earlier generations of x - ray observatories ( _ einstein _ , _ rosat _ , and _ asca _ ) .
potential discriminants between the different models hinge on spatial variations in the x - ray spectral properties of the plasma within superbubbles , which requires observations with an instrument combining high spatial resolution , moderate to high spectral resolution and high sensitivity .
both modern x - ray observatories , the _ chandra x - ray observatory _ and the _ xmm - newton _ , fulfill these requirements .
we have obtained and analyzed _ chandra _ observations of two x - ray - bright lmc superbubbles , with the advanced ccd imaging spectrometer ( acis ) instrument .
the target objects are dem l50 and dem l152 from the@xcite h@xmath3 catalog of lmc hii regions .
dem l152 is also part of the nebula cataloged as n44 by @xcite .
other properties of these superbubbles are highly constrained with existing multi - wavelength observations , properties such as the host stellar population , shell expansion characteristics , local interstellar gas density and intrinsic absorption .
we adopt a distance of 50 kpc to the lmc @xcite . at this distance
an angular offset of @xmath4 corresponds to a physical size of 0.24 pc .
dem l50 has not previously been the target of pointed x - ray observations .
optical and hi observations are reported in @xcite .
it is a roughly elliptical bubble with major and minor axes of @xmath5 and @xmath6 respectively , associated with the superimposed snr n186 d on its northern edge .
it is unclear whether the snr is physically connected to the superbubble or whether this is a chance projection of separate objects along a line of sight .
otherwise dem l50 is relatively isolated from other hii nebulae and young stellar clusters .
in contrast dem l152 is part of the large and relatively crowded n44 complex , a set of shells and hii regions covering @xmath7 of sky @xcite . given the limited @xmath8 field of view of individual _ chandra _ acis ccd chips we choose to concentrate on the x - ray - bright features shell 1 , the south bar , and shell 3 that were identified in earlier _ rosat _ and _ asca _ observations @xcite . by exploiting _
@xmath9 spatial resolution we separate the true superbubble x - ray emission from non - related diffuse x - ray emission in the immediate vicinity of each superbubble and investigate the degree of spatial variations in soft x - ray properties within these bubbles ( see [ sec : spectra ] ) . if the diffuse x - ray emission in these bubbles is very highly structured , arising in small - scale regions associated with the optical shells , then these _ chandra _ observation have the spatial resolution to detect this . in [ sec : models ] we compare our results to the predictions of the standard @xcite model and to 1-d hydrodynamical simulations including sn - shell impacts and discuss superbubble energy budgets in light of our observations .
we discuss the metallicity - enhancement scenario and the possibility of mass - loading from clouds in light of our observations in
[ sec : discussion ] .
dem l152 was observed with the _ chandra _ acis - s instrument on 2002 september 22 for @xmath10 ks ( observation identification number [ obsid ] 3356 ) , and dem l50 was observed for @xmath11 ks on 2003 january 01 ( obsid 3355 , see table .
[ tab : obslog ] ) .
the observations were made to accommodate the majority of the optical nebula of each bubble on the more sensitive back - illuminated acis s3 ccd chip , but parts of each nebula spill onto the adjacent s2 and s4 ccd chips .
llllrll dem l50 & @xmath12 & @xmath13 & 2003 - 01 - 01 & acis - s & 3355 & 37011.9 ( s3 ) , 36720.9 ( s2 ) , 36714.5 ( s4 ) + dem l152 & @xmath14 & @xmath15 & 2002 - 09 - 22 & acis - s & 3356 & 18713.6 ( s3 ) , 18713.6 ( s2 ) , 18713.6 ( s4 ) + data reduction and analysis were performed using the _ chandra _ ciao software package ( version 3.4 ) with the associated calibration database ( caldb version 3.3 ) .
in addition heasoft ( version 6.1.1 ) was used for some tasks , including spectral fitting with xspec ( version 12.5.1 ) .
each raw data set was reprocessed to take account of the latest calibration of important effects such as time - dependent ccd gain , charge transfer inefficiency and contamination of the optical - blocking filter .
periods during the observation that experienced higher - than - normal levels of solar x - ray and particle events ( commonly referred to as background flares ) were removed using the iterative @xmath16 clipping method described in @xcite .
the effective on - source exposure time remaining in each acis ccd chip after this procedure is given in table [ tab : obslog ] .
only a small fraction , @xmath17% , of the total exposure time was lost to background flares in either observation .
this is much lower than the typical @xmath10% fraction of acis observation time affected by flares .
we used the wavelet - based source detection algorithm wavdetect to search for point - like objects in each x - ray observation .
each chip was treated separately , searching for sources in images created in the soft x - ray 0.3 2.0 kev energy - band , hard x - ray 2.0 8.0 kev energy band and total 0.3 8.0 kev energy band . only sources with signal - to - noise ratios
@xmath18 were accepted as point sources in the initial iteration of the data analysis . in a few cases , low s
/ n features that most probably are genuine point sources were missed by the source - detection algorithm .
these sources became apparent only after all brighter point sources had been removed .
we assessed each such feature , and removed those that appeared to be point - like based on our personal scientific judgment .
the point sources are shown marked on soft ( @xmath19 2 kev ) and hard ( @xmath20 8 kev ) energy band images of dem l50 and dem l152 in fig.[fig : point_sources ] .
we screened out the events from detected point sources in the data used for both imaging and spectral analysis of the superbubbles to a radius equivalent to 4 gaussian @xmath21 , based on a fit to the radius of the _ chandra _ psf as a function of off - axis angle .
the holes left in any image by source removal were filled in using the ciao task dmfilth .
the observed distribution of pixel values in a background annulus of thickness 6 acis pixels ( @xmath22 ) around the point source defines the probability density function from which random values were chosen to fill in the source region ( the poisson option in dmfilth ) .
care was taken to ensure that the background annulus chosen did not contain other point sources which would bias the interpolation . for spectral analysis
, sources detected in any of the three bands ( the soft , hard and total bands described above ) were excluded when creating diffuse - emission - only spectra .
the soft band source lists were used for any image of the diffuse emission created within the energy range 0.3 2.8 kev .
the hard and total band source lists were used for any image created within the 2.8 8.0 kev energy band .
we discuss the data reduction of the images in [ sec : images : data ] and the data reduction of the spectra in
[ sec : spectra : data ] .
for the purposes of background subtraction we used the blank - sky datasets provided as part of the _ chandra _ caldb , as the diffuse soft x - ray emission from both dem l50 and dem l152 covers a large fraction of the s3 chip in each observation .
these blank - sky data sets comprise hundreds of kiloseconds of data per ccd chip , providing a high s / n estimate of the typical background experienced by the acis detectors . since the net background experienced in the acis detector does change with time , the blank sky data was re - normalized so that the mean x - ray surface brightness ( excluding point sources ) for each chip in the @xmath23
@xmath24 kev energy band matched that in each observation .
typical variations in the hard x - ray background in the acis instrument are @xmath25% ( root mean square ) from observation to observation @xcite .
we used a hard x - ray band to calculate the renormalization factor for the blank - sky data sets as soft diffuse x - ray emission is present in both the s2 and s3 chips in the two observations ( see fig .
[ fig : point_sources ] ) .
the majority of the superbubble emission falls within the field of view of the s3 chip in both observations .
part of dem l50 spills over onto the s2 chip , however , and n44 shell 3 ( a snr ; see @xcite ) is a strong soft x - ray source in the s2 chip of the dem l152 observation .
excess soft x - ray emission is not apparent on the s4 chip in either observation .
note that apparently diffuse hard x - ray emission in a superbubble has only been detected in 30 doradus c @xcite .
our method of re - normalizing the blank - sky data sets means that there will be no significant diffuse x - ray emission in the @xmath23 6 kev energy band , once point sources and the background have been subtracted .
this approach is valid , in that a visual inspection of the raw , non - background - subtracted data reveals no evidence for hard x - ray emission within dem l50 or dem 152 , and the mean source - free hard x - ray surface brightness in each chip of our observations is less than or equal to that in the blank - sky data .
we can not rule out the presence of diffuse hard x - ray emission in our observations of dem l50 and dem l152 , but if such emission is present , it is very faint . for dem l50 ,
the s3 chip needed essentially no renormalization , as the observed source - free @xmath26 kev surface brightness differed from the blank - sky data sets by only @xmath27% .
for the front illuminated s2 and s4 chips the blank sky backgrounds were normalized downward by 5.9% ( uncertainty 2.1% ) and 14.9% ( uncertainty 2.0% ) respectively .
it is not surprising that the hard background surface brightness in the blank - sky data might be higher than in our observations of dem l50 and dem l152 .
the blank sky data sets are constructed from a large number of data sets and thus from data experiencing a higher ( but more normal ) fraction of mild background flares than our observations . for the dem l152 observation the blank sky background was normalized down by 4.4% ( uncertainty @xmath28% ) .
for the s2 and s4 chips the blank sky backgrounds were normalized downward by 2.0% ( uncertainty 3.0% ) and 16.6% ( uncertainty 2.8% ) respectively .
the magnitude of the difference in the mean source - free x - ray surface brightness in the @xmath23 6 kev energy band for the s4 chip in both observations is somewhat surprising .
this ccd chip is known to suffer from significant flaws in its serial readout that leads to artificial events being registered ( visible as streaks in images ) that can only be partially screened - out in software processing .
we speculate that this effect is normally exacerbated by the unwanted particle events common in background flares , thus accounting for the pronounced difference between the blank sky data for the s4 chip and our observations .
we altered the coordinate system of the blank - sky data to match the point of the appropriate observation .
thus our background subtraction for image analysis consisted of subtracting an image created from the blank - sky data from an image from the real observational data , scaled by the ratio of total exposure times in the observation to the blank - sky data for the appropriate chip .
in figures [ fig : img_deml50 ] and [ fig : img_deml152 ] we present adaptively smoothed x - ray images of dem l50 and dem l152 in a variety of soft x - ray energy bands , along with optical h@xmath3+[nii ] imaging and hi column density maps .
[ fig : rgb ] presents three - color composite images of the superbubbles that combine narrow - band optical h@xmath3+[nii ] , [ oiii ] and soft x - ray emission .
adaptive smoothing using the ciao tool csmooth @xcite attempts to differentially smooth an input image to achieve a relatively uniform local s / n ratio , so that bright regions are smoothed with a smaller smoothing kernel than low surface brightness regions .
we initially smoothed the diffuse image in the @xmath19 2.8 kev energy band , after point source removal but prior to background subtraction , with a target s / n ratio of 3 but with a maximum smoothing kernel equivalent to a gaussian of fwhm@xmath29 . the smoothing map calculated in this step
was then applied to all the background - subtracted soft x - ray diffuse images , so that any differences between images in different energy bands are intrinsic and are not due to the application of different smoothing maps . given claims that csmooth can generate spurious structure in adaptively smoothed images @xcite
, we also investigated adaptively smoothed images using the completely independent adaptive smoothing algorithm implemented in the _ xmm - newton _ data analysis software sas as asmooth .
this algorithm generated images very similar to those produced by csmooth , except that the stated statistical significance of individual features within the images differed significantly between the two methods .
we also generated images using uniform smoothing with a 2-dimensional gaussian mask of fwhm@xmath30 ( not shown ) in order to conservatively verify the existence of features seen in the adaptively smoothed images .
the diffuse soft x - ray emission in dem l50 is clearly confined within the optical superbubble and snr shells ( fig .
[ fig : img_deml50]b ) .
the apparent x - ray surface brightness of the snr n186 d @xcite on the northern edge of the roughly - elliptical superbubble is twice that of the larger superbubble .
if we ignore n186 d and concentrate solely on the superbubble we see that the detected x - ray emission in the @xmath19 2.8 kev energy band is relatively uniformly distributed , although with a slight tendency toward being brighter on the south and eastern interior edges of the optical shell .
the hi column density along the line of sight to dem l50 peaks in the projected center of the bubble , however ( see fig .
[ fig : img_deml50]c ) , raising the possibility that the weak limb brightening is due to central absorption rather than true enhancement of the x - ray emission at the shell walls . that at least some of the hi lies between us and the x - ray emitting plasma is clear from the way the x - ray - bright southeast limb of the superbubble fills in the gap in hi column density at that location ( compare fig . [
fig : img_deml50]a and c ) . based on the observed velocities of the hi gas , @xcite
show that the hi associated with dem l50 is most likely the outer layer of the expanding shell of the bubble .
dem l50 is most prominent in the @xmath31 1.6 kev band .
the snr n186 d is prominent in both this energy band and the softer @xmath19 0.6 kev energy band , despite being associated with a column density of @xmath32 of hi .
this suggests that the snr may be spectrally distinct from the superbubble .
neither the superbubble nor the snr are apparent in the @xmath33 2.8 kev energy band .
in contrast to the relative simplicity of dem l50 , the spatial structure of dem l152 ( n44 ) in the x - ray , optical and hi is complex ( see fig .
[ fig : rgb ] ) .
the field of view of the _ chandra _ observations ( fig .
[ fig : img_deml152 ] ) cover the following features identified in the _ rosat _ pspc observations of @xcite : shell 1 , the main superbubble , and parts of shell 3 , a snr , and the south bar , in particular the suspected break - out region where shell 1 may be venting its contents through a break in its southern edge .
the relationship between the new _ chandra _ observation and the features identified in prior x - ray observations is shown in fig .
[ fig : hri_n44 ] , where the _ chandra _ acis field of view and contours of diffuse x - ray surface brightness are overlaid on a smoothed _ rosat _
hri image of n44 .
the image was taken from the 108.5 ks hri observation rh600913 , obtained from the heasarc data archive .
shell 1 is the most prominent x - ray source in the @xmath19 2.8 kev energy band _
image , and the soft diffuse x - ray emission is particularly bright around the inner edge of the optical shell .
this limb - brightening is not an artifact of adaptive smoothing , as it is also clearly visible in images smoothed with a uniform fwhm@xmath30 gaussian mask .
the limb - to - center brightness contrast is a factor of @xmath25 in the @xmath19 2.8 kev and @xmath31 1.6 kev energy band images and is slightly larger in the softer @xmath19 0.6 kev energy band image ( fig .
[ fig : img_deml152]d ) .
this central decrement in x - ray surface brightness can not be due to absorption by intervening gas as both the x - ray - bright limbs and the fainter center of shell 1 lie within a region of lower hi column density ( fig .
[ fig : img_deml152]c ) .
shell 3 falls partly within the acis s3 and s2 chips , although its eastern edge is not covered in the current observations .
it has a comparable x - ray surface brightness to shell 1 , even though the hi column density along this line of sight ( @xmath34 ) is approximately twice the value toward the center of shell 1 .
bright , diffuse , soft x - ray emission also extends to both the west and south of shell 1 ; the latter feature being known as the south bar @xcite .
it is also noteworthy that both shell 1 and shell 3 are clearly detected in the @xmath33 2.8 kev energy band , in contrast to the complete lack of emission in this energy band from dem l50 and n186 d. the south bar lacks appreciable emission in this energy band , which is consistent with the finding of @xcite that the south bar was spectrally softer than shell 1 .
llrlll dem l50 & bubble & 46.9443 & @xmath35 & @xmath36 & @xmath37 + dem l50 & snr & 9.8197 & @xmath38 & @xmath39 & @xmath40 + dem l152 & bubble & 13.2994 & @xmath41 & @xmath42 & @xmath43 + dem l152 & snr & 4.9459 & @xmath44 & @xmath44 & @xmath45 + dem l152 & south bar & 11.8618 & @xmath46 & @xmath47 & @xmath48 + dem l152 & west & 17.4595 & @xmath49 & @xmath50 & @xmath51 + as a first step to selecting regions for more detailed spectral analysis we constructed three - color images of the two bubbles where the resultant color is representative of the mean energy of the diffuse x - ray emission ( fig .
[ fig : rgb]b & e ) within the soft x - ray band . in these panels
red represents the lowest energy photons ( @xmath19 0.6 kev ) , green intermediate energy photons ( @xmath31 1.6 kev ) and blue relatively harder emission in the @xmath33 2.8 kev band .
as we shall later show , these figures somewhat exaggerate the spectral differences between the two objects and the degree of small - scale spectral variation within each object .
nevertheless they do serve some purpose by more clearly illustrating some of the possible spatial variation in the distribution of diffuse emission at different energies and help inform our choice of regions over which to study the spectral properties of the diffuse emission . in fig .
[ fig : rgb]b the emission from the superbubble dem l50 appears slightly spectrally harder than the brighter emission from the snr n186 d , but there is only a marginal suggestion of spectral hardness variation within dem l50 itself . based on this and the optical morphology of the bubble ,
we choose to separate the x - ray data into two regions for spectral fitting : an elliptical region encompassing snr n186 d and another elliptical region covering the superbubble but excluding the snr .
these are the regions shown in fig .
[ fig : img_deml50]a . in order to examine spectral variations on smaller scales
, we further subdivided dem l50 into a center and limb region when performing spectral fits ( see [ sec : spectra : data ] ) .
note that only the parts of those ellipses within the s3 chip are actually included within the spectral extraction regions .
the soft x - ray emission in the vicinity of dem l152 shows more evident spectral variation than does dem l50 ( figure [ fig : rgb]e ) .
based on this and the optical morphology of the n44 nebula we chose spectral extraction regions similar to those used earlier by @xcite and @xcite : shell 1 ( the main bubble ) , shell 3 ( the snr ) , and the south bar , with most of the remaining emission associated with a new region we simply name west ( see fig .
[ fig : img_deml152]a ) .
diffuse x - ray count rates within the spectral extraction regions of both sets of observations are given in table [ tab : rates ] .
column 3 gives the area of each region .
columns 4 - 6 show the background - subtracted acis s3 count rates in the @xmath52 kev energy band within each specified region , where @xmath53 gives the count rate for the full x - ray emission , @xmath54 gives the estimated diffuse emission count rate , and @xmath55 gives the total point source count rate .
emission from point - like x - ray sources accounts for @xmath56% of the total soft x - ray emission within the boundaries of the dem l50 bubble , and less than @xmath57% of the emission within the boundary of the dem l152 shell 1 .
given _ chandra _ s high spatial resolution and sensitivity we should be able to determine whether the spectral properties of the diffuse emission vary on spatial scales smaller than the entire shell or bubbles . in figures [ fig : rgb]b & e we show energy - color - coded images of the diffuse emission around dem l50 and dem l152 .
such images can be prone to smoothing - related artifacts , so we also mapped the hardness ratios in @xmath58 pixels .
these large pixels were required in order to minimize the statistical uncertainty in the hardness ratio in each pixel . the spectral hardness ratio , which will be discussed in more detail in [ sec : spectra : data ] , is defined as @xmath59 , where @xmath60 and @xmath61 are the count rates in the specified hard ( @xmath62 kev ) and soft ( @xmath63 kev ) energy bands respectively .
thus higher values of @xmath64 correspond to a harder spectrum .
images of the hardness ratio itself tend to be difficult to interpret , as they can be dominated by noise in the lower signal - to - noise regions .
for this reason we constructed maps of the statistical significance of the deviation of the local hardness ratio from the average hardness ratio @xmath65 over each bubble . for a given pixel with hardness ratio @xmath66 and uncertainty @xmath67 ,
figures [ fig : rgb]c & f show @xmath68 .
in other words , the value in each pixel corresponds to the significance of the deviation in gaussian @xmath21 . while figures [ fig : rgb]c & f hint at some of the intrinsic large - scale variations across the superbubbles , variations in column density and the resultant absorption of soft x - ray photons have a substantial effect on the observed hardness ratios . figures [ fig : rgb]c & f should only be used to identify regions that appear spectrally distinct from one another .
although some spatial variation in the spectral properties of the diffuse x - ray emission is present in both objects , it is not particularly strong , and emission in the @xmath31 1.6 kev energy band dominates both sets of observations .
hardness ratio variations are present on large scales , for example between the bubbles and the nearby supernova remnant , or between shell 1 and the southern bar region of dem l152 .
on smaller angular scales ( @xmath69 ) we find that any spatial variations in spectral hardness are of marginal statistical significance in our current x - ray data . for the spectral fits
, we removed background flares from both the s3 and s2 chips with the lc_clean routine in ciao .
we then used the re - normalized acis stowed background files to subtract the approximately constant particle background from both chips .
to re - normalize the background files , we compared the relative count rates of the stowed background and our observations between 8 kev and 9.4 kev and re - normalized the particle background to match the count rate of our data .
the @xmath70 kev energy range was chosen to exclude real x - ray photon events and a line feature above 9.4 kev .
we used the specextract script in ciao to extract spectra from the specified regions and subtract the particle background . to account for the x - ray background ,
we fit the extracted spectra from the s2 chip , following @xcite and @xcite , and used the resulting fits as a component of our subsequent models to the s3 data .
a thermal plasma in collisional ionization equilibrium fits the local x - ray background , while two additional thermal plasmas in equilibrium are used for the galactic background , and the cumulative emission from extragalactic sources is fit by a power law .
our best fits to the s2 background are shown in table [ tab : fits2 ] .
the power law index was fixed to the value found by @xcite , and the column densities were obtained from heasarc from @xcite .
temperatures were left as a free parameter .
the local hot bubble temperature is given in column 3 , the galactic hi column density is shown in column 4 , the temperatures of the two thermal components of the galactic background are given in columns 5 and 6 , and the extragalactic power law index is shown in column 7 .
normalizations were re - scaled to account for the relative difference in area between the s3 extracted regions and the s2 chip .
for the sake of comparison , we include two fits to the s2 background for dem l50 ; the low s2 " fit was a local minimum in @xmath71 before the global best fit was found .
the effect of varying the background fit can be seen by comparing the dem l50 models with the high s2 " and low s2 " background fits .
all extracted regions are shown in figure [ fig : ext_regions ] .
in addition to the regions in figs . [
fig : img_deml50]a and [ fig : img_deml152]a , we extracted spectra for smaller subregions .
we divided dem l50 into limb and center regions a separate spectrum for the bright southern portion of the limb .
for dem l152 , we extracted additional spectra for the bubble ( that is , shell 1 ) limb , west limb , and west blowout regions as well as for the two knots shown in fig .
[ fig : ext_regions ] . while some clumps seen in the smoothed image of dem l152 are artifacts of the smoothing algorithm , these two knots are statistically significant ( see fig . [
fig : rgb]f ) and appear when different smoothing algorithms are used .
we fit each superbubble region and subregion with one- and two - temperature plasmas in collisional ionization equilibrium , using the xspec astrophysical plasma emission code ( apec ) models @xcite with variable abundances ( tables [ tab : fit50 ] and [ tab : fit152 ] ) .
we examined an additional , non - equilibrium ionization model for dem l50 ( table [ tab : fit50ne ] ) . in all models ,
he , c , n , al , and ni abundances were fixed to the solar values .
we varied the @xmath3 elements o , ne , mg , si , s , ar , and ca in proportion to one another , and fe was left as a free parameter . although absolute abundances are not well constrained , the @xmath3/fe ratio is more reliable . to ensure that the fits were not local minima , we looked at possible values of , temperature , and abundances spanning one to two orders of magnitude with the xspec command steppar . as a further check on the validity of our fits and to establish error bars for each variable , we ran the error command on the free parameters . to determine fluxes , we ran the xspec command flux on each model and determined the absorption - corrected fluxes by setting to zero .
we adopted the distance of 50 kpc to the lmc when converting absorption - corrected fluxes to luminosities . observed fluxes and the calculated luminosities
are shown in tables [ tab : fluxobs50]-[tab : lumin152 ] , which we discuss below .
we also calculated hardness ratios and median energies for the diffuse emission from each region , shown in table [ tab : hardness ] ; the hardness ratio @xmath59 , where @xmath60 and @xmath61 are the luminosities in the hard ( @xmath31 1.6 kev ) energy band and soft ( @xmath19 0.6 kev ) energy band .
absorption by intervening hi results in an apparently harder spectrum , and figs .
[ fig : img_deml50]c & [ fig : img_deml152]c show that there is significant variation in the hi column density across these objects .
the column density was a free parameter in our fits , and the resulting luminosities and hardness ratios account for variations between regions .
@xmath64 obtained using the count rate ( i.e. not absorption - corrected ) is shown for comparison in column 4 of table [ tab : hardness ] .
the influence of the column density can be seen by comparing the luminosity - determined and count rate - determined hardness ratios for the snr 0523 - 679 @xcite in dem l152 . while the snr appears to have the hardest spectrum from the count rates , its higher absorption column density masks a much softer spectrum .
@xcite demonstrate that the median energy is a more reliable indicator of x - ray spectral variations than the hardness ratio is . as a result
, we include the median energy of each region in table [ tab : hardness ] .
the median energies generally agree with the hardness ratios .
an exception is the dem l152 snr , which has a higher median energy due to the hi absorption discussed above .
lllllll dem l50 & high s2 fit & 1.08 & 0.132 & @xmath72 & @xmath73 & @xmath74 + dem l50 & low s2 fit & 0.08 & 0.132 & @xmath75 & @xmath76 & @xmath74 + dem l152 & l152 s2 fit & 0.08 & 0.228 & @xmath77 & @xmath78 & @xmath74 + lllllllllllll & * high s2 * & @xmath79 & @xmath80 & * 3.5e-03 * & @xmath81 & @xmath81 & * 0.15 * & @xmath82 & @xmath83 & @xmath84 & * 1.743 * & * 211 * + center 1 t solar & high s2 & 0.31 @xmath85 0.05 & 0.19 @xmath85 0.01 & 1.3e-03 & @xmath86 & @xmath86 & 0.19 & 1.00 & 1.00 & 1.00 & 1.943 & 213 + center 2 t & high s2 & @xmath87 & @xmath88 & 4.7e-03 & 0.17 @xmath85 0.03 & 1.6e-03 & 0.13 & @xmath89 & @xmath90 & @xmath91 & 1.757 & 209 + center 2 t solar & high s2 & @xmath92 & @xmath93 & 3.1e-03 & @xmath94 & 5.9e-04 & 0.13 & 1.00 & 1.00 & 1.00 & 1.835 & 211 + limb 1 t & high s2 & 0.29 @xmath85 0.05 & @xmath95 & 2.7e-03 & @xmath86 & @xmath86 & 0.18 & @xmath96 & @xmath97 & @xmath98 & 1.744 & 302 + * limb 2 t * & * high s2 * & @xmath99 & @xmath100 & * 1.4e-02 * & @xmath101 & * 7.3e-04 * & * 0.09 * & @xmath102 & @xmath103 & @xmath104 & * 1.718 * & * 300 * + limb 2 t solar & high s2 & 0.21 @xmath85 0.04 & @xmath93 & 1.4e-03 & @xmath105 & 2.1e-04 & 0.13 & 1.00 & 1.00 & 1.00 & 1.749 & 302 + limb 1 t & low s2 & 0.36 @xmath85 0.06 & @xmath106 & 2.6e-03 & @xmath86 & @xmath86 & 0.18 & @xmath107 & @xmath108 & @xmath109 & 1.793 & 302 + limb 2 t & low s2 & @xmath110 & @xmath111 & 1.4e-03 & @xmath112 & 7.1e-04 & 0.16 & @xmath113 & @xmath114 & @xmath115 & 1.799 & 300 + * bright limb 1 t * & * high s2 * & @xmath116 & @xmath117 & * 1.9e-03 * & @xmath81 & @xmath81 & * 0.19 * & @xmath118 & @xmath119 & @xmath120 & * 1.516 * & * 137 * + bright limb 2 t & high s2 & 0.37 @xmath85 0.07 & @xmath121 & 1.8e-03 & @xmath122 & 8.3e-04 & 0.14 & @xmath123 & @xmath124 & @xmath125 & 1.532 & 135 + * snr 2 t * & * high s2 * & @xmath126 & @xmath127 & * 4.7e-04 * & @xmath128 & * 2.6e-05 * & * 0.22 * & @xmath129 & @xmath130 & @xmath131&*2.114 * & * 145 * + snr 2 t solar & high s2 & @xmath132 & @xmath133 & 6.2e-03 & 0.29 @xmath85 0.02 & 2.8e-04 & 0.09 & 1.00 & 1.00 & 1.00 & 2.173 & 147 + snr 2 t & low s2 & @xmath134 & @xmath135 & 4.3e-04 & 0.73 @xmath85 0.09 & 3.2e-05 & 0.23 & @xmath136 & @xmath137 & @xmath138 & 2.150 & 145 + snr 2 t solar & low s2 & @xmath139 & @xmath133 & 9.9e-03 & @xmath140 & 3.9e-04 & 0.09 & 1.00 & 1.00 & 1.00 & 2.255 & 148 + llllllllllll center nei & high s2 & @xmath141 & @xmath142 & @xmath143e-04 & 5.57e10 @xmath85 4.68e10 & 22,000 & @xmath144 & @xmath145 & @xmath146 & 1.838 & 210 + limb nei & high s2 & @xmath147 & @xmath148 & @xmath149e-04 & 3.31e10 @xmath85 2.84e10 & 46,000 & @xmath150 & @xmath151 & @xmath152 & 1.718 & 301 + limb nei & low s2 & @xmath153 & @xmath154 & @xmath155e-04 & 6.26e10 @xmath85 4.69e10 & 65,000 & @xmath156 & @xmath157 & @xmath158 & 1.718 & 301 + bright limb nei & high s2 & @xmath159 & @xmath160 & @xmath161e-04 & 8.33e10 @xmath85 2.88e10 & 53,000 & @xmath162 & @xmath163 & @xmath164 & 1.543 & 136 + snr nei & high s2 & @xmath165 & @xmath166 & @xmath167e-04 & 2.60e10 @xmath85 6.39e9 & 11,000 & @xmath168 & @xmath169 & @xmath170 & 2.265 & 146 + llllllllllll & @xmath171&@xmath172 & * 2.5e-03 * & @xmath173 & @xmath173 & * 0.37 * & @xmath174&@xmath175 & @xmath176 & * 1.360 * & * 151 * + bubble 2 t & @xmath177 & @xmath178 & 5.5e-04 & @xmath179 & 2.2e-03 & 0.33 & @xmath180 & @xmath181 & @xmath182 & 1.352 & 149 + * bubble limb 1 t * & @xmath183 & @xmath184 & * 3.3e-03 * & @xmath81 & @xmath81 & * 0.33 * & @xmath185 & @xmath186 & @xmath187 & * 1.794 * & * 138 * + bubble limb 2 t fixed & 0.28 & @xmath188 & 4.7e-04 & @xmath189 & 1.7e-03 & 0.30 & @xmath190 & @xmath191 & @xmath192 & 1.782 & 137 + west 1 t & @xmath193 & 0.30 @xmath85 0.02 & 3.6e-03 & @xmath86 & @xmath86 & 0.30 & @xmath194 & @xmath195 & @xmath196 & 1.088 & 154 + * west 2 t * & @xmath197 & @xmath198&*2.5e-03 * & @xmath199 & * 2.9e-04 * & * 0.31 * & @xmath200 & @xmath201 & @xmath202 & * 1.027 * & * 152 * + west limb 1 t & @xmath203 & @xmath204 & 7.3e-04 & @xmath86 & @xmath86 & 0.32 & @xmath205 & @xmath206 & @xmath207 & 1.387 & 78 + west limb 2 t fixed & 0.28 & @xmath208 & 6.6e-04 & @xmath209 & 5.7e-05 & 0.26 & @xmath210 & @xmath211 & @xmath212 & 1.355 & 77 + * west blowout 1 t fixed * & * 0.28 * & @xmath213 & * 2.0e-04 * & @xmath81 & @xmath214 & * 0.25 * & @xmath215 & @xmath216 & @xmath217 & * 1.143 * & * 77 * + west blowout 2 t fixed & 0.28 & @xmath218 & 2.5e-05 & @xmath219 & 3.7e-05 & 0.22 & 10.10 & @xmath220 & @xmath221 & 1.156 & 75 + south bar 1 t & 0.23 @xmath85 0.02 & @xmath222 & 1.0e-03 & @xmath86 & @xmath86 & 0.35 & @xmath223 & @xmath224 & @xmath225 & 1.054 & 128 + south bar 2 t & 0.21@xmath85 0.02 & @xmath226 & 4.4e-05 & @xmath227 & 1.5e-05 & 0.43 & @xmath228 & 6.84 & @xmath229 & 1.034 & 126 + * snr 1 t * & @xmath230 & @xmath231 & * 9.3e-04 * & @xmath81 & @xmath81 & * 0.38 * & @xmath232 & @xmath233 & @xmath234 & * 1.176 * & * 67 * + snr 2 t & @xmath235 & @xmath236 & 7.4e-04 & 0.86 & 6.3e-05 & 0.39 & @xmath237 & @xmath238 & @xmath239 & 1.198 & 65 + knot 1 1 t solar & 0.22 & 0.30 @xmath85 0.03 & 8.2e-05 & @xmath86 & @xmath86 & 0.30 & 1.00 & 1.00 & 1.00 & 1.471 & 17 + knot 1 1 t free abun & 0.22 & 0.30 @xmath85 0.03 & 7.3e-05 & @xmath86 & @xmath86 & 0.30 & 1.40@xmath240 & @xmath241 & @xmath242 & 1.253 & 15 + knot 2 1 t solar & 0.22 & @xmath243 & 3.5e-05 & @xmath86 & @xmath86 & 0.30 & 1.00 & 1.00 & 1.00 & 1.288 & 8 + llllll center 1 t & 0.24 & 0.23 & 0.28 & 0.26 & 0.032 + center 1 t solar & 0.24 & 0.22 & 0.27 & 0.25 & 0.032 + center 2 t & 0.24 & 0.23 & 0.28 & 0.26 & 0.031 + center 2 t solar & 0.25 & 0.23 & 0.28 & 0.26 & 0.032 + * center nei * & * 0.25 * & * 0.22 * & * 0.28 * & * 0.26 * & * 0.032 * + limb 1 t & 0.39 & 0.36 & 0.45 & 0.42 & 0.057 + limb 2 t & 0.41 & 0.36 & 0.47 & 0.42 & 0.057 + limb 2 t solar & 0.43 & 0.36 & 0.48 & 0.42 & 0.058 + limb 1 t @xmath244 & 0.48 & 0.36 & 0.54 & 0.43 & 0.063 + limb 2 t @xmath244 & 0.55 & 0.36 & 0.55 & 0.43 & 0.064 + * limb nei * & * 0.41 * & * 0.36 * & * 0.46 * & * 0.42 * & * 0.057 * + limb nei @xmath244 & 0.48 & 0.36 & 0.55 & 0.43 & 0.064 + bright limb 1 t & 0.16 & 0.15 & 0.17 & 0.17 & 0.015 + bright limb 2 t & 0.16 & 0.15 & 0.18 & 0.17 & 0.015 + * bright limb nei * & * 0.16 * & * 0.15 * & * 0.18 * & * 0.17 * & * 0.015 * + snr 2 t & 0.37 & 0.26 & 0.39 & 0.28 & 0.021 + snr 2 t solar & 0.34 & 0.26 & 0.36 & 0.28 & 0.020 + snr 2 t @xmath244 & 0.36 & 0.26 & 0.39 & 0.29 & 0.023 + snr 2 t solar @xmath244 & 0.35 & 0.26 & 0.37 & 0.28 & 0.022 + * snr nei * & * 0.35 * & * 0.26 * & * 0.37 * & * 0.28 * & * 0.021 * + lllllll center 1 t & 15.69 & 8.43 & 15.79 & 8.53 & 0.098 & 2.70 + center 1 t solar & 9.41 & 4.93 & 9.51 & 5.03 & 0.099 & @xmath245 + center 2 t & 21.02 & 10.51 & 21.12 & 10.61 & 0.098 & @xmath246 + center 2 t solar & 12.41 & 5.47 & 12.51 & 5.57 & 0.098 & @xmath247 + * center nei * & * 10.25 * & * 4.95 * & * 10.35 * & * 5.05 * & * 0.099 * & * 1.77 * + limb 1 t & 14.42 & 6.67 & 14.60 & 6.85 & 0.18 & @xmath248 + limb 2 t & 23.81 & 8.25 & 23.98 & 8.42 & 0.18 & 2.31 + limb 2 t solar & 10.95 & 3.99 & 11.13 & 4.16 & 0.18 & @xmath249 + limb 1 t @xmath244 & 13.72 & 9.60 & 13.92 & 9.80 & 0.20 & @xmath250 + limb 2 t @xmath244 & 12.54 & 8.73 & 12.74 & 8.93 & 0.20 & @xmath251 + * limb nei * & * 12.54 * & * 5.35 * & * 12.72 * & * 5.53 * & * 0.18 * & * 1.22 * + limb nei @xmath244 & 12.15 & 8.48 & 12.35 & 8.68 & 0.20 & @xmath252 + bright limb 1 t & 7.72 & 4.50 & 7.77 & 4.54 & 0.047 & 2.81 + bright limb 2 t & 8.50 & 4.33 & 8.54 & 4.38 & 0.047 & @xmath253 + * bright limb nei * & * 5.83 * & * 3.43 * & * 5.87 * & * 3.48 * & * 0.048 * & * 2.12 * + snr 2 t & 3.45 & 1.35 & 3.51 & 1.41 & 0.063 & 1.02 + snr 2 t solar & 8.69 & 2.73 & 8.75 & 2.79 & 0.060 & @xmath254 + snr 2 t @xmath244 & 1.70 & 1.11 & 1.77 & 1.18 & 0.070 & @xmath255 + snr 2 t solar @xmath244 & 10.83 & 3.81 & 10.90 & 3.88 & 0.067 & @xmath256 + * snr nei * & * 3.67 * & * 1.47 * & * 3.73 * & * 1.54 * & * 0.064 * & * 1.08 * + totals of 1 t models & 30.11 & 15.11 & 30.39 & 15.38 & 0.27 & @xmath257 + totals of 2 t models & 44.83 & 18.76 & 45.10 & 19.03 & 0.28 & @xmath258 + totals of nei models & 22.79 & 10.30 & 23.07 & 10.58 & 0.28 & 1.42 + llllll & * 0.93 * & * 0.87 * & * 0.95 * & * 0.89 * & * 0.016 * + bubble 2 t & 0.93 & 0.88 & 0.94 & 0.89 & 0.017 + * bubble limb 1 t * & * 0.77 * & * 0.72 * & * 0.78 * & * 0.73 * & * 0.0096 * + bubble limb 2 t fixed & 0.77 & 0.73 & 0.78 & 0.74 & 0.011 + west 1 t & 0.73 & 0.67 & 0.73 & 0.67 & 0.0061 + * west 2 t * & * 0.73 * & * 0.68 * & * 0.75 * & * 0.70 * & * 0.023 * + west limb 1 t & 0.26 & 0.23 & 0.26 & 0.23 & 0.0020 + west limb 2 t fixed & 0.25 & 0.24 & 0.26 & 0.24 & 0.0061 + * west blowout 1 t fixed * & * 0.19 * & * 0.18 * & * 0.20 * & * 0.18 * & * 0.00070 * + west blowout 2 t fixed & 0.20 & 0.18 & 0.20 & 0.18 & 0.00077 + south bar 1 t & 0.53 & 0.48 & 0.54 & 0.48 & 0.0053 + south bar 2 t & 0.54 & 0.48 & 0.56 & 0.49 & 0.016 + * snr 1 t * & * 0.15 * & * 0.14 * & * 0.16 * & * 0.15 * & * 0.0042 * + snr 2 t & 0.15 & 0.15 & 0.16 & 0.15 & 0.0076 + knot 1 1 t solar & 0.049 & 0.046 & 0.049 & 0.046 & 0.00019 + knot 1 1 t free abun & 0.051 & 0.048 & 0.052 & 0.048 & 0.00036 + knot 2 1 t & 0.024 & 0.022 & 0.024 & 0.022 & 0.00012 + lllllll & * 20.20 * & * 9.10 * & * 20.26 * & * 9.16 * & * 0.054 * & * 4.23 * + bubble 2 t & 22.22 & 10.58 & 22.27 & 10.64 & 0.056 & @xmath259 + * bubble limb 1 t * & * 17.88 * & * 8.61 * & * 17.91 * & * 8.64 * & * 0.032 * & * 4.31 * + bubble limb 2 t fixed & 17.66 & 8.54 & 17.69 & 8.58 & 0.035 & @xmath260 + west 1 t & 21.59 & 8.50 & 21.61 & 8.52 & 0.020 & @xmath261 + * west 2 t * & * 22.19 * & * 9.12 * & * 22.27 * & * 9.20 * & * 0.075 * & * 3.55 * + west limb 1 t & 6.23 & 2.42 & 6.24 & 2.43 & 0.0066 & @xmath262 + west limb 2 t fixed & 7.31 & 3.24 & 7.33 & 3.26 & 0.020 & @xmath263 + * west blowout 1 t fixed * & * 7.22 * & * 2.70 * & * 7.22 * & * 2.71 * & * 0.0023 * & * 3.07 * + west blowout 2 t fixed & 7.26 & 2.79 & 7.27 & 2.79 & 0.0025 & @xmath264 + south bar 1 t & 12.81 & 4.72 & 12.82 & 4.73 & 0.017 & @xmath265 + south bar 2 t & 11.99 & 4.29 & 12.04 & 4.34 & 0.050 & @xmath266 + * snr 1 t * & * 5.72 * & * 2.12 * & * 5.73 * & * 2.14 * & * 0.014 * & * 3.22 * + snr 2 t & 5.72 & 2.16 & 5.75 & 2.18 & 0.025 & @xmath267 + knot 1 1 t solar & 0.81 & 0.45 & 0.81 & 0.45 & 0.0006 & @xmath268 + knot 1 1 t free abun & 0.82 & 0.46 & 0.82 & 0.46 & 0.0012 & @xmath269 + knot 2 1 t & 0.43 & 0.21 & 0.43 & 0.21 & 0.0004 & @xmath270 + total of 1 t models & 54.60 & 22.32 & 54.69 & 22.41 & 0.091 & @xmath271 + total of 2 t models & 56.40 & 24.03 & 56.58 & 24.21 & 0.18 & @xmath268 + llllll dem l50 center & -0.67 & -0.72 & -0.63 & ... & 0.723 + dem l50 limb & -0.60 & -0.73 & -0.63 & ... & 0.723 + dem l50 bright limb & -0.48 & -0.55 & -0.43 & ... & 0.767 + dem l50 center + limb & -0.63 & -0.72 & -0.63 & 0.46 @xmath85 0.02 & 0.723 + dem l50 snr & ... & -0.56 & -0.54 & 0.32 @xmath85 0.02 & 0.694 + dem l152 bubble & -0.34 & -0.35 & ... & 0.75 @xmath85 0.02 & 0.869 + dem l152 bubble limb & -0.31 & -0.33 & ... & ... & 0.854 + dem l152 west & -0.50 & -0.51 & ... & 0.67 @xmath85 0.02 & 0.825 + dem l152 west limb & -0.48 & -0.49 & ... & ... & 0.810 + dem l152 west blowout & -0.58 & -0.59 & ... & ... & 0.759 + dem l152 south bar & -0.50 & -0.53 & ... & 0.63 @xmath85 0.03 & 0.825 + dem l152 snr & -0.51 & -0.50 & ... & 0.80 @xmath85 0.07 & 0.942 + modeled fits to dem l50 are shown in tables [ tab : fit50 ] and [ tab : fit50ne ] . the fits in table [ tab : fit50 ]
model the emission as one- and two - temperature thermal plasmas .
the best fit in column 3 is consistent with the observations of the hi in dem l50 by @xcite .
the first temperature component , @xmath272 , appears in column 4 , with its normalization in column 5 , and the second component , @xmath273 , appears in column 6 , with its normalization in column 7 .
column 8 shows the effective kt " , defined as the average @xmath274 of the region weighted by the normalization .
the effective kt " thus shows the dominant temperature component in the region .
columns 9 and 10 give the abundances relative to solar , with the @xmath3/fe ratio in column 11 .
columns 12 and 13 show the reduced @xmath71 and the number of degrees of freedom of the fit .
the statistically best fits , determined by running ftest in xspec , are shown in bold .
figure [ fig:50spectra ] shows a sample fit to the center region of dem l50 . from the best fits for the limb and center regions in table [ tab : fit50 ]
, the center s temperature distribution appears more homogeneous than the limb s , and the two - temperature best - fit to the limb shows a higher temperature component that may correspond to the bright limb subregion .
this correlation would be consistent with shock - heating of the southern part of the limb by an expanding snr , while the rest of the limb remains at a cooler temperature . in each region of dem l50 ,
spectral fits consistently gave low @xmath3/fe ratios ( see table [ tab : fit50 ] ) , with typical values of about 0.3 compared to an average lmc value of 0.6 @xcite .
these low ratios are especially surprising , given that we might have expected a high @xmath3/fe ratio due to an enhancement in @xmath3 elements from core - collapse sne .
the bubble has likely experienced @xmath25 sne @xcite , so an @xmath3 enhancement would be natural . raising
the @xmath3/fe ratio by forcing the abundances to solar values resulted in the models labeled ` solar ' in table [ tab : fit50 ] . in every case
, ftest showed that the free abundance fits were statistically better than the solar values .
the ftest command calculates a _
p_-value , which here denotes the probability that solar - abundance material could randomly produce a spectrum fit by the more complicated , free abundance fit .
the resulting _
p_-values ranged from 0.049 to as low as @xmath275 , showing that the solar fits have a very low probability of being correct .
although the solar abundances are statistically excluded , there may be systematic problems in forward - fitting low - resolution x - ray spectra , which could give us misleading estimates of the abundances and their uncertainties . using the mekal or raymond - smith models instead of apec
did not affect the @xmath3/fe ratios .
another possibility is that collisional ionization equilibrium is not a valid assumption for dem l50 . to test this scenario
, we performed the non - equilibrium ionization fits shown in table [ tab : fit50ne ] .
temperature is shown in column 4 , the normalization is in column 5 , abundances relative to solar are in columns 8 and 9 , and the reduced @xmath71 and number of degrees of freedom are in columns 11 and 12 , respectively .
the @xmath3/fe ratios found with these new fits are all consistent with solar values .
a sample non - equilibrium ionization fit is shown in figure [ fig:50spectra_nei ] for the dem l50 center region . to see whether or not non - ionization equilibrium is a plausible scenario , we consider the ionization timescale .
the ionization timescale in seconds is found by dividing @xmath276 , the density - weighted ionization timescale in s @xmath277 ( column 6 ) , by the electron density , @xmath278 . to find the electron density , we used @xmath279 where @xmath280 is the normalization of the fit , in units of @xmath281 and @xmath282 is the distance to the superbubble , assumed to be 50 kpc .
we also assumed a spherical geometry , uniform density distribution , and @xmath28310% he abundance ( @xmath284 ; see , e.g. , @xcite ) .
the resulting ionization timescale estimates are shown in column 7 of table [ tab : fit50ne ] and are all on the order of @xmath285 years , about 10% of the superbubble s age . while our observation of the bubble in this stage would be a chance occurrence to some extent , this timescale is not unrealistically brief and a recent supernova is also consistent with many of the observed properties of the bubble as we will discuss later .
furthermore , the hypothesized brightening from snr impacts should enhance the detectability of bubbles with recent sne , increasing the likelihood of observing a bubble that is not in equilibrium .
if the low @xmath3/fe abundances are indeed real , they may reflect an fe enrichment in the ism near dem l50 .
dem l50 is located in a large , 1.4 kpc diameter void in the ism @xcite ; a single type ia supernova could not noticeably enrich this entire region , nor could it enrich the estimated @xmath286 of material swept up by dem l50 @xcite . on the other hand ,
if the density is low enough , the interior of dem l50 could be enriched by a single type ia event , and snr n186 d near the bubble would be a likely candidate . the @xmath3/fe ratio of the snr , however , does not seem noticeably different from the rest of the bubble ( see [ sec : snrmodels ] ) and may in fact be less fe - enriched than the center region .
there is no clear spectral contrast between the snr and regions dominated by the shocked ism like that seen in the type ia snr dem l71 @xcite , nor are there any noticeable fe , s , or si features such as those seen in type ia snr n103b @xcite .
furthermore , snr n186d s proximity to young stars and star - forming regions make a core - collapse sn origin more likely ( e.g. * ? ? ?
dust destruction from a supernova shock wave can not explain the low @xmath3/fe ratio .
although dust destruction would increase the fe abundance , the destruction of silicates would concomitantly raise the o abundance ( compigne , m. , private communication ; see @xcite 2003 for o and fe abundances in dust ) .
we note that dem l50 does have an uncommon infrared morphology ( slater et al .
2010 , in prep . ) ; its 8 @xmath287 m emission appears to fill the bubble s h@xmath3 shell , whereas many other lmc superbubbles only show polycyclic aromatic hydrocarbon ( pah ) emission external to the bubble .
this morphology is likely explained by pah emission on the interface between the ionized and the foreground neutral layer , which is seen in projection toward the center of the bubble ( see fig . [
fig : img_deml50 ] and [ sec : images : deml50 ] ) .
an unusual dust environment does not appear to be the cause of the low observed @xmath3/fe ratio .
similarly odd fe abundances have been observed in other objects .
dem l316 consists of a pair of interacting snrs ; one snr has a high fe abundance while the other does not @xcite .
as mentioned previously , the lmc @xmath3/fe ratio is low in general , @xmath2830.6 @xcite ; dem l50 may simply lie in a region of the ism with a lower than average @xmath3/fe ratio .
of course , the @xmath3/fe ratio we observe could also result from inadequate x - ray models or spectra , rather than genuine properties of the emitting material . in x - ray spectral models , parameters such as temperature and column density are degenerate with the metallicity , making the model difficult to constrain @xcite .
a variety of x - ray models may fit the same spectrum equally well .
for instance , other authors have found that the need for unusual abundances in a spectral fit may disappear when a more complicated , multi - component x - ray spectral model is used instead ( e.g. * ? ? ? * ; * ? ? ? * ) .
the @xmath2830.12 kev resolution of _ chandra _ acis spectra and model degeneracies make it unclear which precise combination of models should be used , however .
determining the origin of the unusual abundances in dem l50 requires further investigation of both the lmc ism and the x - ray models used , and the explanation of the observed abundances may have to wait for higher resolution spectra .
our current data suggest , however , that the non - equilibrium ionization scenario is the most plausible explanation for dem l50 s low observed @xmath3/fe ratio .
luminosities for all regions and models are shown in table [ tab : lumin50 ] , with the non - equilibrium ionization fits shown in bold .
the limb contributes roughly 50% of the total emission .
the bright limb subregion provides 13 - 41% of the total .
hardness ratios and median energies appear in table [ tab : hardness ] and indicate that the bright limb is clearly spectrally harder than the limb as a whole , while the limb and center have comparable hardness ratios .
this relative spectral hardness suggests that the bright limb was recently heated .
for the two - temperature equilibrium fit to the limb , the limb appears to have a softer hardness ratio than the center .
the two - temperature fit has a slightly higher absorbing column density and an additional lower temperature component , which account for the difference .
the temperatures of the non - equilibrium fits are less precise , and the larger error bars make the temperature differences between the regions unclear .
the relative hardness of the bright limb is still apparent when the non - equilibrium fits are considered , however , and is confirmed by the median energies of the regions . as was noted in
[ sec : images : deml50 ] , absorption due to intervening hi may be responsible for the apparent limb - brightening observed in dem l50 . when the areas of the extracted regions are considered , the center does appear to have a @xmath28320 - 45% higher surface brightness than the limb as a whole ( table [ tab : lumin50 ] ) .
the bright limb subregion , however , has a surface brightness comparable to or greater than the center .
the hi column density has a significant effect on the calculated luminosity , however , and as can be seen in table [ tab : lumin50 ] , the luminosity varies substantially depending on the model .
the error bars in our models correspond to typical uncertainties in the x - ray luminosity of about 25% , but the uncertainty is up to 50% for the bright limb .
when galactic is included , the contours observed by @xcite , shown in figure [ fig : img_deml50 ] , do agree with the modeled , given the error bars in the fits , variations within the regions , and uncertainties in the observations .
nevertheless , we can not conclusively determine whether the apparent brightness of the bright limb is genuine or due to line - of - sight absorption .
the temperature and hardness of the bright limb subregion support the idea that an off - center snr heated the shell wall in that region , while the center s brightness is most likely a result of mass - loading due to conductive evaporation from the shell walls .
mass - loading from swept - up clouds is unlikely given dem l50 s location in a void in the ism and the lack of observed clouds in x - ray or other wavelengths ( see fig .
[ fig : rgb ] ) .
we discuss this further in
[ sec : discussion : massloading ] .
metallicity enhancement is also not observed ; while fe is enhanced , the @xmath3 elements , which should have the greatest effect on the cooling function , are not . given the contribution of the limb to the total luminosity and the lack of observed clumps in the ism near dem l50 ( see fig . [
fig : rgb ] ) , an off - center snr and thermal conduction from the shell walls into the center are a more viable explanation for dem l50 s brightness than ablation or evaporation of ism cloudlets . like dem l50 ,
dem l152 is an x - ray - bright superbubble , although as seen in figures [ fig : img_deml50 ] and [ fig : img_deml152 ] , its hi environment and x - ray and optical morphology differ substantially from that of dem l50 . consequently , dem l152 s derived x - ray properties also generally differ from dem l50 ; the characteristics the two objects have in common may help resolve the question of why some superbubbles become x - ray - bright .
we modeled the emission from dem l152 with one- and two - temperature thermal plasmas .
table [ tab : fit152 ] reports the resulting fits ; the column headings are the same as in table [ tab : fit50 ] , except that the background fit , which is the same for all fits , is not listed . in the case of the west limb and south bar ,
it is not clear which fit is statistically better , so neither fit is shown in bold .
figure [ fig:152spectra ] shows the best fit to the bubble region as an example .
dem l152 is a few @xmath288 k hotter than dem l50 on average , a result of the higher number of o stars and supernovae in dem l152 @xcite .
the bubble limb appears slightly cooler than the bubble region as a whole , which is reasonable since it makes up the outer edge of the bubble and likely consists of denser , cooler material .
the west region shows a 0.25 kev component as well as an extremely hot 0.79 kev component .
this hot component appears in the west limb region two - temperature fit , but not the west blowout , and may also account for a substantial portion ( @xmath28325% ) of the emission from the south bar .
the localization of this hot component to the adjacent west limb and south bar regions suggests recent heating in this area , perhaps from a supernova explosion .
an expanding snr would also explain the limb brightening seen in the west region .
interestingly , @xcite also suggest the presence of an off - center supernova remnant in the southwest due to the x - ray brightness in the south bar , and @xcite find high - velocity gas in the region .
as expected , the two blowout regions are slightly cooler than the bubble region .
most of the south bar s emission is fit by a 0.29 kev component , while the hot component mentioned above accounts for the remainder .
likewise , a 0.25 kev plasma characterizes the west blowout .
the blowouts are cooler than the 0.37 kev bubble region and cooler than ( or at least comparable to ) the 0.33 kev bubble limb region , presumably having cooled due to adiabatic expansion .
abundances for each region show enrichment from core - collapse supernovae ( see table [ tab : fit152 ] ) , which raises the @xmath3/fe ratio .
we discuss the implications of this enrichment for the expected luminosity in
[ sec : discussion : metallicity ] .
two clumps of x - ray emitting material , a few parsecs in diameter , appear in the south bar region ( see fig . [
fig : ext_regions ] ) .
we extracted and fit spectra for these two knots ( see table [ tab : fit152 ] ) , fixing the column density to the best fit value for the south bar region .
due to low signal - to - noise , we only used single temperature fits and fixed abundances to solar values for the second knot .
the resulting luminosities should be viewed as estimates only . nevertheless , the luminosities are one to two orders of magnitude lower than the luminosity of the other regions and clearly do not have much effect on the total emission .
we plot the median energy versus total count rate for the knots and regions of dem l152 in fig .
[ fig : knotplot ] .
the knots do not exhibit any detectable spectral difference relative to the rest of the superbubble .
the luminosities for each region and subregion of dem l152 are shown in table [ tab : lumin152 ] ; the error bars correspond to an uncertainty of @xmath28315% in the calculated luminosities .
the bubble region provides @xmath283 37% of the total x - ray luminosity ( 33% from the bubble limb , 4% from diffuse emission ) , the west region contributes @xmath283 41% ( 14% from the west limb , 13% from the west blowout , and 14% from diffuse emission ) , and the south bar provides @xmath283
21% ( 19% from diffuse emission , 2% from the knots ) .
the intrinsic surface brightnesses are unclear , as they depend on the model and column density assumed .
for example , the best fits for the bubble region imply some slight limb - brightening , while the two - temperature fits indicate the opposite . if the limb - brightening observed in fig .
[ fig : rgb ] is not intrinsic , it must instead result from higher absorption towards the bubble center .
this hypothetical absorption would conflict with the hi contours shown in fig .
[ fig : img_deml152 ] , however , which show a lower column density towards the center of the bubble .
thus , we expect that the observed limb - brightening in the bubble region is real .
a comparison of the hardness ratios and median energies in table [ tab : hardness ] shows that the south bar and west blowout regions are softer than the rest of the superbubble , consistent with a blowout origin and with the previous x - ray observations of @xcite and @xcite .
this variation in hardness can also be seen in figure [ fig : rgb]e .
the bubble and bubble limb regions are noticeably harder than the other regions of dem l152 . in general
, we find softer hardness ratios for dem l152 than previous authors ( e.g. * ? ? ?
* ) , a result of fitting the column density instead of assuming a fixed conversion between count rate and luminosity . as with dem l50 , limb regions account for roughly half of the observed x - ray emission .
the south bar and west limb regions alone provide a third of the total emission and may be associated with a recent off - center snr as discussed above .
in contrast , the two knots make up only 2% of the total x - ray emission , suggesting that emission from clumps can not explain dem l152 s x - ray luminosity .
the snr n186 d appears at the northern edge of dem l50 .
the fact that n186 d is detected in [ oiii ] while dem l50 is not suggests they may be two physically distinct objects @xcite ; similar hi kinematics in both objects , however , may indicate that the snr is associated with the larger bubble @xcite , and thus the exact relationship between the snr and the superbubble remains unclear .
snr n186 d s x - ray spectra can be fit by a non - equilibrium model or a two - temperature equilibrium model . with free abundances
, the equilibrium fits show a 0.19 kev component and a weaker 0.72 kev component .
these temperatures are typical of snrs in the lmc ( see * ? ? ?
* ) and in our own galaxy ( e.g. * ? ? ?
* ; * ? ? ?
if abundances are forced to solar values , however , the two temperature components are 0.08 kev and 0.29 kev .
the dominant 0.08 kev component is a lower temperature than any of the lmc remnants studied by @xcite and even the 0.29 kev component is on the lower end of her sample .
the free abundance fit is statistically better than this solar fit , with a _
p_-value of 0.05 , and shows the same low @xmath3/fe ratio seen in the rest of dem l50 .
although we can not rule out the possibility that this snr is a type ia remnant , the consistency of the @xmath3/fe ratio throughout dem l50 suggests the abundances are associated with the ism environment of dem l50 or alternatively , that the snr is not in ionization equilibrium . as discussed in
[ sec : deml50_abun ] , the non - equilibrium scenario is the more likely explanation .
a non - equilibrium ionization fit gives an ionizing timescale of 11,000 years , a temperature of 0.58 kev , and an @xmath3/fe ratio of 0.64 , close to the typical lmc value of @xmath2830.6 @xcite .
the non - equilibrium ionization model gives a total x - ray luminosity in the @xmath289 kev band of @xmath290 .
the luminosity varies by an order of magnitude among the potential models , however , from @xmath291 to @xmath292 , a result of the higher in the models with solar abundances .
these possible luminosities are all within the range of @xmath293-@xmath294 found by @xcite for snrs in the lmc and by @xcite for m33 snrs .
the median energy for snr n186 d , shown in table [ tab : hardness ] , is softer than its hardness ratio would suggest .
the hi appears patchy in this region of dem l50 , and the observations by @xcite show a substantial gradient in the hi column density near the snr .
these variations may be affecting our modeled temperatures in this region .
dem l152 s snr , 0523 - 679 @xcite , consists of a 0.38 kev plasma , close to the median temperature value of williams sample of lmc snrs @xcite .
the snr shows a high @xmath3/fe ratio characteristic of core - collapse sne .
the extremely high and unconstrained value of the snr s @xmath3/fe ratio is a result of the lower signal - to - noise in this small region of the bubble .
the total x - ray luminosity is @xmath295 , typical of other snrs ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
the snr does appear to differ spectrally from the rest of the bubble , as seen in fig .
[ fig : knotplot ] .
we modeled the central temperature and x - ray luminosity in the [ 0.3 - 2.0 kev ] band for dem l50 and dem l152 using the similarity solutions of @xcite as implemented by @xcite .
energy and mass input were not assumed to be constant ; time - dependent energy input from stellar winds and sne was calculated using the mass loss rates from @xcite and the stellar population shown in table [ tab : stelpop ] ( see @xcite for details ) . for dem l50 , we assumed an ambient density of 1.4 as explained in @xcite , a metallicity of 0.4 and an age of 5 myr , at which point 2 sne are expected to have occurred if a @xcite initial mass function ( imf ) is assumed @xcite .
we modeled dem l152 using a density of 2.5 @xcite , a metallicity of 0.4 and an age of 6 myr , incorporating the effects of 4 sne as suggested in @xcite s ( 1995 ) analysis of dem l152 s mass function .
the resulting x - ray luminosities and central temperatures are shown in table [ tab : models ] and compared with the observed values .
the observed x - ray luminosities are an order of magnitude higher than the standard model predicts .
varying the input ambient density or metallicity did increase the x - ray luminosity , but even increasing these parameters by an order of magnitude still resulted in an x - ray luminosity lower than the observed value .
the radii calculated with the weaver model likewise do not match the observed values ; both bubbles have predicted radii of more than 100 pc ( 130.6 pc for dem l50 , 146.6 pc for dem l152 ) , whereas the observed radii are around 50 pc . this
growth - rate discrepancy " has been observed in bubbles with a range of sizes , from bubbles around individual wolf - rayet stars to the superbubbles we consider here ( e.g. @xcite ; see review by @xcite ) .
one possible explanation for the smaller observed radii is that excess radiative cooling has caused the bubble to lose more energy than predicted . to see
if the high x - ray luminosities of our objects provide enough cooling to slow the bubble s growth , we use the formula for the bubble radius given by @xcite : @xmath296 where @xmath297 is the ambient density in @xmath277 , @xmath298 is the mechanical luminosity in @xmath299 , and @xmath300 is time in @xmath288 years . assuming that supernovae dominate the mechanical energy input , we can estimate the mechanical luminosity as the energy provided by sne divided by the age of the superbubble .
two sne and an age of 5 myr ( dem l50 ) or four sne and an age of 6 myr ( dem l152 ) yield a mechanical luminosity of order @xmath294 .
the x - ray radiative losses for both bubbles , on the other hand , are on the order of @xmath299 .
although our superbubbles are x - ray - bright , the observed x - ray luminosity is therefore insufficient to account for the small radius observed in both bubbles .
in contrast to the weaver model , the one - dimensional hydrodynamic model of dem l50 from @xcite does correctly fit the radius and velocity of the bubble .
the mechanical input is the same as before , but this model assumes a higher ambient pressure of @xmath301 @xmath277 k , neglects thermal conduction at the shell walls , and includes the effect of a snr impact on the shell . while the higher ambient pressure results in the correct radius , the x - ray luminosity predicted by this model is far too low , an order of magnitude lower than the weaver model , which was itself lower than the observed luminosity .
this low luminosity suggests that thermal conduction does play a role in dem l50 and is not strongly suppressed by magnetic fields .
if thermal conduction is occurring , the bubble temperature should be lowest at the outer edge , where dense , cold evaporated material is present , and should increase toward the bubble center .
the equilibrium fits to the dem l50 limb and center regions do show this temperature contrast ; the effective temperature of the limb is only 0.09 kev , while the value increases to 0.15 kev for the center region .
the non - equilibrium fits , on the other hand , suggest that the limb is slightly hotter than the center .
the possibility of a cooler limb is within the error bars , however , and the contribution of the heated bright limb region may raise the average temperature of the limb . the x - ray luminosity at different times is shown in table [ tab : hydro ] and compared with the weaver model luminosity for dem l50 and the observed luminosity .
a supernova goes off at @xmath302=5.26 myr , and the snr increases the bubble s luminosity by several orders of magnitude .
the emissivity profiles of the bubble before and after the snr hits the shell walls are shown in figure [ fig : modelprofile ] .
the snr impact enhances the emissivity of the bubble s limb and also increases the shell velocity from 1 to 23 in agreement with the observed value .
although the luminosities of this model are unrealistic , the response of the luminosity and shell velocity to the snr impact suggests that snr impacts may be important in our superbubbles . neither the weaver model nor the 1-d hydrodynamic model is able to match the observed luminosities of the superbubbles .
the hydrodynamic model suggests , however , that mass - loading via thermal evaporation from the shell walls must play a role in these bubbles .
mass - loading from the shell walls and heating from snr impacts are consistent with our observations and could account for most of the superbubbles x - ray emission .
llllllll dem l50 & 1 & 1 & 3 & 1 & 8 & 6 & 9 + dem l152 & 1 & 3 & 4 & 11 & ... & 12 & ... + lllllll dem l50 weaver & 1.4 & 5 myr & 1.31 @xmath303 & 0.27 kev & @xmath304 & 0.15 kev or 0.28 kev + dem l152 weaver & 2.5 & 6 myr & 5.05 @xmath303 & 0.32 kev & @xmath305 & 0.37 kev + llllllll 0.3 - 2.0 kev & @xmath306 & @xmath307 & @xmath308 & @xmath309 & @xmath310 & @xmath311 & @xmath312 + 2.0 - 8.0 kev & @xmath313 & @xmath314 & @xmath315 & @xmath316 & @xmath317 & ... & @xmath318 +
@xcite suggests that since emission lines of @xmath3 elements dominate the soft x - ray emission , metallicity enhancement from sne could explain the anomalously high x - ray luminosities observed in many superbubbles . as discussed earlier (
[ sec : deml50models]-[sec : deml152models ] ) , dem l152 shows an @xmath3 enhancement , but dem l50 may not , indicating that a different explanation would be required for that bubble . assuming the initial stellar population proposed by @xcite , approximately four sne have gone off in dem l152 , from one 85 progenitor and three 60 progenitors ( see table [ tab : stelpop ] ) .
these sne should input a total 65 of o @xcite and 0.2 - 0.8 of fe ( see discussion in @xcite ) . assuming a number density of 0.01 - 0.2 , spherical geometry
, a radius of 50 pc , and 10% he abundance , the total mass enclosed by the bubble should be between 100 and 3000 . as the inferred number density from the observed spectrum ( see equation [ eqn : dens ] ) is 0.2 , we assume an enclosed mass of 3000 in what follows .
we use equations 9 and 10 from @xcite to calculate the resulting metallicity : @xmath319 @xmath320 @xmath321 is the mass ejected in sne , @xmath322 is the mass evaporated from the shell walls , and @xmath323 and @xmath324 are the solar abundances by mass from @xcite , @xmath325 and @xmath326 .
@xcite predict that the evaporated mass will be the dominant source of mass in the bubble interior .
we assume that , apart from the mass ejected by the four sne , the remainder of the mass was contributed by evaporation .
assuming stellar remnant masses of @xmath327 @xcite and the initial masses from @xcite , the total mass returned to the ism by the four most massive stars is 257 , resulting in a @xmath322 of @xmath283 2700 .
then , from equations [ eqn : silicho ] and [ eqn : silichfe ] , the predicted metallicities are @xmath328 and @xmath329 .
if the bubble contains less evaporated mass , these metallicities will be higher . as noted by @xcite ,
the x - ray emissivity is proportional to the metallicity . assuming the @xmath3 elements dominate the soft x - ray emission
, the factor of 10 increase from @xmath330 of 0.4 to @xmath331 of 3 would be sufficient to explain the factor of 10 increase in dem l152 s luminosity ( see table [ tab : models ] . )
this predicted metallicity does not match the observed @xmath3 abundance in dem l152 , however ( see table [ tab : fit152 ] ) , which is subsolar in all cases with reasonable error bars .
the @xmath3/fe ratio is a more reliable diagnostic of the metallicity , and it is possible that the @xmath3 abundance from the spectral fits is incorrect . given the additional uncertainties in the bubble s density , geometry , and resulting mass and in the number of sne that have occurred , this discrepancy between the predicted and observed metallicities is not unexpected .
the predicted @xmath3/fe ratio from the 4 sne is 5 - 8 , while table [ tab : fit152 ] shows typical @xmath3/fe ratios of around 3 .
when errors are taken into account , the observed @xmath3/fe ratio could be consistent with the predicted value , and therefore it is possible that the higher @xmath3 abundance is also present .
if we use the observed @xmath3 abundances , which have an average value of @xmath332 and are generally consistent with 1 , the enhancement over the weaver model only accounts for 14 - 20% of the observed emission . at most ,
the observed @xmath3 abundance can account for the diffuse emission in the bubble and west regions and the west blowout , but not for the emission from the bubble limb , west limb , or south bar .
if , on the other hand , we accept that this observed @xmath3 abundance is inaccurate and use the predicted @xmath331 , the enhanced metallicity explains the x - ray luminosity but not the limb - brightened features or enhanced expansion velocities .
a similar analysis for dem l50 yields a predicted @xmath331 of 5 and predicted @xmath3/fe of 7 - 13 , where we have assumed @xmath333 , @xmath334 , a bubble radius of 50 pc , and @xmath335 , derived from the non - equilibrium fit normalization .
the predicted metallicities are much higher than the approximately solar values obtained from the non - equilibrium fit , and as with dem l152 , the observed metallicities can not account for the majority of the x - ray enhancement .
another possible explanation for the high x - ray luminosities of dem l50 and dem l152 is mass loading from the evaporation or ablation of overrun clouds .
while the distribution of mass from evaporation at the shell walls should be uniform over relatively large scales , evaporating clouds should appear as small clumps of emission in our data .
dem l50 is located in a void and does not exhibit any noticeable clumpiness in its emission .
the two knots we observed in dem l152 and its more crowded hi environment , on the other hand , show that mass - loading by clumps could be occurring in that bubble .
@xcite investigate clouds with @xmath336and radii of 3 - 5 pc , while @xcite find their modeled shocked clouds have number densities of 4 , radii of 1 pc , and are surrounded by a corona with density 0.4 .
to encompass this range of possible cloud parameters , we calculated the emission from clouds with each of these three densities ( @xmath337=0.4 , 4 , and 10 ) , a temperature of @xmath338 k , a metallicity of 0.4 , and radii of 0.5 pc , 1 pc , and 5 pc , all of which should be resolved at chandra s @xmath339 " resolution .
the x - ray emissivity in the 0.2 - 4.5 kev band is approximately @xmath340 erg ps .
@xcite ; this emissivity will result in more emission than we expect to see in the 0.3 - 2.0 kev band , but it should allow us to roughly estimate the type of clouds that could be detectable . assuming a median photon energy of 0.55 kev @xcite , we can calculate the predicted count rate per arcsec@xmath341 ( table [ tab : knotcts ] ) .
a comparison with the x - ray contours in figures [ fig : img_deml50 ] and [ fig : img_deml152 ] shows that the knots with radii of 0.5 pc are potentially detectable if they have a density of @xmath342 , while no knots with a density of @xmath343are detectable .
interestingly , the detected knots " in dem l152 have a count rate of approximately @xmath344 s@xmath345 arcsec@xmath346 , which is within the range of values in table [ tab : knotcts ] and a density of @xmath339 , found using equation [ eqn : dens ] , assuming a radius of 2.5 pc .
these values suggest that these knots may indeed be genuine clouds swept up by the bubble .
another possibility is that some clumps appeared as point sources in the images ( fig .
[ fig : point_sources ] ) .
we would expect predominantly soft emission from overrun clumps , while other point sources , such as background agn or x - ray binaries , should appear in the hard band as well .
only two point sources , one in dem l50 and one in dem l152 , appear in the soft band alone .
the point source in dem l50 is quite faint , containing only @xmath347 counts , and its identity is unclear .
the soft point source in dem l152 is located outside the main bubble .
this source is likely associated with the highly ionized n44 c region rather than an independent cloud swept up by dem l152 . in any case
, the lack of multiple soft point sources suggests that dense , overrun globules are not present to any significant degree in either bubble .
llll 0.4 & @xmath348 & @xmath349 & @xmath350 + 4.0 & @xmath351 & @xmath352 & @xmath353 + 10.0 & @xmath354 & @xmath355 & @xmath356 + although dem l50 and dem l152 may have overrun additional clouds which are below our detection limit or already destroyed , their morphology and spectral properties suggest that mass - loading by clouds is not dominating the bubbles evolution .
modeled bubbles with high levels of mass loading tend to show a smoother distribution of density and temperature across the bubble , as the clouds add cooler gas to the system ( see e.g. * ? ? ?
* ; * ? ? ?
the increased central density and reduced central temperature may result in a flat surface brightness profile @xcite or even a centrally - peaked profile @xcite .
a limb - brightened morphology , on the other hand , appears correlated with supernova activity .
no pre - supernova superbubbles yet studied show limb - brightening , while this morphology is common in post - supernova superbubbles @xcite .
dem l50 and dem l152 s limb - brightened morphologies likely reflect their supernova history rather than a history of high mass - loading .
mass - loading should reduce the central temperature of a bubble ( e.g. * ? ? ?
* ) , whereas dem l152 s central temperature is quite high ( @xmath357 kev ) , higher than predicted by the weaver model .
dem l50 s lower central temperature of 0.15 kev in the equilibrium fit could be consistent with mass - loading , although uncertainties in the number of sne or energy input in dem l50 could also account for the slightly lower temperature .
the temperature derived from the non - equilibrium fit to dem l50 is completely consistent with the weaver model predictions , however , requiring no additional mass - loading .
mass - loading by overrun clouds or evaporation from the shell walls is likely occurring to some extent in both bubbles , but it does not fully explain the bubbles x - ray emission or large - scale properties .
dense cloudlets have been detected in other superbubbles , but their presence does not necessarily imply a high x - ray luminosity .
for instance , although clouds are detected in the superbubble dem l192 and the bubble appears to be expanding into an inhomogeneous ism , its x - ray luminosity in the 0.3 - 3.0 kev band is only moderate , @xmath358 @xcite .
@xcite note a correlation between the h@xmath3 and x - ray surface brightnesses in this bubble , however , and suggest that mass - loading is occurring through evaporation at the shell walls .
this analysis is consistent with our findings in this section and in [ sec : models : hd ] ; thermal conduction and the subsequent evaporation of shell material is important in dem l50 and dem l152 , while the contribution of swept - up clouds to the x - ray emission is negligible .
thus , we see evidence that both metallicity enhancement and mass - loading by clouds has taken place in these superbubbles .
the metallicity enrichment is too low to account for the observed luminosities , however .
likewise , few candidate clouds are observed in the superbubbles and the x - ray emission is inconsistent with a high mass - loading scenario .
metallicity and mass - loading by clouds do not completely explain the observed x - ray luminosities .
we present high spatial resolution _ chandra _
acis - s images and spectra of the diffuse emission in two x - ray - bright lmc superbubbles , dem l50 and dem l152 .
these objects have x - ray luminosities on the order of @xmath299 , an order of magnitude higher than predicted by the standard @xcite wind - blown bubble model .
the level of detail in the _ chandra _ observations allows us to evaluate the relative importance of the various mechanisms that contribute to the x - ray luminosity by constraining the spatial origin of the emission and revealing x - ray spectral variations across the superbubbles . with these high - quality images , we isolated the diffuse x - ray emission by excluding point sources and examined the spectra of separate regions within each superbubble .
x - ray images of both bubbles show limb - brightened morphologies . most of the limb - brightening in dem l50 is the result of a higher central absorption column density , while the limb - brightening at the bubble s southern end may be real . likewise
, the intervening absorption does not account for the limb - brightening in shell 1 of dem l152 .
hardness ratios indicate that these same regions , the bright limb subregion of dem l50 and shell 1 of dem l152 , are spectrally harder than the other regions in the superbubbles .
spectral fits to the superbubbles account for the variations in absorption column density from region to region and reveal the temperature differences and luminosity contribution of each region .
the spectra of both objects show evidence of snr impacts on the shell walls .
in particular , dem l50 s bright limb has a harder spectrum , higher temperature , and higher surface brightness than the rest of the bubble .
these spectral properties support the idea that a snr recently heated the limb region .
the limb - brightened shell 1 region in dem l152 also has a higher temperature and surface brightness compared to other regions of the superbubble .
a high temperature component in the limb - brightened west region and the adjacent south bar blowout may be another indicator of recent heating due to sne .
our spectral fits also revealed unexpected abundance ratios in dem l50 , which suggest the superbubble is not in collisional ionization equilibrium .
dem l50 s low @xmath3/fe ratio is the opposite of the expected @xmath3-enrichment from core - collapse sne .
while we can not rule out the possibility of a type ia sn in the area , the consistency of the ratio across the bubble suggests that either the bubble happens to lie in an unusual ism environment or , more likely , dem l50 is not in ionization equilibrium . assuming a non - equilibrium situation , the derived ionization timescales are on the order of 10@xmath359 years , which is consistent with the other indications of a recent supernova in the bubble .
we compared our observations with both the standard wind - blown bubble model @xcite and a one - dimensional hydrodynamic model based on @xcite .
both models include the time - dependent energy input of the stellar population . as shown by the weaver model
predicts a total x - ray luminosity an order of magnitude below the observed value . while the 1-d hydrodynamic simulation reproduces the size and expansion velocity of dem l50 by assuming a higher ambient pressure and including the effects of snr impacts , the neglect of thermal conduction produces x - ray luminosities far below our observed luminosities .
this discrepancy indicates that thermal conduction is important in these bubbles , as evaporation of material from the shell walls enhances the density of the interior .
the snr impact does significantly brighten the modeled bubble , particularly at the limb , which supports the off - center snr explanation for the existence of x - ray - bright superbubbles @xcite .
our observations do not reveal signs of significant metal enrichment , an alternative explanation for the high observed luminosities @xcite .
a higher proportion of metals could enhance the x - ray emissivity , but neither superbubble exhibits a sufficiently high metallicity to account for the anomalous x - ray luminosity .
abundances determined from x - ray observations are uncertain , however , and the bubbles may be more enriched than the spectral fits suggest . while a high metallicity could raise the luminosity , it would still not explain the observed morphologies or large expansion velocities of the objects .
we consider the hypothesis that mass - loading through the ablation or evaporation of swept - up clouds could increase the density of the superbubbles and raise the x - ray luminosity ( e.g. * ? ? ?
* ; * ? ? ?
* ) . with _ chandra _
s @xmath283 1 " resolution , we expect to detect clouds a few parsecs in diameter .
few candidate clouds are observed , and they contribute negligible emission to the superbubbles .
although some mass - loading is likely present and the superbubbles may have overrun more clouds in the past , the effect of clouds on the x - ray emission is insignificant compared to the luminosity enhancement from snr impacts .
an examination of the energy budget for both superbubbles further clarifies the role of off - center snrs .
the relatively small bright limb subregion of dem l50 generates roughly a quarter of the total luminosity , while the limb as a whole accounts for half of the total x - ray luminosity . in dem l152 ,
the shell 1 limb and west limb regions alone provide about half of the total x - ray luminosity .
all these regions exhibit signs of supernova activity . on the other hand ,
the contribution of metallicity enrichment and mass - loading from clouds to the x - ray luminosity is much less .
metallicity enrichment accounts for at most 20% of the excess emission , while dense clouds account for less than 2% of the total emission .
snr impacts and evaporation of material from the shell walls dominate the observed x - ray luminosities .
our _ chandra _ images and spectra of dem l50 and dem l152 are the first observations of x - ray - bright superbubbles at a sufficiently high spatial and spectral resolution to distinguish between various emission mechanisms .
while mass - loading and metallicity enrichment due to sne must be occurring to some degree , our observations confirm that thermal evaporation from the shell walls and off - center snr impacts are the primary sources of the enhanced x - ray luminosity .
snr impacts increase the shell expansion velocity , lead to higher radiative losses , and may contribute to blowouts and the subsequent mixing of ism and hot , enriched gas .
although the x - ray luminosities are not high enough to explain the superbubble growth - rate discrepancy , x - ray radiative losses will decrease the energy available for transport to the ism . unlike the cloud mass - loading scenario
, snr impacts and thermal conduction should affect most superbubbles at some point in their evolution , irrespective of their environment .
off - center snr impacts in superbubbles play a critical role in the evolution of the ism by enhancing the mixing of hot , enriched gas with ambient material , while at the same time leading to x - ray cooling that decreases the available energy .
, c. e. , arnal , e. m. , cichowolski , s. , goss , w. m. , & pineault , s. 2003 , in iau symposium , vol .
212 , a massive star odyssey : from main sequence to supernova , ed .
k. van der hucht , a. herrero , & c. esteban ( san francisco , ca : asp ) , 596 | the hot , x - ray - emitting gas in superbubbles imparts energy and enriched material to the interstellar medium ( ism ) and generates the hot ionized medium , the ism s high - temperature component .
the evolution of superbubble energy budgets is not well understood , however , and the processes responsible for enhanced x - ray emission in superbubbles remain a matter of debate .
we present _ chandra _ acis - s observations of two x - ray - bright superbubbles in the large magellanic cloud ( lmc ) , dem l50 ( n186 ) and dem l152 ( n44 ) , with an emphasis on disentangling the true superbubble x - ray emission from non - related diffuse emission and determining the spatial origin and spectral variation of the x - ray emission .
an examination of the superbubble energy budgets shows that on the order of 50% of the x - ray emission comes from regions associated with supernova remnant ( snr ) impacts .
we find some evidence of mass - loading due to swept - up clouds and metallicity enrichment , but neither mechanism provides a significant contribution to the x - ray luminosities .
we also find that one of the superbubbles , dem l50 , is likely not in collisional ionization equilibrium .
we compare our observations to the predictions of the standard weaver et al .
model and to 1-d hydrodynamic simulations including cavity supernova impacts on the shell walls .
our observations show that mass - loading due to thermal evaporation from the shell walls and snr impacts are the dominant source of enhanced x - ray luminosities in superbubbles .
these two processes should affect most superbubbles , and their contribution to the x - ray luminosity must be considered when determining the energy available for transport to the ism . |
quantum coherence and its destruction by coupling to a dissipative environment play an important role in the transport phenomena of a particle moving in a potential @xcite .
well known examples include electron transfer in molecular and biological systems @xcite , many chemical reactions @xcite , squid rings @xcite , quantum ratchets @xcite , nonlinear optical processes @xcite , and tunneling processes in device systems @xcite .
such systems are commonly modeled as one - dimensional or two - dimensional potential systems coupled to heat bath degrees of freedom , which drive the systems toward the thermal equilibrium state .
the heat bath degrees of freedom are then reduced using such methods as the projection operator method or the path integral method , for example .
many equations of motion have been derived for the purpose of understanding the quantum aspects of dissipative dynamics @xcite .
because a complete picture of quantum dissipative dynamics must treat phenomena that can only be described in real time , a great deal of effort has been dedicated to the problem of numerically integrating these equations of motion in real time @xcite .
although these equations are analogous to the classical kinetic equations , which have proved to be useful for classical transport problems , such equations can not be derived in a quantum mechanical framework without significant approximations and/or assumptions .
for example , the quantum boltzmann equation is based on the assumption that the effects of collisions between electrons can be described by the rates determined from fermi s golden rule , and hence it is regarded as a semi - classical equation @xcite .
similarly , the quantum fokker - planck equation can be derived from the caldeira - leggett hamiltonian under a markovian approximation , but in order for this to be possible , the heat bath must be at a sufficiently high temperature , in which case most of the important quantum dynamical effects play a minor role @xcite .
treatments of these kinds are therefore not sufficient to construct fully quantum mechanical descriptions of broad validity
. to circumvent this problem , we present a quantum mechanical approach , which is valid for arbitrary temperatures .
this treatment employs the reduced hierarchy equations of motion ( heom ) , and it can be used in application to systems for which fully quantum mechanical description is necessary @xcite . in particular , the reduced heom approach can be used to numerically treat non - markovian system - bath coupling in a non - perturbative manner @xcite . here , we apply this approach to study the dynamics of a resonant tunneling diode ( rtd ) described by the caldeira - leggett hamiltonian . the rtd system that we consider is modeled by a double barrier structure with an electrostatic potential representing a region consisting of an undoped layer positioned between two doped layers .
( see figure .
[ fig1 ] ) .
the double barrier structure constitutes a single quantum well with discretized energy states for electrons which are called resonant levels .
the number of resonant levels depends on the height of the barriers .
when a bias is applied to the rtd system , as long as the energy of electrons which flow in the rtd is lower than the energy of the resonant level , the most of electrons are reflected by the barrier because of the small transmission coefficient .
when the energy of the electrons matches the resonant energy , electrons can go through the barriers efficiently due to resonant tunneling , and the current acquires the maximum value . on the other hand , the current decreases
after the energy of the electrons exceeds the resonant energy . as a result , rtd systems exhibit characteristic negative differential resistance ( ndr ) in the current - voltage ( i - v ) relation @xcite . until now , rtds have been mostly used as high - frequency oscillators device using ndr characteristics because the tunneling is the fastest charge - transport mechanism in semiconductors @xcite .
resonant tunneling diodes are presently the highest - frequency active semiconductor devices in existence @xcite and oscillation frequencies above 1 thz have recently been realized @xcite . from a fundamental physics point - of - view
, the rtd system provides a simple and convenient `` context '' for studying and testing various methods of analysis for nanoscale quantum devices @xcite .
frensley discovered ndr in the i - v curve through a numerical computation treating a quantum liouville equation in the wigner representation that adopted open boundary condition and ignored phonon - scattering processes @xcite .
kluksdahl et al
. incorporated dissipative and self - consistent effects , employing the poisson - boltzmann equation by adopting a relaxation time approximation , and succeeded in modeling the experimentally observed hysteresis behavior of the i - v curve @xcite .
jensen and buot developed a numerical scheme to treat systems of the same kind and found evidence that the current oscillation and plateau - like behavior arise from intrinsic bistability @xcite . when plateau - like behavior and hysteresis of the i - v curve in the ndr region , which were thought to arise from the feedback of the electrostatic field , were experimentally observed @xcite
, sollner claimed that they result merely from resonance with the external circuit @xcite .
although theoretical calculations have provided evidence of intrinsic bistability and self - excited current oscillations in the ndr region , such phenomena have not been justified by experimental means . in addition , because there exists no well - established methodology that can be applied rigorously to this type of model and includes the effect of dissipation , which is the origin of joule heat , previous theoretical results have not been well justified .
the heom approach is ideal to clarify a role of bistability in the ndr region and for detailed analysis of the rtd system .
this paper is organized as follows . in sec .
2 , we introduce the reduced hierarchy equations of motion applicable to the resonant tunneling problem .
we then present the computational details for the numerical simulations in sec .
numerical results for the i - v curves and current oscillations are presented in secs .
section 5 is devoted to concluding remarks .
we consider the following caldeira - leggett hamiltonian @xcite , which describes the dynamics of an electron subjected to a thermal environment : @xmath0 .
\label{clhamiltonian}\end{aligned}\ ] ] here , @xmath1 , @xmath2 , and @xmath3 are the mass , momentum and position variables of the electron , and @xmath4 and @xmath5 are the mass , momentum , position and frequency variables of the @xmath6th phonon oscillator mode . in equation ( [ clhamiltonian ] ) , the electron - phonon interaction is given by @xmath7 here , @xmath8 is any function of @xmath3 and the quantities @xmath9 are coefficients that depend on the nature of the electron - phonon coupling .
the heat bath can be characterized by the spectral distribution function , defined by @xmath10 and the inverse temperature , @xmath11 , where @xmath12 is the boltzman constant .
we assume the drude distribution , given by @xmath13 where the constant @xmath14 represents the width of the spectral distribution of the collective phonon modes and is the reciprocal of the correlation time of the noise induced by phonons .
the parameter @xmath15 is related to the electron - phonon coupling strength . for the collective heat bath coordinate @xmath16 , the canonical and symmetrized correlation functions , respectively defined by @xmath17 and @xmath18 , where @xmath19 is the heisenberg representation of @xmath20 , and @xmath21 represents the thermal average over the bath modes ,
are given by @xcite @xmath22 and @xmath23 here , @xmath24 are the matsubara frequencies , and we have @xmath25 , \label{eq : c_0}\end{aligned}\ ] ] and @xmath26 the function @xmath27 is analogous to the classical correlation function of @xmath28 and corresponds to the correlation function of the bath - induced noise , whereas @xmath29 corresponds to dissipation .
the noise @xmath27 is related to @xmath29 through the quantum version of the fluctuation - dissipation theorem , @xmath30 = \hbar \omega \coth(\beta \hbar \omega/2 ) /2
\psi[\omega]$ ] , which insures that the system exists in the thermal equilibrium state for finite temperatures @xcite .
note that in the high temperature limit , @xmath31 , the noise correlation function reduces to @xmath32 .
this indicates that the heat bath oscillators interact with the system in the form of gaussian - markovian noise . to derive the equation of motion for the electron
, we use the reduced density operator of the system by taking the trace over the heat bath degrees of freedom : @xmath33 in the path integral representation , the reduced density matrix elements are written @xmath34\int { d[q ' ( \tau ) ] \int { dq_i } \int { dq'_i } } } \rho ( q_i , q'_i ) \rho_{\mathrm{cs } } ( q,\,q_i,\,q',\,q'_i;\,t ) \nonumber \\ \times { \rm e } ^{\frac{\rm i } { \hbar } s_a [ q;\,t ] } f[q,\,q';\,t]{\rm e}^ { - \frac{\rm i } { \hbar } s_a [ q';\,t ] } , \label{eq : rho_reduced}\end{aligned}\ ] ] where @xmath35 $ ] is the action for the hamiltonian of the system , @xmath36 , expressed as @xmath37 \equiv \int_{t_i}^{t } d\tau \left [ \frac{1}{2 } m \dot q^2(\tau ) - u(q ( \tau ) ; \tau ) \right ] , \label{eq : actionfunc}\end{aligned}\ ] ] @xmath38 is the initial state of the system at time @xmath39 , @xmath40 $ ] is the influence functional @xcite , and @xmath41 is the initial correlation function between the system and the heat bath @xcite .
the functional integrals for @xmath42 and @xmath43 are carried out from @xmath44 to @xmath45 and from @xmath46 to @xmath47 , respectively . in the heom approach
, we can specify @xmath48 by nonzero hierarchy elements . to simplify the derivation of the heom ,
here we set @xmath49 and regard @xmath50 as a temporal initial condition .
then , after deriving the heom , we take into account @xmath48 through implementation of a hierarchy of initial conditions that can be evaluated numerically @xcite .
the influence functional for the inverse temperature @xmath51 is given by @xcite @xmath52 = { \rm exp}\left\ { \left ( { - \frac{\rm i}{\hbar } } \right)^2 \int_{t_i } ^t { d\tau } v^{\times}(q,\,q';\,\tau ) \right .
\times \left . \left
[ \frac{\partial } { \partial \tau } \int_{t_i } ^{\tau } { d\tau ' } \,\frac{i \hbar}{2 } \psi ( \tau - \tau ' ) v^{\circ}(q,\,q';\,\tau ' ) + \int_{t_i } ^{\tau } { d\tau ' } c(\tau - \tau ' ) v^{\times}(q,\,q';\,\tau ' ) \right ] \right\ } , \nonumber \\ \label{eq : influenceq}\end{aligned}\ ] ] where @xmath53 and @xmath54 .
if we choose @xmath55 so as to satisfy @xmath56 , where @xmath57 is the characteristic frequency of the system such as the frequency of self - excited current oscillations , the factor @xmath58 in equation ( [ eq : l_2gmdef ] ) can be replaced with dirac s delta function , using the approximation @xmath59 .
therefore , @xmath27 can be expressed as @xmath60 by choosing @xmath61 , the above expression allows us to evaluate @xmath27 for finite @xmath55 with negligible error at the desired temperature @xmath62 .
the reduced hierarchy equations of motion ( heom ) can be obtained by considering the time derivative of the reduced density matrix with the kernel given in equations ( [ eq : l_1bargm ] ) and ( [ eq : l_2gmdef2 ] ) .
the heom have been used to study chemical reactions @xcite , linear and nonlinear spectroscopy @xcite , exciton transfer @xcite , electron transfer @xcite , quantum dots @xcite , quantum ratchet@xcite , and quantum information @xcite .
a variety of numerical techniques have been developed for the heom approach in order to accelerate numerical calculations @xcite .
the accuracy of the heom approach has been justified for a brownian oscillator system @xcite , a displaced brownian oscillators system @xcite , and a spin - boson system @xcite via linear and nonlinear response functions by comparing the analytical solutions of the response functions @xcite .
the validity of the heom are also confirmed with other numerically methods such as the iterative quasi - adiabatic propagator path - integral scheme and a time - convolution less master equation in the relevant crossover regime from weak to strong system - bath coupling @xcite .
the heom are ideal for studying quantum transport systems , in conjunction with the wigner representation , characterized by the wigner distribution function , @xmath63 because they allow us to treat continuous systems utilizing open boundary conditions and periodic boundary conditions @xcite .
although the wigner distribution function is not positive definite , it is the quantum analogue of the classical distribution function in the phase space @xcite .
its classical limit can be computed readily .
this is helpful , because knowing the classical limit allows us to identify the purely quantum mechanical effects @xcite .
while we can handle any form of @xmath64 @xcite , here we consider the linear - linear system bath coupling case defined by @xmath65 . in the wigner representation ,
the equations of motion are expressed in hierarchical form as follows @xcite : @xmath66w^{(n)}_{j_1 , \dots , j_k}(t ) \nonumber \\
+ \hat{\phi}\left[w^{(n+1)}_{j_1 , \dots , j_k}(t ) + \sum_{k=1}^k w^{(n)}_{j_1 , \dots , ( j_k+1 ) , \dots j_k}(t ) \right ] \nonumber \\
+ n\gamma\hat{\theta}_0 w^{(n-1)}_{j_1 , \dots , j_k}(t ) + \sum_{k=1}^k j_k \nu_k \hat{\theta}_k w^{(n)}_{j_1 , \dots , ( j_k+1 ) , \dots , j_k}(t ) \label{heom}\end{aligned}\ ] ] for non - negative integers @xmath67 , where we have chosen @xmath55 such that @xmath68 . in equation ( [ heom ] )
, @xmath69 is the quantum liouvillian in the wigner representation , given by @xmath70 with @xmath71 .
\label{wigner_pot } \label{liouville2}\end{aligned}\ ] ] the other operators appearing in equation ( [ heom ] ) are the bath - induced relaxation operators , defined as @xmath72 @xmath73 , \label{heomt1}\end{aligned}\ ] ] @xmath74 and @xmath75 + \sum_{k=1}^k \frac{c_k}{\nu_k } \right\ } % # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # 3 \frac{\partial^2 } { \partial p^2}. % # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # \label{heomt3}\end{aligned}\ ] ] in the case that the quantity @xmath76 satisfies the relation @xmath77 , this infinite hierarchy can be truncated with negligible error at the desired temperature @xmath62 by the terminator @xcite @xmath78 the validity of the above truncation scheme and its extension for efficient numerical calculations have been discussed for the spin - boson system @xcite .
note that only @xmath79 has physical meaning , and the other elements @xmath80 with @xmath81 are auxiliary operators introduced to avoid the explicit treatment of the inherent memory effects that arise in the time evolution of the reduced density matrix .
if the noise correlation is very short ( @xmath82 ) and the temperature is high ( i.e. , the noise is white ) , the quantum fokker - planck equation can be derived in a form similar to that of kramers equation @xcite . in the present case , however , we can not employ the white noise approximation , because quantum effects play a dominant role in the low temperature regime ( @xmath83 ) @xcite .
( a ) the structure of the rtd .
the well consists of undoped gaas ( 4.520 nm ) , the barriers consist of undoped algaas ( 2.825 nm ) , the spacer layers consist of undoped gaas ( 2.825 nm ) , and the contact regions consist of doped gaas ( with a doping concentration of @xmath84 @xmath85 ) . in order to elucidate the dependence of the current on the width of the contact regions , we carried out computations for three values of this widths : ( i ) 16.950 nm ; ( ii ) 33.900 nm ; ( iii ) 42.375 nm .
this figure depicts the case of 16.950 nm .
( b ) the structure of the conduction band edge .
the height of the potential barriers is 0.27 ev .
, width=491 ] in equation ( [ wigner_pot ] ) , @xmath86 is the effective potential for the electron , which can be written @xcite @xmath87 where @xmath88 and @xmath89 are the static and self - consistent parts , respectively . as the static potential
, we employ the double - barrier structure depicted in figure 1 ( b ) . the self - consistent part , @xmath90 ,
is calculated from the electron distribution at each time step in the integration of equation ( [ heom ] ) using the poisson equation @xmath91 = e \left[n^{+}(q ) - p(q , t ) \right ] , \label{poisson}\end{aligned}\ ] ] where @xmath92 is the dielectric constant , @xmath93 is the doping density and @xmath94 is the electron density calculated from the wigner distribution .
coupling the heom to the poisson equation , we obtain a fully self - consistent model of quantum electron transport .
this allows us to examine charge redistribution effects .
the equations of motion given in ( [ heom ] ) were numerically evaluated using finite mesh representations of the wigner distribution functions .
the spatial derivative of the kinetic term in the liouville operator , @xmath95 , was approximated by using a third - order left - handed or right - handed difference scheme .
note that the first - order difference scheme is not sufficiently accurate for the present problem , because this scheme introduces false diffusion for the wavepacket dynamics , which suppresses the self - excited current oscillations . depending on the sign of the momentum ,
the expressions are given by @xmath96 for @xmath97 , and @xmath98 for @xmath99 , in order to treat continuous systems utilizing the inflow and outflow boundary conditions with use of the first - order difference scheme at @xmath100 ( @xmath101 ) and @xmath102 ( @xmath103)@xcite .
the inflow boundary conditions were set by stipulating that @xmath104 and @xmath105 are given by the equilibrium distribution of a free particle calculated from the heom with periodic boundary conditions .
due to fluctuations and dissipation , the flow of a wavepacket reaches a steady state even when there exists a non - zero bias voltage .
the validity of the boundary conditions was verified by considering several system sizes .
other derivatives with respect to @xmath106 were approximated using the fourth - order centered difference scheme given by @xmath107 and @xmath108 the mesh size for the position , @xmath109 , and momentum,@xmath110 , are respectively 0.2825 nm and 3.540 @xmath111 .
we set the parameters used in the heom as @xmath112 thz ( @xmath113 fs ) , @xmath114 ghz ( @xmath115 ps ) , and @xmath116 k in order to create conditions close to those used in previous theoretical studies . in appendix b
, we also report the results of calculations for the strong coupling case , with @xmath117 ghz ( @xmath118 ps ) , to elucidate the role of dissipation .
the depth of the hierarchy and the number of matsubara frequencies were chosen as @xmath119 and @xmath120 , respectively . to model gaas
, the effective mass of the electron was assumed to be constant across the device and equal to @xmath121 , where @xmath122 is the electron mass in vacuum . the dielectric constant in equation ( [ poisson ] )
was set as @xmath123 . as the static double - barrier potential , which models the hetero - structure of gaas sandwiched between two thin algaas layers , we set the widths of quantum well ( undoped gaas ) , barrier ( undoped algaas ) , and spacer layer ( undoped gaas ) to be 4.520 nm , 2.825 nm , and 2.825 nm , respectively .
the height of the potential barriers was 0.27 ev .
the conduction band edge consists of a single quantum well bounded by tunneling barriers ( figure 1(b ) ) .
the widths of the contact regions ( the yellow parts in figure 1(a ) , where gaas is doped with a concentration of @xmath84 @xmath85 ) were chosen as 16.950 nm , 33.900 nm , and 42.375 nm .
note that , to adapt the one dimensional model , we rescaled the concentration by multiplying the doping density by unit area .
in a previous study @xcite , we chose a smaller value of @xmath14 , @xmath124 thz ( @xmath125 fs ) , and fixed the width of the contact region as 42.375 nm .
note that , since the effective system - bath coupling strength is estimated as @xmath126 , where @xmath57 is the characteristic frequency of the system @xcite , the damping strength in the present case is slightly larger than the previous case . in that case , we found hysteresis , double plateau - like behavior , and self - excited current oscillation in the negative differential resistance ( ndr ) region of the current - voltage curve .
we found that while most of the current oscillations decay in time in the ndr region , there exists a non - transient oscillation characterized by a tornado - like rotation in the wigner space in the upper plateau of the ndr region . in this paper ,
we explore the cause of such current oscillations by considering several values of the width of the contact regions in cases of both weak and strong coupling , characterized by different values of @xmath15 .
we determined the characteristics of the current - voltage ( i - v ) according to the following procedure .
first , we integrated equation ( [ heom ] ) at zero bias voltage without the self - consistent part of the effective potential under the inflow boundary conditions specified above . when we obtained the _ temporal _ steady state , the obtained distribution was then used as the initial distribution for the self - consistent calculation , and we then integrated equation ( [ heom ] ) again , with the effective potential @xmath86 evaluated iteratively using the poisson equation given in ( [ poisson ] ) . under this procedure ,
when the distributions reached the _ genuine _ steady state @xmath127 , the current was calculated by @xmath128 and then the _ genuine _ state was used as the initial distributions for the next bias step . while the _
temporal _ steady states were obtained for the static potential , @xmath88 , the _ genuine _ steady states were calculated from the effective potential , @xmath129 . since the value of the effective potential depended on the hysteresis of a physical process and since we wanted to use a uniquely determined steady state , we chose the _
temporal _ steady state as a temporal initial state to have the _ genuine _ steady state . following the above steps , we increased the bias from 0.000v to 0.500v , and then decreased it to 0.000v with bias steps of 0.01 v in the normal region and 0.002 v in the negative differential resistance ( ndr ) region .
the corresponding sweeping rates were @xmath130 in the normal region and @xmath131 in the ndr region .
we found that the profiles of the i - v curves did not change for slower sweeping rate than the present values , whereas the width of plateau observed in the ndr region often became smaller for a faster sweeping rate . at each step , we integrated the equation of motion until the system exhibited the steady current .
however , in some cases , in the ndr region , steady current oscillation arose .
in such cases , the value of current in figure 2(a ) was evaluated as a time average after stable oscillations were realized ( between 30 to 40 ps in most cases ) .
( a ) the i - v characteristics for three sizes of the contact regions : ( i ) 16.950 nm ; ( ii ) 33.900 nm ; ( iii ) 42.375 nm . the black curve with circles represents the case in which the bias is increasing , and the blue curve with the @xmath132s represents the case in which the bias is decreasing .
( b ) time evolution of the current for the value of the bias voltage at which current oscillation occurs .
the red curves represent the current oscillation with large amplitude , and the green curves represent the current oscillation with small amplitude .
the inset in ( ii - b ) contains a close - up view of one portion of the current .
the snapshots of the wigner distribution at the time points marked with purple circles are presented in figures 4 and 5.,width=377 ] in figure 2 , we present ( a ) current - voltage ( i - v ) relations and ( b ) the time evolution of the self - excited current oscillation in the weak coupling case ( @xmath114 ghz ) for three values of the width of the contact regions ( the yellow parts in figure 1(a ) ) . in these plots , the sizes of the contact regions are ( i ) 16.950 nm , ( ii ) 33.900 nm , and ( iii ) 42.375 nm , respectively , with a fixed doping concentration of gaas .
these graphs reveal ndr behavior , hysteresis , and plateaus in the i - v curve .
moreover , self - excited current oscillation appears in some regions of the plateau . while jensen and buot observed only a single plateau similar to that in figure 2(i - a ) with current oscillation @xcite , our results in figure 2(ii - a ) and 2(iii - a ) exhibit a double plateau structure .
while the experimental result shown by @xcite is similar to figure 2(i - a ) , those shown by @xcite are similar to figures 2(ii - a ) and 2(iii - a ) . for the sake of comparison
, we present a graph corresponding to figure 2 calculated from the boltzmann equation in appendix c. in the case of a single plateau structure ( figure 2(i - a ) ) , the width of the plateau is large , while in the case of a double plateau structure ( figures 2(ii -
a ) and ( iii - a ) ) , the width is small .
the i - v curves obtained in the strong coupling case ( @xmath117 ghz ) are presented in appendix b. in that case , ndr behavior , hysteresis , and a single plateau in the i - v curve are observed , but steady current oscillation does not appear , even in the plateau .
( a ) the frequency distribution of the current oscillation that arises in the plateau region in the case of increasing bias . the size of contact region and the bias voltage are as follows : ( i ) 16.950 nm and 0.332 v ( red curve in figure 2(i - b ) ) ; ( ii ) 33.900 nm and 0.280 v ( green curve in figure 2(ii - b ) ) ; ( iii ) 42.375 nm and 0.292 v ( green curve in figure 2(iii - b ) ) ; ( iii@xmath133 ) 42.375 nm and 0.302 v ( red curve in figure 2(iii - b ) ) .
the insets depict the corresponding transitions .
the colored lines represent the eigenstates in the emitter basin given in the right graph , while the thick grey line represents the continuous energy band .
( b ) the time - averaged effective potential ( black solid curve ) and time averaged electron density ( black dashed curve ) for ( i)-(iii ) .
the red , green , blue , orange , and purple curves represent the eigenstates in order of increasing eigenenergy calculated using the averaged effective potential without the heat bath .
basin - like structures denoted in the red squares are formed on the emitter side of the potential ( emitter basin ) .
, width=453 ] as in the case considered in the previous paper @xcite , we find that most of the current oscillations decay in time in the ndr region , but there also exist non - decaying oscillations in some regions of the plateau , as seen in figures 2 ( i - b ) - ( iii - b ) .
the fourier components of each persistent oscillation are plotted in figures 3 ( i - a ) - ( iii-a ) .
we can classify these oscillations into two types , according to the current amplitude .
the first type is observed in the single plateau ( figure 2(i ) ) and the lower part of the double - plateau ( figure 2(iii ) ) with large amplitude ( red ) .
the plateau in this case is located in the middle of the ndr region .
the second type is observed in the upper part of the double - plateau ( figures 2(ii ) and 2(iii ) ) with small amplitude ( green ) .
the plateau in this case is located at a current approximately three - quarters of the peak current .
the second type of oscillation contains two fourier components ( figures 3 ( ii - a ) and ( iii - a ) ) , whereas the first type contains just a single component ( figures 3 ( i - a ) and ( iii-a ) ) .
we find that as the size of the contact regions increases , the frequency of each peak decreases . as mentioned by kluksdahl _
_ @xcite and zhao _ et al . _
@xcite , the quantum well formed on the emitter side of the effective potential plays an important role in the realization of hysteresis and plateau - like structure in the ndr region . the time averaged electron densities ( black dashed curves ) and the effective potentials ( black solid curves ) calculated from the wigner distribution function in the case of increasing bias are plotted in figures 3 ( i - b ) - ( iii-b ) . for reference , in appendix a , we present a graph corresponding to figure 3(iii-b ) depicting the situation in the case of decreasing bias .
a basin - like potential on the emitter side ( emitter basin ) is observed in the case of increasing bias , while the emitter basin does not exist in the case of decreasing bias . in the case of increasing bias ,
when the bias exceeds the peak point of the i - v curve , the kinetic energy of the inflowing electron becomes larger than the eigenenergy of the resonant tunneling state . as a result ,
the reflection of the current from the emitter side of the barrier becomes large , and the electron distribution function becomes concentrated near the barrier , as depicted in figure 3 ( i - b ) - ( iii-b ) .
when the electron density increases , the effective potential decreases . as a result
, the emitter basin appears .
when the emitter basin becomes sufficiently deeper , there appear resonant tunneling states between the emitter basin and the double - barrier well . in such a case , we find current oscillation and a plateau of the i - v curve in the ndr region . in the case of decreasing bias ,
however , because there is no resonant tunneling state between the emitter basin and double - barrier well , the current is much smaller than in the case of the increasing bias .
this difference causes the hysteresis behavior . to elucidate this point more clearly ,
we solve the steady - state schrdinger equation for the regions of emitter basin and the double - barrier well , to obtain approximate eigenstates and eigenenergies of an electron whose energy is lower than the continuous energy band on the emitter side . since bound states are not formed in the collector side of the potential , we exclude this region from the calculations . it should be noted that we employ the time - averaged potential for the purpose of the graph , but the effective potential and the corresponding eigenstates actually vary in time , because they depend on the electron distribution function , and it varies in time . thus , for example , the identifications of the first ( red ) and second ( green ) eigenstates and the third ( blue ) and fourth ( orange ) eigenstates change in time , often becoming degenerate and interchanging . in addition , we ignore the continuous band on the collector side and the influence of the heat bath when calculating the eigenstates and eigenenergies . for this reason ,
the calculated eigenenergies are not precise , but we find that they are sufficient for determining the cause of the current oscillation , because each resonant frequency is rather isolated when estimating the oscillation frequency .
we find that each oscillation peak in both the large and small oscillation cases can be attributed to transitions between eigenstates in the emitter basin , as depicted in the insets of figures 3(i - a)-3(iii-a ) .
as shown in figures 3 ( i - b ) and 3 ( iii-b ) , the first and second eigenstates are the tunneling states in the large oscillation case , while the third eigenstate in figures 3 ( ii - b ) and ( iii - b ) is the tunneling state in the small oscillation case .
when we compare the profiles of each eigenstate and the electron density distributions , we find that the tunneling state and the higher energy eigenstate close to the tunneling state are populated in both cases .
thus , we conclude that the current oscillation results from transitions between these two states , with the frequency of the oscillation determined by the frequency of these transitions . since both the first and second eigenstates change in time , the amplitude of current becomes large in the large oscillation case . when the structure of the emitter basin becomes stable with respect to change of the bias voltage , a plateau forms . as the size of the contact regions increases ,
the transition frequencies decrease , because the size of the emitter basin increases , while the depth of the basin does not change .
this accounts for the peak shift seen in figures 3 ( i - a ) and 3 ( iii-a ) and figures 3 ( ii - a ) and 3 ( iii - a ) . for a small size of contact regions , the emitter basin becomes more stable with respect to change in the bias voltage and as the result the plateau becomes larger . both figures
2 ( ii ) and ( iii ) exhibit double plateau - like features in the ndr region , but we find current oscillation only in the upper plateau in the case of figure 2 ( ii ) .
this is because the dissipation in the large oscillation case of figure 2(ii ) is not sufficiently strong to create significant population of the tunneling states in the case of a small basin , for which the resonant frequencies between the tunneling states and adjacent states are large . in our previous study @xcite
, we considered the same condition as in figure 2 ( iii ) , with a value of @xmath14 half as large ( @xmath134 thz ) .
in that case , however , there was only upper plateau oscillation , while we observed the bistability and the double plateau - like feature .
this is because the effective system - bath coupling strength in the previous case is weaker than in the present case @xcite .
thus , in that case , the ground tunneling states are not populated from the conduction band through dissipation .
note , however , that if the system - bath coupling is too strong , dissipation suppresses the current oscillation , as explained in appendix b. current oscillation with a large amplitude observed in the lower part of the double - plateau structure at the times marked on the red curve in figure 2 ( iii - b ) depicted as snapshots of the wigner distribution .
( a ) current flows into the system from the emitter side of the boundary , and then it is scattered ( @xmath135 , where @xmath136 is the momentum of the inflow current ) by the emitter side of the barrier almost elastically .
the scattered electron flows into peak b. ( b ) peaks a and b become higher and lower in turn , exhibiting motion similar to that of pistons in a two - piston engine .
( c ) the current becomes large whenever peak a becomes large due to the tunneling .
, width=264 ] one important aspect of the present methodology is that it allows elucidation of the dynamical behavior of the system through the time evolution of the wigner distribution function .
we have been able to characterize the patterns of the time evolution of the wigner distribution for two types of current oscillations , with large and small oscillation amplitude . here
, we describe these two types in detail , using as one reference , the large oscillation case in figure 2 ( iii - a ) and the small oscillation case in figure 2 ( ii - a ) . in figure 4 , we display snapshots of the wigner distribution function for the case of large oscillation at the times marked on the red curve in figure 2 ( iii - b ) . as illustrated in figures 3 ( i - b ) and 3 ( iii-b )
, the characteristic feature of this type of oscillation is the large electron density near the emitter side of the barrier , which is observed as a distinct peak separated from the conduction band ( at @xmath137 18 nm in figure 3 ( i - b ) and at @xmath137 43 nm in figure 3 ( iii-b ) ) . as a result of this feature ,
the effective potential possesses a deep emitter basin next to the barrier .
the profiles of the eigenstates depicted in figure 3 ( iii-b ) indicate that this peak consists of the first ( red ) to fourth ( orange ) eigenstates .
we find that a small peak near the edge of the conduction state ( at @xmath137 35 nm in figure 3(iii-b ) ) , which arises from the third ( blue ) and fourth ( orange ) eigenstates in figure 3(iii-b ) , also plays an important role in the current oscillation . in the wigner distribution plotted in figure 4(a ) , these two peaks are denoted by a and b , respectively . in the situation depicted in figure 4 ( a ) , the current flows into the system from the emitter side of the boundary , and then it is scattered by the emitter side of the barrier almost elastically , because the kinetic energy of the current electron is much higher than that of the tunneling state .
the scattered current is trapped by the emitter basin and rotates clockwise around the peaks a and b , but , due to dissipation , some of it flows into peak b while losing energy . in the eigenstate representation given in figure 3(iii-b ) , this behavior corresponds to population transfer from the fourth ( orange ) to the third ( blue ) eigenstate . in figure 4 ( b ) , when the third ( blue ) state decays further to the first ( red ) and second ( green ) tunneling states , the height of a increases . as a result ,
peak a becomes higher , while peak b becomes lower . in figure 4(c ) , because peak a is related to the tunneling state , the outflow current becomes larger whenever peak a becomes higher . throughout this process ,
the peaks a and b become higher and lower by turns , in a manner reminiscent of the piston in a two - piston engine . as a result of this motion , the current exhibits oscillation .
this behavior is typical for this large oscillation case . because there is no current in the case of fig .
4 ( a ) , the oscillation amplitude is large compared to that in the small oscillation case .
if the dissipation is too strong , however , the piston - like motion is suppressed , and there is only steady current , as described in appendix b. current oscillation with small amplitude observed in the upper part of the double - plateau structure at the times marked on the green curve in figure 2 ( ii ) depicted as snapshots of the wigner distribution .
( a ) current flows into the system from the emitter side of the boundary . then , a part of the current is scattered ( @xmath135 , where @xmath136 is the momentum of the inflow current ) by the emitter side of the barrier .
the other part of the current flows to the collector side in the form of steady current through tunneling .
( b ) the scattered electron flows in a tornado - like manner to peak c in the emitter basin due to dissipation .
( c ) the shaking motion of the effective potential periodically accelerates the component at c to the tunneling state , and the current is thus enhanced.,width=264 ] in figure 5 , we display snapshots of the wigner distribution for the case of small oscillation at the times marked in figure 2 ( ii - b ) . in figure 5 ( a ) , while current flows into the system from the emitter side of the boundary , that part of it with energy larger than that of the tunneling state is scattered by the emitter side of the barrier .
the remaining current , i.e. , that whose energy is closer to the energy of the tunneling state , flows to the collector side in the form of steady current , through tunneling . in figure 5 ( b ) , it is seen how the scattered electron moves back and forth in the emitter basin , while losing energy due to dissipation . as a result , the electron flows into peak c , exhibiting tornado - like motion .
figure 5 ( c ) depicts shaking motion of the effective potential that periodically accelerates the electron distribution in peak c to the tunneling state .
through this effect , current flows to the collector side through the barrier . due to synchronization with this shaking motion ,
the current is enhanced periodically .
this tornado - like motion is typical for the small oscillation case . because there is a large contribution from steady current , the oscillation amplitude here is smaller than in the large oscillation case .
in summary , we investigated current oscillations in the plateau structures of the ndr region for three sizes of the contact regions with a model that includes damping , employing the caldeira - leggett hamiltonian .
we found two distinct types of current oscillations .
the first type is observed in the single plateau and in the lower part of the double - plateau structure .
it is characterized by a large oscillation amplitude and a single fourier component .
the other type is observed in the upper part of double - plateau structure .
it is characterized by a small oscillation amplitude and two fourier components .
an emitter basin that forms on the emitter side of the effective potential plays a key role in creating the current oscillation .
eigenstate analysis indicates that the first type is caused by transitions between the ground tunneling state and the adjacent excited state in the emitter basin , while the second type is caused by transitions between the intermediate tunneling state and higher states .
because the transition frequencies are large in the case of narrow emitter basin , there is high frequency oscillation in the case of small contact regions for a fixed basin depth . in wigner space ,
these two types of oscillation are characterized by the two types of motion : two - piston engine - like motion and tornado - like motion .
dissipation plays an important role in the realization of current oscillation . in order for the ground tunneling state to be populated in the case of large oscillation , there must be fairly strong dissipation , whose strength is determined by the system - bath coupling and the noise correlation time . if the dissipation is too strong , however , the current oscillation vanishes due to damping .
the key to have non - trivial behaviors such as hysteresis , single / double plateaus and self - excited current oscillations is on the existence of the resonant tunneling states , the charge redistribution effects and the dissipation .
the present results may be helpful to design nano - devices including a molecular junction system @xcite .
although many efforts have been made to improve wigner transportation theory @xcite , the quantum boltzmann equation @xcite , and other formalisms @xcite to study quantum dissipative dynamics in nano - devices , there are still a number of limitations and many subtle problems on such formalisms , which deserve further attention .
on the basis of the reduced hierarchy equations of motion ( heom ) approach , our investigation was carried out through highly accurate numerical calculations applied to a tunneling device system in a non - markovian environment at finite temperature .
we have provided evidence of intrinsic bistability and self - excited current oscillations in the ndr region rigorously .
while the current oscillation experimentally observed in the rtd induced by the resonance with external circuit@xcite , however , these effects are not accounted for in our approach like many other theories based on the boltzmann approach . to investigate the relation between the experimentally observed current oscillations and the existence of intrinsic current oscillation
, further investigation is necessary .
although the validity of the caldeira - leggett hamiltonian in the description of electron motion is not yet well established , we believe that the present results provide insight into the role of quantum mechanical phenomena in the type of system studied here . the present approach can be used to treat a strong system - bath coupling non - perturbatively . in addition , any time - dependent external field can be added while taking into account the system - bath quantum coherence through the hierarchy elements .
such features are ideal for studying squid rings @xcite and quantum ratchet systems @xcite .
yt is grateful for financial support received from the humboldt foundation and a grant - in - aid for scientific research ( b2235006 ) from the japan society for the promotion of science . as acknowledges the research fellowship of kyoto university
( a ) the effective potential ( black solid curve ) and electron density ( black dashed curve ) in the steady - state in the case of decreasing bias . here , the width of the contact region is 42.375 nm , and the bias voltage is 0.302 v. in contrast to the cases considered in figures 3 ( i - b)-(iii-b ) , in the case considered here , the emitter basin is very small .
the red and green curves represent the first and second eigenstates .
( b ) the steady - state wigner distribution .
the arrow indicates the direction of the steady current .
, width=491 ] to understand the origin of hysteresis in the ndr region , we plot the effective potential , electron density and energy eigenstates of the emitter basin and double - barrier well in the case of decreasing bias in figure a1 ( a ) . in this case
, the emitter basin is so shallow that there is no tunneling state between the emitter basin and the double - barrier well .
the existence of the peak near the barrier indicates that the second excited state is significantly populated .
the current arises through the transition from the second excited state to the ground quantum state through dissipation .
the wigner distribution is plotted in figure a1 ( b ) . due to dissipation
, the distribution is in a steady state .
the peak near the barrier arises because the barrier impedes the flow .
the electron density then leaks to the collector side without oscillation through tunneling in the form of steady current .
to see the effect of dissipation , we determined the i - v characteristics for the case of a stronger system - bath coupling , @xmath117 ghz ( @xmath118 ps ) .
the values of the other parameters are the same as in figure 2 .
in contrast to the case considered in figure 2 , in the present case current oscillation is not observed even in the plateau region .
also , the upper plateau does not exist .
this is because the current oscillation decays quickly through the damping . because the heat bath is strongly coupled to the system , the eigenstate picture of the electron system itself , as depicted in figure 3(b ) , is of questionable validity . as a result
, the plateau is lost .
the i - v characteristics for the case of strong system - bath coupling , @xmath117 ghz ( @xmath118 ps ) .
the values of the other parameters are the same as in figure 2 .
the black curve with the circles represents the case in which the bias is increasing , and the blue curve with the @xmath132s represents the case in which the bias is decreasing . comparing this figure with figure 2(a ) , it is seen that the size of the ndr region is smaller and the plateau structure is less pronounced here than in the weak - coupling case .
current oscillation is not observed even in the plateau region .
, width=188 ]
for the sake of comparison , we present the i - v characteristics calculated from the boltzmann equation and the poisson equation @xcite for the same physical conditions as in figure 2 .
the boltzmann equation commonly used in the rtd problem is expressed as @xcite @xmath138 where @xmath139 is the quantum liouvillian defined by equations ( [ liouville1 ] ) and ( [ liouville2 ] ) and @xmath140 is the modified collision operator under the relaxation time approximation . here , @xmath141 is the relaxation time , @xmath142 is the equilibrium wigner distribution function , @xmath143 is the density of the electron , and @xmath144 is that of the equilibrium distribution , respectively .
because the collision term is determined by the wigner distribution at time @xmath145 , @xmath146 , and does not depend upon the previous history of distribution , this equation describes markovian dynamics . because the boltzmann equation does not have a fluctuation term that is related to a dissipation term through the quantum version of the fluctuation - dissipation theorem ,
the equilibrium distribution is not an intrinsic state of the boltzmann equation .
moreover , we have to determine what the equilibrium distribution @xmath142 is in an _ ad hoc _ manner .
the i - v characteristics calculated from the boltzmann equation and poisson equation with @xmath147 ps .
the values of the other parameters are the same as in figure 2 .
the black curve with circles represents the case in which the bias is increasing , and the blue curve with the @xmath132s represents the case in which the bias is decreasing .
the wigner distributions at the voltage marked by the red circles in ( i ) are given in figure c2(a ) .
, width=188 ] we solve the boltzmann equation for the effective potential calculated from equation ( [ effectivepote ] ) and the poisson equation ( [ poisson ] ) following the same procedure as in sec . 4 with the same set of system parameters as in figure 2 . here
, the equilibrium distribution , @xmath142 , is obtained from the quantum liouville equation for the effective potential with the bias voltage zero @xcite .
the boundary conditions is given by @xcite @xmath148.\end{aligned}\ ] ] in the boltzmann equation approach , the time constant @xmath141 was estimated from other theory and buot et al . set it to be @xmath149 fs at @xmath150k @xcite . to compare with the heom result , here we solve the boltzmann equation at @xmath151 k for various @xmath141 to find the case that exhibits similar i - v profiles as in figures 2(i)-2(iii ) .
note that the difference between the fermi - dirac and bose - einstein distributions is minor at this temperature .
the obtained results for @xmath147 fs are presented in figure c1 . in figures
c1(i)-c1(iii ) , while we observed hysteresis , we could not find any plateau - like behavior and current oscillation in the ndr region as was shown in the heom calculations in figures 2(i)-2(iii ) .
the present results are also different from the result obtained from the boltzmann equation at @xmath150k for @xmath149 fs , in which a single plateau behavior and current oscillations in the ndr region were observed @xcite . the steady - state wigner distribution for ( i ) the increasing bias case and ( ii ) the decreasing bias case calculated from ( a ) the boltzmann equation and ( b ) the heom depicted in figure c1(i ) and 2 , respectively . the bias voltage of ( i - a ) and ( ii - a ) is 0.32 v ( marked by the red circles in figure c1(i ) ) , while that in ( a - ii ) and ( b - ii ) is 0.30 v. , width=453 ] to analyze the difference between the boltzmann and heom results , we display snapshots of the steady - state wigner distribution near the maximum and minimum of the i - v curves in figures c2(i ) and 2(ii ) . we find that the wigner distribution in the collector region is smooth in the heom case , while there are many small peaks disturbing the flow in the boltzmann case .
this difference arises because the equilibrium state in the boltzmann approach is fixed even when the effective potential is changed from the original one due to the self - consistent calculations . as the result , the difference between the imposed equilibrium state and the true equilibrium state becomes large especially in the collector region , where the distribution is far from the assumed equilibrium distribution , @xmath142 , due to the scattering from the potential barriers . as indicated in this appendix , dynamics described by the boltzmann equation approach is different from the heom approach .
this difference arises because the thermal equilibrium state of the boltzmann equation is introduced as an assumption , while the thermal equilibrium state of the heom approach is an intrinsic state of the equation that is determined through the balance between the fluctuation term and dissipation term .
this difference becomes significant for a system that exhibits hysteresis , since the equilibrium state of the system depends upon the pathway of process .
this difference may also be significant if the system is driven by a time - dependent external field , in which the equilibrium state is not well - defined .
mukamel s 1995 _ principles of nonlinear optical spectroscopy _
( new york ; oxford university press ) tanimura y and ishizaki a 2009 _ acc .
res . _ * 42 * 1270 tanimura y and mukamel s 1993 _ j. chem .
phys . _ * 99 * 9496 tanimura y and mukamel s 1993 _ phys . rev .
_ e * 47 * 118 tanimura y and okumura k 1997 _ j. chem .
_ * 106 * 2078 datta s 1995 _ electronic transport in mesoscopic systems _ ( new york ; cambridge university press ) ferry d k , goodnick s m and bird j 2009 _ transport in nanostructures _ ( new york ; cambridge university press , 2nd ed ) kadanoff l p and baym g 1962 _ quantum statistical mechanics _
( new york ; benjamin ) mahan g d 2000 _ many - particle physics _ ( berlin ; springer ) haug h and jauho a p 2008 _ quantum kinetics in transport and optics of semiconductors _
( new york ; springer - verlag ) caldeira a o and leggett a j 1983 _ physica _ a * 121 * 587 chang l - d and waxman d 1985 _ j. phys .
_ c * 18 * 5873 cederbaum l s , gindensperger i and burghardt i 2005 _ phys .
lett . _ * 94 * 113003 hughes k h , christ c d and burghardt i 2009 _ j. chem .
phys . _ * 131 * 024109 chin a w , rivas a , huelga s f and plenio m b 2010 _ j. math
* 51 * 092109 chin a w , rivas a , prior j , rosenbach r , caycedo - soler f , huelga s f and plenio m b 2013 _ nature physics _
doi:10.1038/nphys2515 garcia - palacios j l and zueco d 2004 _ j. phys . a math .
gen . _ * 37 * 10735 coffey w t , kalmykov yu p , titov s v and mulligan b p 2007 _ phys .
_ e * 75 * 041117 coffey w t , kalmykov yu p , titov s v and mulligan b p 2007 _ phys .
_ * 9 * 3361 shit a , chattopadhyay s and chaudhuri j r 2011 _ chem .
_ * 386 * 56 tanimura y 2006 _ j. phys .
soc . jpn .
_ * 75 * 082001 tanimura y and kubo r 1989 _ j. phys
jpn . _ * 58 * 101 tanimura y and wolynes p g 1991 _ phys . rev . _ a * 43 * 4131 tanimura y and wolynes p g 1992 _ j. chem .
_ * 96 * 8485 ishizaki a and tanimura y 2005 _ j. phys .
_ * 74 * 3131 tanimura y and maruyama y 1997 _ j. chem .
phys . _ * 107 * 1779 tanimura y and steffen t 2000 _ j. phys .
jpn . _ * 69 * 4095 kato t and tanimura y 2002 _ j. chem .
phys . _ * 117 * 6221 kato t and tanimura y 2004 _ j. chem
. phys . _ * 120 * 260 ishizaki a and tanimura y 2007 _ j. phys .
chem . _ a * 111 * 9269 sakurai a and tanimura y 2011 _ j. phys . chem . _ a * 115 * 4009 kato a and tanimura y 2013 _ j. phys .
_ b * 117 * 13132 chang l l , esaki l and tsu r 1974 _ appl .
* 24 * 593 sollner t c l g , tannenwald p e , peck d d and goodhue w d 1984 _ appl .
lett . _ * 45 * 1319 slight t j , romeria b , wang l , figueiredo j
m l , wasige e and ironside c n 2008 _ ieee j. quantum electron .
_ * 44 * 1158 feiginov m , sydlo c , cojocari o and meissner p 2011 _ appl .
lett . _ * 99 * 233506 feiginov m , sydlo c , cojocari o and meissner p 2012 _ eur . phys
. lett . _ * 97 * 58006 asada m , suzuki s and kishimoto n 2008 _ jpn .
_ * 47 * 4375 suzuki s , asada m , teranishi a. sugiyama h and yokoyama h 2010 _ appl .
lett . _ * 97 * 242102 kanaya h , shibayama h , sogabe r , suzuki s and asada m 2012 _ appl . phys . exp . _ * 5 * 124101 kluksdahl n c , kriman a m and ferry d k 1989 _ phys .
_ b * 39 * 7720 jensen k l and buot f a 1991 _ phys .
_ * 66 * 1078 jansen k l and buot f a 1991 _ ieee trans .
_ * 38 * 2237 biegel b a and plummer j d 1996 _ phys .
_ b * 54 * 8070 zhao p , cui h l , woolard d , jansen k l and buot f a 2000 _ j. appl .
* 87 * 1337 yang s , cao j and field r w 2004 _ j. chem .
phys . _ * 121 * 6599 yoder d p , grupen m and smith r k 2010 _ ieee trans .
electron devices _ * 57 * 3265 chen l , zheng r h , shi q and yan y - j 2009 _ j. chem .
phys . _ * 131 * 094502 chen l , zheng r q. shi and yan y - j 2010 _ j. chem .
phys . _ * 132 * 024505
jiang f , jin j , wang s and yan y - j 2012 _ phys .
_ b * 85 * 245427 ishizaki a and fleming g r 2009 _ proc .
* 17255 chen l , zheng r , jing y and shi q 2011 _ j. phys .
chem . _ * 134 * 194508 kreisbeck c and kramer t 2012 _ j. phys .
lett . _ * 3 * 2828 strmpfer j and schulten k 2012 _ j. phys .
chem . _ * 137 * 065101 dijkstra a g and tanimura y 2010 _ new j. phys . _
* 12 * 055005 dijkstra a g and tanimura y 2012 _ new j. phys . _
* 14 * 073027 xu r - x , chen y , cui p , ke h w and yan y - j 2007 _ j. phys .
a * 111 * 9618 tanaka m and tanimura y 2009 _ j. phys .
* 78 * 073802 tanaka m and tanimura y 2010 _ j. chem .
phys . _ * 132 * 214502 tanimura y 2012 _ j. chem .
phys . _ * 137 * 22a550 shi q ,
chen l p , nan g j , xu r x and yan y - j 2009 _ j. chem .
phys . _ * 130 * 084105 hu j , xu r x and yan y - j 2010 _ j. chem .
* 133 * 101106 tian b l , ding j j , xu r x and yan y - j 2010 _ j. chem .
phys . _ * 133 * 114112 kreisbeck c , kramer t , rodriguez m , and hein b 2011 _ j. chem . theor . comp . _ * 7 * 2166 zhu j , kais s , rebentros p and aspuru - guzik a 2011 _ j. phys . chem . _
b * 115 * 1531 strmpfer j and schulten k 2012 _ j. chem .
* 8 * 2808 sahrapour m m and makri n 2010 _ j. chem .
phys . _ * 132 * 134506 nalbach p , ishizaki a , fleming g r and thorwart m , 2011 _ n. j. phys . _
* 13 * 063040 ishizaki a and tanimura y , 2008 _ chem .
_ * 347 * 185 ravaioli u , osman m a , ptz w , kluksdahl n and ferry d k 1985 _ physica _ b * 134 * 36 jacoboni c and bordone p 2004 _ rep .
phys . _ * 67 * 1033 jiang h , cai w and tsu r 2011 _ j. comp . phys . _ * 230 * 2031 sakurai a and tanimura y 2013 _ j. phys .
jpn . _ * 82 * 033707 vacchini b and hornberger k 2009 _ phys .
* 71 hornberger k and vacchini b 2008 _ phys . rev . _ a * 77 * 022112 feiginov m , sydlo c , cojocari o and meissner p 2011 _ appl .
lett . _ * 99 * 133501 knezevic i and novakovic b 2013 _ j. comput .
_ * 12 * 363 | the quantum dissipative dynamics of a tunneling process through double barrier structures is investigated on the basis of non - perturbative and non - markovian treatment .
we employ a caldeira - leggett hamiltonian with an effective potential calculated self - consistently , accounting for the electron distribution . with this hamiltonian ,
we use the reduced hierarchy equations of motion in the wigner space representation to study non - markovian and non - perturbative thermal effects at finite temperature in a rigorous manner .
we study current variation in time and the current - voltage ( i - v ) relation of the resonant tunneling diode for several widths of the contact region , which consists of doped gaas . hysteresis and both single and double plateau - like behavior are observed in the negative differential resistance ( ndr ) region . while all of the current oscillations decay in time in the ndr region in the case of a strong system - bath coupling , there exist self - excited high - frequency current oscillations in some parts of the plateau in the ndr region in the case of weak coupling .
we find that the effective potential in the oscillating case possesses a basin - like form on the emitter side ( emitter basin ) and that the current oscillation results from tunneling between the emitter basin and the quantum well in the barriers .
we find two distinct types of current oscillations , with large and small oscillation amplitudes , respectively .
these two types of oscillation appear differently in the wigner space , with one exhibiting tornado - like motion and the other exhibiting a two piston engine - like motion . |
the cauchy problem for the camassa holm ( ch ) equation @xcite , @xmath10 where @xmath11 is a constant , has attracted much attention due to the fact that it serves as a model for shallow water waves @xcite and its rich mathematical structure .
for example , it has a bi - hamiltonian structure , infinitely many conserved quantities and blow - up phenomena have been studied , see , e.g. , @xcite , @xcite , and @xcite .
we here focus on the construction of the lipschitz metric for the semigroup of conservative solutions on the real line .
this problem has been recently considered by grunert , holden , and raynaud @xcite in the periodic case , and here we want to present how the approach used there has to be modified in the non - periodic case . for simplicity , we will only discuss the case @xmath12 , that is , @xmath13 and from now on we refer to as the ch equation .
however , the approach presented here can also handle the generalized hyperelastic - rod equation , see remark [ rem : genhyp ] .
in particular , it includes the case with nonzero @xmath14 .
the generalized hyperelastic - rod equation has been introduced in @xcite .
it is given by @xmath15 where @xmath9 and @xmath16 are smooth functions .
is assumed to be strictly convex or concave . ] with @xmath17 and @xmath18 , we recover for any @xmath14 . with @xmath19 and @xmath20
, we obtain the hyperelastic - rod wave equation : @xmath21 which has been introduced by dai @xcite .
equation can be rewritten as the following system @xmath22 where we choose @xmath3 to be an element of @xmath23 .
the @xmath6 norm is preserved as @xmath24 for any smooth solution @xmath3 . however , even for smooth initial data , the solution may break down in finite time . in this case , the solution experiences wave breaking ( @xcite ) : the solution remains bounded while , at some point , the spatial derivative @xmath25 tends to @xmath26 .
this phenomenon can be nicely illustrated by the so called multipeakon solutions .
these are solutions of the form @xmath27 let us consider the case with @xmath28 and one peakon @xmath29 ( moving to the right ) and one antipeakon @xmath30 ( moving to the left ) . in the symmetric case ( @xmath31 and @xmath32 ) the solution @xmath3 will vanish pointwise at the collision time @xmath33 when @xmath34 , that is , @xmath35 for all @xmath36 . at time @xmath37 , the whole energy is concentrated at the origin , and we have @xmath38 , with @xmath39 denoting the dirac delta distribution at the origin . in general
we have two possibilities to continue the solution beyond wave breaking , namely to set @xmath3 identically equal to zero for @xmath40 , which is called a dissipative solution , or to let the peakons pass through each other , which is called a conservative solution and which is depicted in figure [ fig : peakcol ] .
we are interested in the latter case , for which solutions have been studied by bressan and constantin @xcite and holden and raynaud @xcite
. since the @xmath6-norm is preserved , the space @xmath6 appears as a natural space for the semigroup of solutions .
however , the previous multipeakon example reveals the opposite .
indeed , @xmath35 for all @xmath41 .
thus , the trivial solution @xmath42 , that is , @xmath43 for all @xmath44 , which is also a conservative solution , coincides with @xmath3 at @xmath37 . to define a semigroup of conservative solutions ,
we therefore need more information about the solution than just its pointwise values , for instance , the amount and location of the energy which concentrates on sets of zero measure .
this justifies the introduction of the set @xmath45 of eulerian coordinates , see definition [ def : d ] , for which a semigroup can be constructed @xcite . furthermore , the @xmath6 norm is not well suited to establish a stability result .
consider , e.g. , the sequence of multipeakons @xmath46 defined as @xmath47 , see figure [ fig : peakcol ] .
then , assuming that @xmath48 , we have @xmath49 so that the flow is clearly discontinuous with respect to the @xmath6 norm . , which vanishes at @xmath33 , for @xmath50 ( left ) and
@xmath37 ( middle ) and @xmath51 ( right ) .
the solid curve depicts the multipeakon solution given by @xmath52 for some small @xmath53 ( the ch equation is invariant with respect to time translation).,title="fig:",width=151 ] , which vanishes at @xmath33 , for @xmath50 ( left ) and @xmath37 ( middle ) and @xmath51 ( right ) .
the solid curve depicts the multipeakon solution given by @xmath52 for some small @xmath53 ( the ch equation is invariant with respect to time translation).,title="fig:",width=151 ] , which vanishes at @xmath33 , for @xmath50 ( left ) and @xmath37 ( middle ) and @xmath51 ( right ) .
the solid curve depicts the multipeakon solution given by @xmath52 for some small @xmath53 ( the ch equation is invariant with respect to time translation).,title="fig:",width=151 ] [ fig : peakcol ] the aim of this article is to present a metric for which the semigroup of conservative solutions on the line is lipschitz continuous . a more extensive discussion about lipschitz continuity with examples from ordinary differential equations ,
can be found in @xcite .
a detailed presentation for the camassa
holm equation in the periodic case is presented in @xcite , thus we here focus on explaining the differences between the periodic case and the decaying case .
however , we first present the general construction .
the construction of the metric is closely connected to the construction of the semigroup itself .
let us outline this construction .
we rewrite the ch equation in lagrangian coordinates and obtain a semilinear system of ordinary differential equations : let @xmath54 denote the solution and @xmath55 the corresponding characteristics , thus @xmath56 .
then our new variables are @xmath55 , as well as @xmath57 where @xmath58 corresponds to the lagrangian velocity while @xmath59 can be interpreted as the lagrangian cumulative energy distribution .
the time evolution for any @xmath60 is described by @xmath61 where @xmath62 and @xmath63 this system is well - posed as a locally lipschitz system of ordinary differential equations in a banach space , and we can define a semigroup of solution which we denote @xmath64 . from standard theory for ordinary differential equations
we know that @xmath64 is locally lipschitz continuous , that is , given @xmath65 and @xmath66 , @xmath67 for any @xmath68 $ ] , @xmath69 depends only on @xmath66 and @xmath65 .
the mapping from lagrangian to eulerian coordinates is surjective but not bijective .
the discrepancy between the two sets of coordinates is due to the freedom of relabeling in lagrangian coordinates .
the relabeling functions form a group , which we denote @xmath70 , and which basically consists of the diffeomorphisms of the line with some additional assumptions ( see definition [ def : gr ] ) .
given @xmath60 , the element @xmath71 is the relabeled version of @xmath72 by the relabeling function @xmath73 . using the fact that the semigroup @xmath64 is equivariant with respect to relabeling ,
that is , @xmath74 we can construct a semigroup of solutions on equivalence classes from @xmath64 . finally ,
after establishing the existence of a bijection between the eulerian coordinates and the equivalence classes in lagrangian coordinates , we can transport the semigroup of solutions defined on equivalence classes and construct a semigroup , which we denote @xmath75 , of conservative solutions in @xmath45 .
we want to find a metric which makes @xmath75 lipschitz continuous . for that purpose
, we introduce a pseudometric which is symmetric , @xmath76 , for which the triangle inequality @xmath77 holds , and satisfies @xmath78 for @xmath79 . ] in lagrangian coordinates which does not distinguish between elements of the same equivalence class and which , at the same time , leaves the semigroup @xmath64 locally lipschitz continuous .
this strategy has been used in @xcite for the hunter saxton equation and in @xcite for the camassa
holm equation in the periodic case . in @xcite ,
the pseudometric is defined by using ideas from riemannian geometry . here , we follow the approach of @xcite and first
introduce a pseudosemimetric which is symmetric , @xmath76 and satisfies @xmath78 for @xmath80 . ] which also identifies elements of the same equivalence class and leaves @xmath64 lipschitz continuous .
a natural choice , which was applied in @xcite , is to consider the pseudometric @xmath81 defined as @xmath82 the pseudometric @xmath81 identifies elements of the same equivalence class , as @xmath83 .
moreover , it is invariant with respect to relabeling , that is , @xmath84 for any @xmath85 .
it remains to prove that the pseudosemimetric makes the semigroup @xmath64 locally lipschitz , that is , given @xmath66 and @xmath65 , there exists a constant @xmath86 depending on @xmath66 and @xmath65 such that @xmath87 for all @xmath68 $ ] and @xmath88 .
the proof follows almost directly from the stability and equivariance of @xmath64 .
we outline it here . for every @xmath53
, there exist @xmath85 such that @xmath89 and we get [ eq : notusstab ] @xmath90 and follows by letting @xmath91 tend to zero .
however , the use of the lipschitz stability of @xmath64 relies on bounds on @xmath92 and @xmath93 that are unavailable .
the problem is that the norm @xmath94 of the banach space is _ not _ invariant with respect to relabeling and therefore , since @xmath9 and @xmath16 are a priori arbitrary , we can not obtain any bound depending on @xmath66 for @xmath92 and @xmath93 .
this motivates the introduction in this paper of the pseudosemimetric @xmath95 defined as @xmath96 as expected , the pseudosemimetric @xmath95 identifies equivalence classes ( we have @xmath97 ) but we lose the nice relabeling invariance property . at the same time
, this definition of @xmath95 implies some implicit restrictions on the diffeomorphisms @xmath98 and @xmath99 which allow us to bound the relabeled versions @xmath100 and @xmath101 so that the approach sketched in can be carried out .
it remains to explain why , in the periodic case @xcite , we could use the definition of @xmath102 , which is a more natural definition and moreover simplifies the proofs . in the periodic case
( we take the period equal to one ) , the stability of the semigroup @xmath64 is established in the space @xmath103)$ ] equipped with the norm @xmath104)}={\left\vertu\right\vert}_{l^\infty([0,1])}+{\left\vertu_\xi\right\vert}_{l^1([0,1])}.\ ] ] note that , in order to keep these formal explanations as simple as possible , we just consider the second component of @xmath60 . since @xmath105 and @xmath106 we have @xmath107 , for any @xmath73 , so that the norm defined in is relabeling invariant . now ,
if the norm of the banach space is relabeling invariant , we have @xmath108 and the pseudosemimetrics @xmath95 and @xmath81 are equivalent . however , the natural banach space for @xmath58 is not @xmath103)$ ] but @xmath109)=h^1([0,1])$ ] . in the periodic case ,
it is not an issue as @xmath110)\subset w^{1,1}([0,1])$ ] but the corresponding embedding does not hold in the case of the real line .
this also shows that the approach that we present here for the real line can also be used in the periodic case and that the novelty in this article is that we handle a norm which is not relabeling invariant .
the final step consists of deriving a pseudometric from the pseudosemimetric @xmath95 .
this can be achieved by the following general construction : let @xmath111 where the infimum is taken over all finite sequences @xmath112 with @xmath113 and @xmath114 .
the pseudometric @xmath115 inherits the lipschitz stability property from @xmath95 .
finally , identifying elements belonging to the same equivalence class , the pseudometric @xmath115 turns into a metric on the set of equivalence classes . by bijection ,
it yields a metric in @xmath45 which makes the semigroup of conservative solutions lipschitz continuous . in the last section , section [ sec : top ]
, we compare this new metric with the usual norms in @xmath6 and @xmath7 .
in this section , we recall from @xcite the construction of the semigroup in lagrangian coordinates . the camassa
holm equation reads @xmath116 and can be rewritten as the following system @xmath117 @xmath118 next , we rewrite the equation in lagrangian coordinates .
therefore we introduce the characteristics @xmath119 the lagrangian velocity @xmath58 reads @xmath120 we define the lagrangian cumulative energy as @xmath121 as an immediate consequence of the definition of the characteristics we obtain @xmath122 the last term can be expressed uniquely in terms of @xmath123 , @xmath58 , and @xmath59 . from we
obtain the following explicit expression for @xmath124 , @xmath125 setting @xmath126 and writing @xmath127 , we obtain @xmath128 and @xmath129 moreover we introduce another variable @xmath130 .
thus we have derived a new system of equations , which is up to that point only formally equivalent to the camassa
holm equation : @xmath131 let @xmath132 be the banach space defined by @xmath133 where @xmath134 and the norm of @xmath132 is given by @xmath135 .
of course @xmath136 but the converse is not true as @xmath132 contains functions that do not vanish at infinity .
we will employ the banach space @xmath137 defined by @xmath138 with the following norm @xmath139 for any @xmath140 .
[ def : f ] the set @xmath141 is composed of all @xmath142 such that [ eq : lagcoord ] @xmath143 ^ 3,\\ \label{eq : lagcoord2 } & y_\xi\geq0 , h_\xi\geq0 , y_\xi+h_\xi>0 \text { almost everywhere , and } \lim_{\xi\rightarrow-\infty}h(\xi)=0,\\ \label{eq : lagcoord3 } & y_\xi h_\xi = y_\xi^2u^2+u_\xi^2\text { almost everywhere},\end{aligned}\ ] ] where we denote @xmath144 . given a constant @xmath145 , we denote by @xmath146 the ball @xmath147 [ th : global ] for any @xmath148 , the system admits a unique global solution @xmath149 in @xmath150 with initial data @xmath151 .
we have @xmath152 for all times .
if we equip @xmath141 with the topology induced by the @xmath137-norm , then the mapping @xmath153 defined by @xmath154 is a continuous semigroup .
more precisely , given @xmath145 and @xmath155 , there exists a constant @xmath156 which depends only on @xmath66 and @xmath65 such that , for any two elements @xmath157 , we have @xmath158 for any @xmath68 $ ] .
we have @xmath159 where @xmath160 is locally lipschitz ( see proof ( * ? ? ?
* theorem 2.3 ) ) .
it implies , by gronwall s lemma , that we have @xmath161 for @xmath68 $ ] , where the constant @xmath86 only depends on @xmath162}{\left\verts_t(x_\alpha)\right\vert}$ ] and @xmath162}{\left\verts_t(x_\beta)\right\vert}$ ] . in (
* theorem 2.8 ) it is proved that @xmath162}{\left\verts_t(x_\alpha)\right\vert}$ ] only depends of @xmath163 and @xmath65 , and thus on @xmath66 and @xmath65 .
[ def : gr ] we denote by @xmath70 the subgroup of the group of homeomorphisms from @xmath164 to @xmath164 such that [ eq : hcond ] @xmath165 where @xmath166 denotes the identity function . given @xmath167 , we denote by @xmath168 the subset @xmath168 of @xmath70 defined by @xmath169 the subsets @xmath168 do not possess the group structure of @xmath170 .
the next lemma provides a useful characterization of @xmath168 .
[ lem : charh ] let @xmath171 .
if @xmath9 belongs to @xmath168 , then @xmath172 almost everywhere .
conversely , if @xmath9 is absolutely continuous , @xmath173 , @xmath9 satisfies and there exists @xmath174 such that @xmath175 almost everywhere , then @xmath176 for some @xmath14 depending only on @xmath177 and @xmath178 .
we define the subsets @xmath179 and @xmath180 of @xmath141 as follows @xmath181 and @xmath182 for @xmath12 , @xmath183 .
as we shall see , the space @xmath184 will play a special role .
these sets are relevant only because they are preserved by the governing equation as the next lemma shows .
in particular , while the mapping @xmath185 may not be a diffeomorphism for some time @xmath186 , the mapping @xmath187 remains a diffeomorphism for all times @xmath186 .
[ lem : fpres ] the space @xmath180 is preserved by the governing equation .
more precisely , given @xmath188 , there exists @xmath189 which only depends on @xmath65 , @xmath14 and @xmath190 such that @xmath191 for any @xmath192 .
let @xmath193 , we denote by @xmath149 the solution of with initial data @xmath194 and set @xmath195 , @xmath196 . by definition , we have @xmath197 and , from lemma [ lem : charh ] , @xmath198 almost everywhere , for some constant @xmath199 depending only on @xmath14 .
we consider a fixed @xmath200 and drop it in the notation .
applying gronwall s inequality backward in time to we obtain @xmath201 for some constant @xmath86 which depends on @xmath202,e)}$ ] , which itself depends only on @xmath190 and @xmath65 . from , we have @xmath203 hence , since @xmath204 and @xmath205 are positive , gives us @xmath206 and @xmath207 .
similarly , by applying gronwall s lemma forward in time , we obtain @xmath208 .
we have @xmath209,e)}\leq c$ ] and @xmath210 for another constant @xmath86 which only depends on @xmath190 and @xmath65 .
hence , applying lemma [ lem : charh ] , we obtain that @xmath211 and therefore @xmath212 for some @xmath213 depending only on @xmath14 , @xmath65 , and @xmath190 . for the sake of simplicity , for any @xmath214 and any function @xmath73
, we denote @xmath215 by @xmath216 .
this operation corresponds to relabeling .
we denote by @xmath217 the projection of @xmath180 into @xmath184 defined as @xmath218 for any @xmath214 .
the element @xmath217 is the unique relabeled version of @xmath72 that belongs to @xmath184 .
[ lem : equivpi ] the mapping @xmath64 is equivariant , that is , @xmath219 this follows from the governing equation and the equivariance of the mappings @xmath220 and @xmath221 , where @xmath124 and @xmath222 are defined in and , see @xcite for more details . from this lemma
we get that @xmath223 we define the semigroup @xmath224 on @xmath184 as @xmath225 the semigroup property of @xmath224 follows from . from @xcite , we know that @xmath224 is continuous with respect to the norm of @xmath137 .
it follows basically from the continuity of the mapping @xmath226 , but @xmath226 is not lipschitz continuous and the goal of the next section is to find a metric that makes @xmath224 lipschitz continuous .
[ rem : genhyp ] the details of the construction of the semigroup of solutions in lagrangian coordinates for the generalized hyperelastic - rod equation is given in @xcite .
the construction is based on a reformulation of the equation in lagrangian coordinates which leads to a semilinear system of equations , similar to for the camassa
holm equation . in the case of the generalized hyperelastic - rod equation ,
the equation can be rewritten as @xmath227 where @xmath228 is given by @xmath229 and @xmath230 @xmath231 section [ sec : semilag ] outlines the construction of the semigroup of solutions in lagrangian coordinates .
the construction of the metric in eulerian coordinates , which is given in the following sections , relies basically on two fundamental results of this section : the lipschitz stability of the semigroup of solution in lagrangian coordinates ( theorem [ th : global ] ) and the equivariance of the semigroup ( lemma [ lem : equivpi ] ) .
the same results hold for the generalized hyperelastic - rod equation , see ( * ? ? ? * theorem 2.8 and theorem 3.6 ) so that it is possible to define a lipschitz stable metric for this equation in the same way as we do it for the ch equation .
[ def : j ] let @xmath232 , we define @xmath233 as @xmath234 the mapping @xmath95 is symmetric .
moreover , if @xmath235 and @xmath236 are equivalent , then @xmath237 .
our goal is to create a distance between equivalence classes , and that is the reason why we introduce the pseudosemimetric @xmath81 as follows in the periodic case ( @xcite ) .
[ def : jt ] let @xmath232 , we define @xmath238 as @xmath239 the pseudosemimetric @xmath81 is relabeling invariant , that is , @xmath240 . with definition [ def : j
] , we lose this important property .
however , definition [ eq : defj ] allows us to obtain estimates that can not be obtained by definition [ def : jt ] , see the proof of theorem [ th : stab ] .
in addition , it turns out that we do not actually need the relabeling invariance property to hold strictly and the estimates contained in the following lemma are enough for our purpose .
[ lem : jrelabbound ] given @xmath241 and @xmath176 , we have @xmath242 so that @xmath243 for some constant @xmath86 which depends only on @xmath14 .
let us prove .
let @xmath244 and @xmath245 .
we have @xmath246 and @xmath247 so that @xmath248 . hence , @xmath249 . by definition
we have @xmath250 and hence @xmath251 .
thus @xmath252 so that @xmath253 .
we have @xmath254 this concludes the proof of . for any @xmath176 and any @xmath255
, we have @xmath256 hence , after taking @xmath257 , @xmath258 which implies , after taking the infimum , @xmath259 from the pseudosemimetric @xmath95 , we obtain a metric @xmath115 by the following construction .
let @xmath260 , we define @xmath261 as @xmath262 where the infimum is taken over all sequences @xmath263 which satisfy @xmath113 and @xmath114 .
[ lem : linfbdj ] for any @xmath260 , we have @xmath264 first , we prove that , for any @xmath260 , we have @xmath265 we have @xmath266 it follows from the definition of @xmath184 that @xmath267 , @xmath268 and @xmath269 so that @xmath270 .
we also have @xmath271 hence , from , we get @xmath272 in the same way , we obtain @xmath273 for any @xmath73 . after adding these two last inequalities and taking the infimum
, we get . for any @xmath53
, we consider a sequence @xmath263 such that @xmath113 and @xmath114 and @xmath274 .
we have @xmath275 after letting @xmath91 tend to zero , we get .
the mapping @xmath276 is a distance on @xmath184 , which is bounded as follows @xmath277 the symmetry is embedded in the definition of @xmath95 while the construction of @xmath115 from @xmath95 takes care of the triangle inequality . from lemma [ lem : linfbdj ] , we get that @xmath278 implies @xmath279 .
the first inequality in follows from lemma [ lem : linfbdj ] while the second one follows from the definition of @xmath95 and @xmath115 .
indeed , we have @xmath280 we need to introduce the subsets of bounded energy in @xmath184 .
note that the total energy is equal to @xmath281 as @xmath282 and @xmath59 is increasing , see definition [ def : f ] .
we denote by @xmath283 the set @xmath284 and @xmath285 the ball @xmath146 ( see ) is not preserved by the equation while the set @xmath283 is preserved because of the conservation of energy , namely , @xmath286 the set @xmath283 is also conserved by relabeling as , for any @xmath73 , @xmath287 .
the ball @xmath146 is included in @xmath283 but the reverse inclusion does not hold . however , as the next lemma shows , when we restrict ourselves to @xmath184 , the sets @xmath288 and @xmath289 are in fact equivalent .
[ lem : bdef0 m ] for any element @xmath290 , we have @xmath291 for some constant @xmath292 depending only on @xmath66 .
since @xmath293 , @xmath294 , @xmath295 , we get @xmath296 and @xmath297 . hence , @xmath298 .
since @xmath299 , we get @xmath300 . by ,
we get @xmath301 and therefore @xmath302 . finally , we have to show that @xmath303 .
therefore observe that by , @xmath304 .
this together with the fact that @xmath305 yields @xmath306 thus it is left to estimate @xmath307 , which can be done as follows , @xmath308 where we used that @xmath309 , when @xmath310 by . for almost every @xmath200 such that @xmath311 , we have @xmath312 from and hence @xmath313 and @xmath314 . [ def : metric ]
let @xmath315 be the distance on @xmath316 which is defined , for any @xmath317 , as @xmath318 where the infimum is taken over all the sequences @xmath319 which satisfy @xmath113 and @xmath114 .
we can now prove our main stability theorem .
[ th : stab ] given @xmath155 and @xmath145 , there exists a constant @xmath156 which depends only on @xmath66 and @xmath65 such that , for any @xmath317 and @xmath68 $ ] , we have @xmath320 by the definition of @xmath315 , for any @xmath91 such that @xmath321 there exists a sequence @xmath112 in @xmath316 such that @xmath113 , @xmath114 , @xmath322 hence , there exist functions @xmath323 , @xmath324 in @xmath70 such that @xmath325 let us denote @xmath326 by lemma [ lem : fpres ] , we have @xmath327 for some @xmath14 which depends only on @xmath66 and @xmath65 .
the sequence @xmath328 has endpoints given by @xmath329 and @xmath330 .
since @xmath331 and the set @xmath283 is preserved by the flow of the equation and relabeling , we have @xmath332 so that the sequence @xmath328 is in @xmath316 , as required in the definition of @xmath315 . for @xmath333
, we have @xmath334 to use the stability result , we have to bound @xmath335 and @xmath336 . by lemma [ lem : bdef0 m ]
, there exists @xmath292 such that @xmath337 for any @xmath290 .
hence , @xmath338 as @xmath339
. since @xmath340 is a priori arbitrary , it may seem difficult to bound @xmath341 , and it is important to note here that the relabeling invariant pseudosemimetric @xmath81 , see , would not provide us with a bound on this term and the following estimates in fact motivate the definition [ eq : defj ] .
indeed , by , we obtain which yields @xmath342 as @xmath317 .
therefore , by the triangle inequality , @xmath343 so that @xmath344 and @xmath345 are bounded by a constant depending only on @xmath66 .
thus , we can use and get from that @xmath346 where from now on @xmath156 denotes some constant dependent on @xmath66 and @xmath65 .
similarly for @xmath347 , we get that @xmath348 finally , we have @xmath349 the result follows by letting @xmath91 tend to zero .
we now introduce a second set of coordinates , the so called eulerian coordinates .
therefore let us first consider @xmath214 .
we can define eulerian coordinates as in @xcite and also obtain the same mappings between eulerian and lagrangian coordinates ( see also figure [ fig : explfig ] ) . for completeness we will state the results here . where the lagrangian variables are defined is represented by the interior of the closed domain on the left .
the equivalence classes @xmath350 $ ] and @xmath351 $ ] ( with respect to the action of the relabeling group @xmath170 ) of @xmath72 and @xmath352 , respectively , are represented by the horizontal curves . to each equivalence class
there corresponds a unique element in @xmath184 and @xmath45 ( the set of eulerian variables ) .
the sets @xmath184 and @xmath45 are represented by the vertical curves.,width=377 ] [ def : d ] the set @xmath45 consists of all pairs @xmath353 such that 1 .
@xmath354 , and 2 .
@xmath355 is a positive radon measure whose absolutely continuous part @xmath356 satisfies @xmath357 we can define a mapping , denoted by @xmath358 , from @xmath45 to @xmath184 : for any @xmath353 in @xmath45 let , @xmath359 then @xmath360 , and we denote by @xmath361 the map which to any @xmath353 associates @xmath305 .
thus from any initial data @xmath362 , we can construct a solution of in @xmath180 with initial data @xmath363 .
it remains to go back to the original variables , which is the purpose of the mapping @xmath66 defined as follows : given any element @xmath364 in @xmath184 , then @xmath353 defined as follows @xmath365 @xmath366 belongs to @xmath45 .
we denote by @xmath367 the map which to any @xmath72 in @xmath184 associates @xmath353 .
in fact , @xmath66 can be seen as a map from @xmath368 , as any two elements belonging to the same equivalence class in @xmath180 are mapped to the same element in @xmath45 ( cf .
moreover , identifying elements belonging to the same equivalence class , the mappings @xmath358 and @xmath66 are invertible and @xmath369 we will now use these mappings for defining also a lipschitz metric on @xmath45 . let @xmath370 next we show that @xmath75 is a lipschitz continuous semigroup by introducing a metric on @xmath45 . using the map @xmath358 we can transport the topology from @xmath184 to @xmath45 .
define the metric @xmath371 by @xmath372 the lipschitz stability of the semigroup @xmath75 follows then naturally from theorem [ th : stab ] .
it holds on sets of bounded energy , which are given as follows .
[ def : dm ] given @xmath145 , we define the subsets @xmath373 of @xmath45 , which correspond to sets of bounded energy , as @xmath374 on the set @xmath373 we define the metric @xmath375 as @xmath376 where the metric @xmath315 is defined as in definition [ def : metric ]
. definition [ def : dm ] is well - posed as we can check from the definition of @xmath358 : if @xmath377 , then @xmath378 .
the semigroup @xmath379 is a continuous semigroup on @xmath45 with respect to the metric @xmath380 .
the semigroup is lipschitz continuous on sets of bounded energy , that is : given @xmath145 and a time interval @xmath381 $ ] , there exists a constant @xmath156 , which only depends on @xmath66 and @xmath65 such that for any @xmath353 and @xmath382 in @xmath373 , we have @xmath383 for all @xmath384 $ ] .
first we prove that @xmath75 is a semigroup .
since @xmath385 is a mapping from @xmath184 to @xmath184 , we have @xmath386 where we also used and the semigroup property of @xmath224 .
we now prove the lipschitz continuity of @xmath75 . by using theorem [ th :
stab ] , we obtain that @xmath387 by a weak solution of the camassa holm equation we mean the following .
let @xmath388 that satisfies 1 .
@xmath389 , 2 . the equations @xmath390 and @xmath391 hold for all @xmath392 .
then we say that @xmath3 is a weak global solution of the camassa
holm equation . given any initial condition @xmath362 ,
we denote @xmath393 .
then @xmath54 is a weak , global solution of the camassa
holm equation .
after making the change of variables @xmath394 we get on the one hand @xmath395y_\xi(t,\xi)d\xi
dt\\{\nonumber } & = -\iint_{{\mathbb{r}}_+\times { \mathbb{r}}}[u(t,\xi)y_\xi(t,\xi)(\phi(t , y(t,\xi)))_t-\phi_\xi(t , y(t,\xi))u(t,\xi)^2]d\xi dt\\ & = \int_{\mathbb{r}}u(0,\xi)\phi(0,y(0,\xi))y_\xi(0,\xi)d\xi\\{\nonumber } & \quad + \iint_{{\mathbb{r}}_+\times { \mathbb{r } } } [ u_t(t,\xi)y_\xi(t,\xi)+u(t,\xi ) y_{\xi t}(t,\xi)]\phi(t , y(t,\xi))d\xi dt \\{\nonumber } & \quad + \iint_{{\mathbb{r}}_+\times { \mathbb{r } } } u^2(t,\xi)\phi_\xi(t , y(t,\xi))d\xi dt\\{\nonumber } & = \int_{\mathbb{r}}u(0,x)\phi(0,x)dx\\{\nonumber } & \quad -\iint_{{\mathbb{r}}_+\times { \mathbb{r } } } ( q(t,\xi)y_\xi(t,\xi)+u_\xi(t,\xi)u(t,\xi))\phi(t , y(t,\xi))d\xi dt , \end{aligned}\ ] ] while on the other hand @xmath396 which shows that is fulfilled .
equation can be shown analogously @xmath397\phi(t , y(t,\xi))d\xi dt\\{\nonumber}&= \iint_{{\mathbb{r}}_+\times { \mathbb{r}}}[\frac{1}{2 } u_x^2(t , x)+u^2(t , x)-p(t , x)]\phi(t , x ) dx dt .
\end{aligned}\ ] ] in the last step we used the following @xmath398 for almost every @xmath399 the set @xmath400 is of full measure and hence @xmath401 which is bounded by a constant for all times .
thus we proved that @xmath3 is a weak solution of the camassa
holm equation .
the mapping @xmath402 is continuous from @xmath23 into @xmath45 . in other words , given a sequence @xmath403 converging to @xmath354 , then @xmath404 converges to @xmath405 in @xmath45 .
* acknowledgments . *
k. g. gratefully acknowledges the hospitality of the department of mathematical sciences at the ntnu , norway , creating a great working environment for research during the fall of 2009 . | we study stability of solutions of the cauchy problem on the line for the camassa
holm equation @xmath0 with initial data @xmath1 . in particular
, we derive a new lipschitz metric @xmath2 with the property that for two solutions @xmath3 and @xmath4 of the equation we have @xmath5 . the relationship between this metric and the usual norms in @xmath6 and @xmath7
is clarified .
the method extends to the generalized hyperelastic - rod equation @xmath8 ( for @xmath9 without inflection points ) . |
the anticipated ilc physics program indicates that identification of heavy fermions with high efficiency and purity is of primary importance .
the requirements in terms of track extrapolation accuracy are @xmath1 .
the vertex tracker design and the sensor r&d is driven by this requirement .
cmos pixel sensors are an attractive option for the ilc vertex tracker . in particular
, they can be back - thinned to 50 @xmath0 m or less , without significant performance deterioration @xcite . while there has already been some experience with the reconstruction of well - isolated , high momentum particles with monolithic pixel sensors @xcite
, no data exists on low momentum particle tracking with thin pixel sensors with occupancy conditions comparable to those expected at the ilc .
we present here results obtained with a thin pixel prototype telescope ( tppt ) on the 1.5 gev @xmath2 beam at the lbnl advanced light source ( als ) accelerator complex .
the tppt is the first beam telescope made of thinned cmos pixel sensors .
it consists of three planes of thin pixel sensors ( layers 1 to 3 ) , each spaced by 17 mm .
one additional detector ( layer 4 ) is added 17 mm downstream of the third layer . the mimosa 5 chip @xcite , developed at iphc in strasbourg , france , has been selected for the tppt .
this chip , fabricated in the 0.6 @xmath0 m ams process , features a large active area of [email protected] @xmath4 and more than 1 m pixels .
the epitaxial layer is 14 @xmath0 m thick and the pixel pitch 17 @xmath0m.one sector of each mimosa 5 chip , corresponding to a 510@xmath3512 pixel array , is readout through a custom fpga - driven acquisition board .
the beam spill consists of a single bunch with a repetition rate of 1 hz and the extraction signal is used as a trigger . for each spill
we acquire three frames , one before the particles arrive on the tppt .
four 14 bits , 40 msample / s adcs simultaneously read the four sensors , while an array of digital buffers drive all the required clocks and synchronisation signals . a 32 bits
wide bus connects the fpga to a digital acquisition board installed on a control pc .
data is processed on - line to perform correlated double sampling , pedestal subtraction , noise computation and cluster identification . to reduce the amount of data written to disk only the addresses and pulse heights of the pixels in a fixed matrix around the centre of a cluster candidate
are recorded .
the data is then converted in the lcio format and the offline analysis is performed by processors developed within the marlin framework @xcite .
each event is scanned for pixels with pulse height over a signal - to - noise ( s / n ) threshold of 4.5 , these are designated as cluster ` seeds ' .
seeds are then sorted according to their pulse height values and the surrounding , neighbouring pixels are tested for addition to the cluster .
pixels with a pulse height in excess to 2.0 time the noise are accepted .
the neighbour search is performed in a 5@xmath35 matrix .
clusters are not allowed to overlap and we require that clusters are not discontinuous , i.e. pixels associated to a cluster can not be interleaved by any pixel below the neighbour threshold .
the point of impact of the particle track on the detector is determined by reconstructing the charge centre - of - gravity of the cluster .
the tppt telescope is operated in an optical enclosure mounted on an optical rail and aligned on the als btf beam line .
we report here results of the first data taking performed in fall 2006 .
the temperature was kept constant during operation at @xmath5 27@xmath6c by forced airflow .
the cluster s / n averaged on the tppt layers is 14.5 . due to a drop of the voltage supplied to the chip , one of the planes of the telescope was not fully efficient during the als data taking . because of that we also accept particle tracks reconstructed on two layers , provided they have a hit on the reference plane within 150 @xmath0 m .
c0.5 detailed simulation of the tppt has been carried out , using the geant-4 package @xcite to generate the particle points of impact and energy deposits on the sensitive planes .
these have been stored in lcio format and used as input to a cmos pixel simulation program implemented in marlin @xcite .
noise values have been matched to those measured for the detectors in the telescope .
digitised simulation has been passed through the same cluster reconstruction program as the real data and reconstructed hits used to fit straight particle tracks , after pattern recognition .
the single detector response is well reproduced by the simulation .
figure [ fig:1 ] compares the cluster pulse height and the pixels multiplicity in a cluster for simulation and als beam test data .
the tppt at the als beam - line allows us to perform detailed studies of particle tracking with various , controllable , levels of track density under realistic conditions .
data have been collected at the bts with different beam intensities ranging from 0.5 particles mm@xmath7 up to about 5 particles mm@xmath7 .
these particle densities resemble those expected in the core of hadronic jets at @xmath8 = 500 gev . in particular , the distribution of the distance between a hit associated to a particle track and its closest hit reconstructed on the same layer in the high intensity runs at the als reproduces that predicted by simulation for @xmath9 for @xmath10 = 120 gev at 500 gev . after data taking
, the beam telescope geometry has been surveyed using an optical metrology machine .
the results of this survey have been used as starting point of the alignment procedure , performed on a sample of approximately 20000 well - isolated particle tracks with four correlated hits .
after alignment , track candidates have been defined by matching hits on two layers .
each track candidate has been extrapolated to the third layer , where the closest hit has been added , if its residual was less than 50 @xmath0 m .
pattern recognition ambiguities have been solved based on the number of associated hits and on the difference between the track slope and the expected beam slope , determined from the settings of the beam - line final dipoles .
tracks have then been re - fitted using a modified least - square , to account for kinks due to multiple scattering on the measuring planes @xcite and extrapolated to the reference layer , where the residual to the closest reconstructed hit has been computed . since the beam contains a fraction of lower momentum particles , which increases moving away from the beam axis , only the central region of the beam was used and an additional cut was imposed on the residual on the second coordinate .
r0.70 the tracking performance has been characterised by the residuals between the extrapolation of a track reconstructed using at least two layers and the point reconstructed on the reference layer using two different geometries .
the first uses the extrapolation of the track back to the first layer .
this resembles closely the case of track extrapolation from the vertex detector to the interaction point . using all accepted tracks we measure a residual of ( [email protected] ) @xmath0 m ( see figure [ fig:2 ] ) , which becomes ( [email protected] ) @xmath0 m when restricting to three - hit tracks
this should be compared to 6.8 @xmath0 m obtained from the simulation , which assumes perfect geometry .
subtracting in quadrature the estimated single point resolution of x @xmath0 m , we obtain an extrapolation resolution of 8.5 @xmath0 m for a 1.5 gev particle , which is consistent with the impact parameter resolution required for the ilc . the second geometry adopted extrapolates the track on the second detector layer .
in this case the multiple scattering effect is reduced by having measurements on both side of the extrapolation plane and we measure a residual of ( [email protected] ) @xmath0 m . despite the multiple scattering effect ,
the extrapolation resolution of the tppt at the als is significantly smaller than the mimosa-5 pixel pitch .
this allows to perform studies of cluster shape as a function of the track point of impact .
we compared the number of pixels , along the horizontal coordinate , in clusters reconstructed on the first layer which are associated to two sets of tracks . in the first set tracks with extrapolation along the @xmath12 horizontal axis within 4 @xmath0 m from the pixel centre
are chosen : the average pixel multiplicity is 1.5 . in the second set
the track intercepts the detector more than 8 @xmath0 m away from the pixel centre : the average pixel multiplicity increases to 2.3 .
a second thin pixel telescope ( tppt-2 ) has been built for use in the t-966 beam test experiment of the 120 gev proton beam at the fermilab mtest facility .
the tppt-2 consists of four layers of 50 @xmath0m - thin mimosa-5 sensors mounted on new mezzanine cards with low profile components and larger clear region in the pc board below chip .
the four layers are mounted using precision mechanics and are spaced by 15 mm .
the chips have been positioned on the mezzanine boards using a precision vacuum chuck which gives a mounting accuracy better than 50 @xmath0 m .
downstream from the tppt-2 , a detector under test ( dut ) can be mounted on a computer - controlled xy stage which allows to remotely align it to the telescope .
the dut spacing from the tppt-2 can be varied from 5 mm to 20 mm .
the tppt-2 has been tested in may 2007 on a 1.23 gev @xmath2 beam extracted from the als .
an average s / n of 15.5 has been measured operating at 27@xmath6c . for the first t-966 data taking , planned for july and august 2007 , an operating temperature of 20@xmath6c is foreseen , which will reduce the noise .
this work was supported by the director , office of science , of the u.s .
department of energy under contract no.de-ac02-05ch11231 and used resources of the national energy research scientific computing center , supported under contract no.de-ac03-76sf00098 .
we are indebted to the staff of the lbnl advanced light source for their help and the excellent performance of the machine .
99 m. battaglia , d. contarato , p. giubilato , l. greiner , l. glesener and b. hooberman , nucl .
instrum .
a * 579 * ( 2007 ) 675 [ arxiv : physics/0611081 ] .
s. stanic _ et al .
_ , nucl .
instrum .
meth . a * 568 * ( 2006 ) 181 .
m. trimpl _ et al .
_ , nucl .
instrum .
a * 568 * ( 2006 ) 201 .
et al . _ ,
instrum . and meth .
a * a 513 * ( 2003 ) , 291 .
f. gaede , nucl .
instrum .
a * 559 * ( 2006 ) 177 .
s. agostinelli _ et al .
_ , nucl .
instrum .
a * 506 * ( 2003 ) , 250 .
m. battaglia , nucl .
instrum .
a * 572 * ( 2007 ) 274 .
g. lutz , nucl .
instrum .
a * 273 * ( 1988 ) 349 . | we report results on a tracking performance study performed using a beam telescope made of 50 @xmath0m - thick cmos pixel sensors on the 1.5 gev electron beam at the lbnl als . |
cosmological observations of the last decade surely assert that the main part of the energy density of the universe more than 70% belongs to the unknown essence , called dark energy .
its cosmological mission is to provide the accelerated expansion of the universe , revealed from exploration of sn ia s in the distant galaxies and temperature fluctuation power spectrum of cosmic microwave background .
the cosmological @xmath4cdm - model , based on the einstein equations with cosmological constant ( see @xcite and references therein ) , describes very well almost whole set of the observational data on dynamics of expansion of the universe and formation of its large - scale structure .
but physical interpretation of the cosmological constant is rather problematic @xcite . therefore alternative approaches new physical fields ( classical scalar field
quintessence , tachyon field , k - essence , phantom field , quintom field ) , chaplygin gas , gravity and general relativity modifications , multidimensional gravity , branes and others are intensively analysed ( see reviews @xcite and special issue of gen .
gravit . , 2008 , v.40 ) now .
up to now none of them has crucial preferability from observational or theoretical point of view . therefore each of them must be comprehensively studied . here
we restrict ourselves to quintessential scalar fields with classical lagrangian @xmath9 in the dark energy matter dominated universe .
the quintessence model can be defined by setting of the appropriate potential @xmath10 or equation of state ( eos ) parameter @xmath11 .
there is a dozen or more physically - motivated shapes of the potential @xmath10 : exponential , double exponential , exponential with inverse power , power - law , etc .
the dynamics of such scalar fields is intensively studied ( see review @xcite ) .
the eos parameter of dark energy completely defines the background dynamics as well as the evolution of cosmological perturbations @xcite . since observational data on sn ia magnitude
redshift relation and cosmic microwave anisotropy give relatively narrow ranges of dark energy density and eos parameter values , it looks quite attractive to establish the potential @xmath10 using these data and analyse the background dynamics and perturbative properties of such scalar field which are not studied widely enough . in our previous papers
we have constructed the potentials of scalar fields with classical and tachyonic lagrangian leading to the constant eos parameter @xmath12 @xcite and analysed the background dynamics and perturbative properties of such scalar fields @xcite .
it was shown that cosmological model with cold dark matter and such types of the scalar field ( @xmath13-model ) agrees slightly better with the accessible today observable data than the @xmath4cdm - model . but
difference of quantitative merits of goodness is not large enough to pick out one of them at confidential level of @xmath14 .
since the degeneracies between model parameters of dark energy and cosmological parameters @xcite exist for the background dynamics , the complete analysis of linear density perturbations in both dark matter and dark energy components is important for improvement of dark energy observational tests . among the large number of free quintessence parameters and unknown initial values of quintessence perturbation modes there is only small part of models , for which the evolution of perturbations has been studied .
the general conclusion is that magnitudes of dark energy density perturbations on scales smaller than horizon are essentially lower than corresponding magnitudes of matter density ones .
but character of their evolution depends strongly on the scalar field model ( its potential , time variation of eos parameter , sound speed , etc . ) , initial conditions , scale of perturbations and gauge ( see for example @xcite ) . here
the special attention should be paid to the eos parameter of dark energy @xmath15 , which can be constant or varying in time .
the temporal variation of the dark energy eos parameter is often presented by linear fitting formula with two @xcite or three @xcite parameters to be estimated .
other functional dependences of @xmath16 on scale factor or redshift can be found in @xcite .
here we study the parametrization of the equation of state , which needs only 1 additional quantity with clear physical meaning the adiabatic speed of sound @xmath17 ( the analysis of generalized dark sector components can be found in the early works @xcite ) . in general , @xmath18 is the unknown function of time .
however , taking into account the simplicity we restrict ourselves to @xmath19 , so it is regarded only as the second physical parameter defining the equation of state of dark energy ( the first one the present value of @xmath16 ) . in this paper
we undertake the comparative analysis the evolution of gauge - invariant variables of the scalar perturbations in the model with non - relativistic matter ( @xmath20 ) and scalar field which we define by classical lagrangian with potential constructed for two cases ( @xmath12 and @xmath3 ) in the concordance cosmological models .
these cases have been chosen because they allow us to obtain analytical solutions , which seems to look very attractive in the world of numerical computations .
we assume the adiabatic initial conditions for matter and dark energy scalar perturbations .
we consider the homogeneous and isotropic flat universe with metric of 4-space @xmath21 where the factor @xmath22 is the scale factor , normalized to 1 at the current epoch @xmath23 , @xmath24 is conformal time ( @xmath25 ) .
henceforth we also put @xmath26 , so the time variable @xmath27 has the dimension of a length . here and below the latin indices
@xmath28 run from 0 to 3 , the greek ones over the spatial part of the metric : @xmath29=1 , 2 , 3 . if the universe is filled with non - relativistic matter ( cold dark matter and baryons ) and quintessence which interact only gravitationally ( minimal coupling ) then the dynamics of its expansion is completely described by the einstein equations @xmath30 where @xmath31 is the ricci tensor and @xmath32 , @xmath33 energy - momentum tensors of matter @xmath34 and quintessence @xmath35 .
if these components interact only gravitationally then each of them satisfy the differential energy - momentum conservation law separately : @xmath36 ( here and below `` ; '' denotes the covariant derivative with respect to the coordinate with given index in the space with metric @xmath37 ) . for the perfect
fluid with density @xmath38 and pressure @xmath39 , related by the equation of state @xmath40 , it gives @xmath41 ( here and below a dot over the variable denotes the derivative with respect to the conformal time : `` @xmath42''@xmath43 ) .
the matter is considered to be non - relativistic , so @xmath44 and @xmath45 ( here and below `` 0 '' denotes the present values ) .
we assume the quintessence to be a scalar field @xmath46 with classical lagrangian @xmath47 where @xmath10 is the field potential . we suppose also the background scalar field to be homogeneous ( @xmath48 ) , so its energy density and pressure depend only on time : @xmath49 then the conservation law ( [ conserv - eq ] )
gives the scalar field evolution equation ( called the klein - gordon one ) @xmath50 where @xmath51 is the hubble parameter for any moment of conformal time @xmath24 .
we specify the model of quintessence using two thermodynamical parameters : the eos parameter @xmath52 and the adiabatic speed of sound @xmath53 . in general case
they are connected by equation @xmath54 ( here and below we omit index @xmath0 for @xmath15 ) .
if the time dependence of @xmath55 is known then @xmath18 is defined unambiguously , if @xmath18 is determined then the initial value @xmath56 must be defined additionaly , so , the eos parameter has 2 degrees of freedom : a function and a constant . for other parametrizations see @xcite . since the constraints for time dependence of @xmath55 or @xmath57
are not established well we consider two simple cases : @xmath2 and @xmath58 . in the first case @xmath59 and in the second one @xmath60 this equation has obvious asymptotical behaviour : when @xmath61 @xmath62 and when @xmath63 @xmath64 .
so , at early epoch the dark energy mimics dust matter ( @xmath65 ) for @xmath3 or radiation ( @xmath66 ) for @xmath67 . in future such scalar field will mimic cosmological constant ( @xmath68 ) .
the time dependences of eos parameter for both cases are shown in fig.[fig1 ] .
the equation ( [ eqconsm ] ) has the analytical solutions for two cases : * @xmath2 : @xmath69 and * @xmath3 : @xmath70 $ ] , so it s possible to simplify formulae and calculations and we will analyse only this two cases now . if the parametrization of eos parameter is given , it is possible to apply reverse engineering and construct the fields @xmath0 and potentials @xmath10 . from ( [ rho_q ] )
one simply obtains : @xmath71 if the integral for @xmath0 can be expressed via functions that could be inverted to obtain @xmath72 , then @xmath73 can be easily written in analytical form .
so , from the einstein and field equations we deduce the time dependences of the hubble @xmath74 and acceleration @xmath75 parameters as well as the evolution of the scalar field @xmath0 and potential @xmath10 : @xmath76^{2\frac{1+w}{w}}\end{aligned}\ ] ] for @xmath2 and @xmath77 @xmath78 ^ 2-\frac{3h_0 ^ 2}{8\pi g}\omega_qw_0\end{aligned}\ ] ] for @xmath3 . on scale factor @xmath79 for @xmath80 and @xmath81 .
middle : the dynamics of expansion of the homogeneous universe in the model with non - relativistic matter and quintessential scalar field with @xmath3 and @xmath81 with best fitting cosmological parameters from spergel et al .
( 2007 ) ( @xmath82 , @xmath83 , @xmath84 , @xmath85 ) : matter and quintessence densities in units of the critical one .
bottom : the evolution of acceleration parameter . for comparison
we show also the corresponding dependences for @xmath4cdm - model with @xmath86 , @xmath87 and @xmath88 ( spergel et al .
( 2007 ) , apunevych et al .
( 2007 ) ) .
, width=302 ] for both models there are 2 independent solutions for the field ( the growing one corresponds to sign `` + '' and the decaying one to sign `` - '' ) and 2 symmetrical with respect to @xmath89 potentials exist .
however , the physical consequences of both these solutions are the same @xcite , so from now we restrict ourselves only to the growing one .
the variety of scalar field potentials was presented in @xcite .
the potential for @xmath2 can be also found in @xcite , the potential for @xmath90 belongs generally to the family of double exponential potentials ( with additional constant term ) , but both they differ from the physically - motivated ones , for which the evolution of scalar linear perturbations was studied by other authors .
we must note that the asymptotic behaviour at @xmath91 of the expansion rate @xmath92 and acceleration parameter @xmath93 in both cases is the same and similar to that in @xmath4cdm - model : @xmath94 , @xmath95 . at current epoch the parameters of expansion dynamics are the same ( @xmath96 and @xmath97 ) for both models .
but their asymptotic behaviour at @xmath98 is different : in @xmath2 quintessence @xmath99 , @xmath100 , @xmath101 and in @xmath3 quintessence @xmath102 , @xmath103 , @xmath104 .
the energy densities of both fields evolve similarly but have different asymptotic regimes : in the quintessence with @xmath2 @xmath105 always while in the @xmath3 quintessence at @xmath91 @xmath106 and at @xmath98 @xmath107 .
so , the scalar field with @xmath3 behaves as cold dark matter at the early epoch and will mimic the cosmological constant in far future .
the different asymptotic behaviour of these fields is caused by their intrinsic properties . in the @xmath2 quintessence
the negative pressure stiffly follows its energy density and their relation is always constant . in the @xmath3 quintessence
the negative pressure is always constant : @xmath108 .
so , it is insignificant in the early epoch when @xmath91 and @xmath109 for the model of the universe filled only with dust matter and quintessential dark energy , and important in the late one when @xmath110 .
the dynamics of expansion of homogeneous universe in the model with non - relativistic matter and quintessential scalar field with @xmath3 and @xmath81 is shown in fig.[fig1 ] .
for both models we assume best fitting cosmological parameters from @xcite ( @xmath82 , @xmath111 , @xmath84 , @xmath85 ) . for comparison we show also the corresponding dependences in @xmath4cdm - model with @xmath112 , @xmath87 and @xmath88 @xcite .
we have constructed the potentials of quintessential scalar fields with @xmath2 @xcite and @xmath3 for qcdm cosmological model with best fitting parameters obtained from wmap and snia data @xcite .
the evolution of fields @xmath113 , potentials @xmath114 and rolling down of the fields @xmath0 to the minimum which is located at @xmath115 ( @xmath63 ) are shown in fig.[fig3 ] .
the discussion of influence of parameter determination uncertainties on potential of field with @xmath2 can be found in @xcite .
so , the difference of the homogeneous universe expansion dynamics in @xmath4cdm- and such qcdm - models is too small to discriminate them using the avialable datasetets .
that s why in the next sections we will analyse the linear stage of growth of scalar perturbations of matter and dark energy .
for this we will use gauge - invariant approach developed by @xcite .
for analysis of the scalar linear perturbations the conformal - newtonian gauge with space - time metric @xmath116\end{aligned}\ ] ] is convenient . here
@xmath117 and @xmath118 are gauge - invariant perturbations of metric @xcite called bardeen s potentials . if proper anisotropy of medium equals zero then @xmath119 .
dust matter and scalar fields have such property @xcite . in the linear perturbation theory
the fourier decomposition is used , so spatial dependences of all variables can be substituded by corresponding fourier amplitudes .
for example , @xmath120 , where @xmath121 is wave number . henceforth , saying about metric @xmath117 , matter density @xmath122 , its peculiar velocity @xmath123 , scalar field @xmath124 , its energy density perturbations @xmath125 etc we mean their fourier amplitudes @xmath126 , @xmath127 , @xmath128 , @xmath129 , @xmath130 , etc .
the metric ( @xmath126 ) , matter density and velocity perturbations ( @xmath127 , @xmath128 ) as well as scalar field perturbations ( @xmath129 , @xmath130 , @xmath131 ) in the conformal - newtonian gauge are gauge - invariant variables @xcite . the energy density and velocity perturbations of quintessence , @xmath132 and @xmath8 , are connected with the perturbation of field variable @xmath133 in following way : @xmath134 other non - vanishing gauge - invariant perturbations of scalar field are isotropic pressure perturbation @xmath135 and intrinsic entropy @xmath136 the density perturbation of any component in the conformal - newtonian gauge @xmath137 , which is gauge - invariant variable , is related to the other gauge - invariant variables of density perturbations @xmath138 and @xmath139 as : @xmath140 where @xmath141 , @xmath138 , @xmath139 and @xmath142 correspond to either @xmath143- or @xmath0-component . here
@xmath139 is the density perturbation in the rest frame in which the fluctuations of the curvature scalar of the constant time hypersurface vanish and @xmath138 corresponds to the rest frame in which the 4-velocity is orthogonal to constant time hypersurface @xcite .
the intrinsic entropy of quintessence @xmath144 can be presented via gauge - invariant @xmath0-perturbations as follows : @xmath145 this equation shows that the intrinsic entropy for scalar perturbations of quintessence with @xmath146 is non - zero when proper energy density perturbation @xmath7 ( measured in synchronous comoving gauge ) of quintessence is non - vanishing . in the first case ( @xmath2 )
@xmath147 , in the second one ( @xmath90 ) @xmath148 . in the case of perturbed quintessence dissipative processes
generate entropic perturbations , so we have the sound speed @xmath149 defined by more general relation : @xmath150 .
the intrinsic entropy perturbation can be presented in the form : @xmath151 @xcite . for the scalar fields with classical lagrangian @xmath152 @xcite .
evolution equation for scalar field perturbation @xmath129 can be obtained either from lagrange - euler equation or from energy - momentum conservation law @xmath153 : @xmath154 thus , evolution of quintessence perturbation depends on field model ( @xmath10 ) , gravitational potential @xmath155 , expansion rate of the universe @xmath74 and scale of perturbation @xmath121 .
the linearised einstein equations for gauge - invariant perturbations of metric and energy - momentum tensor components are @xmath156 here and below a prime denotes the derivative with respect to the scale factor @xmath79 .
the conservation equations for matter density and velocity perturbations @xmath157 in terms of the gauge - invariant variables @xmath158 and @xmath159 are following : @xmath160 they are connected with the dark energy ones only via @xmath155 and are the same for both models of quintessence .
so , in each case we have the system of 5 first - order ordinary differential equations for 5 unknown functions @xmath161 , @xmath162 , @xmath163 , @xmath164 and @xmath165 . from this systems of equations
it is easy to obtain the systems of 2 second - order ordinary differential equations for 2 unknown functions @xmath161 and @xmath166 : @xmath167 @xmath168 for @xmath2 or @xmath169 for @xmath3 . using their solutions ( four fundamental ) for @xmath161 and @xmath170 and the constraint equation @xmath171 it is possible to find the values of @xmath162 , @xmath163 , @xmath164 , @xmath165 and @xmath172 .
the equations ( [ cons_m1])-([cons_m2 ] ) can be substituted by one second - order equation @xmath173 the systems of equations ( [ e2psiw])-([e2phiw ] ) and ( [ e2psi0])-([e2phi0 ] ) describe the evolution of perturbations of gravitational @xmath155 and quintessentional @xmath174 fields and their coupling .
a few important conclusions can be deduced from qualitative analysis of these systems : * the coupling of @xmath175 and @xmath176field is modulated by the value @xmath177 for @xmath2 and @xmath178 for @xmath3 .
so , if @xmath179 then both fields evolve independently .
since observational data prefer current @xmath55 close to @xmath180 , so their coupling is weak .
* evolution of @xmath176field depends on relation of scale of perturbation to horizon explicitly while the dependence of @xmath175field is implicit ( through the latter ) .
* the system of equations for @xmath55-quintessence ( [ e2psiw])-([e2phiw ] ) allows the asymptotic behaviour @xmath181 , @xmath182 when @xmath98 with relation between them @xmath183 . *
the system of equations for @xmath57-quintessence ( [ e2psi0])-([e2phi0 ] ) allows the asymptotic behaviour @xmath181 , @xmath182 when @xmath91 with initial relation between them @xmath184 . * from equation ( [ cons_m3 ] ) it follows that @xmath185 for superhorizon perturbations ( @xmath186 ) .
if @xmath187 and @xmath188 at @xmath91 then either @xmath189 for @xmath2 or @xmath190 for @xmath90 and for @xmath191 it begins to grow slowly from the constant value .
the decay of @xmath155 and transition from deceleration to acceleration slow the growth of @xmath158 .
analysis of the background dynamics presented in the previous section has shown that both qcdm - models are matter - dominated in the early universe ( fig .
[ fig1 ] ) . in the qcdm - model with @xmath2 the ratio @xmath192 when @xmath91 , while in the qcdm - model with @xmath3 @xmath193 and @xmath194 when @xmath91 .
the adiabatic growing mode of perturbation in the non - relativistic matter - dominated universe can be specified by the condition @xmath187 ( @xmath195 ) .
adiabaticity condition in two - component model ( @xmath196 @xcite ) gives @xmath197 .
these conditions , constraint equations written for hypersurface @xmath198 ( @xmath199 ) and the analytic asymptotic solutions ( see next subsection ) lead to the following adiabatic initial conditions : @xmath200 for @xmath2 and @xmath201 for @xmath3 .
therefore , the growing mode of adiabatic perturbations in two - component ( non - relativistic matter and quintessence ) medium is defined by single value initial gravitational potential @xmath202 .
since the non - adiabatic initial perturbations are strongly constrained by wmap data , in this paper we restrict ourselves only to adiabatic initial conditions . in order to analyse the evolution of gauge - invariant variables of matter and quintessence perturbations
we must solve the system of equations ( [ edgqw])-([epsiw ] ) together with ( [ cons_m1])-([cons_m2 ] ) numerically for initial conditions ( [ ivqw])-([idgmw ] ) or ( [ ivq0])-([idgm0 ] ) respectively . but before we propose the analysis of these systems of equations in the early epoch ( @xmath203 ) , for which the analytical solutions are known .
so , the system of equations ( [ e2psi0])-([e2phi0 ] ) for @xmath203 can be simplified as @xmath204 this system of equations has 4 fundamental solutions , so it is possible to write the general solution in the form : @xmath205.\end{aligned}\ ] ] the first two solutions , noted by the constants of integration @xmath206 and @xmath207 , are well known growing and decaying modes of adiabatic perturbations in the dust matter - dominated universe .
the next two solutions , noted by the constants of integration @xmath208 and @xmath209 , are due to possible entropy initial conditions and intrinsic non - vanishing entropy of quintessence .
really , the condition @xmath210 leads to 1 second - order equation which has two dust - like fundamental solutions : @xmath211 for the quintessence with @xmath2 solutions for @xmath155 are the same and @xmath212 .
the quantities @xmath162 , @xmath163 and @xmath165 can be found using the equations ( [ ddgds])-([cons_m2 ] ) and ( [ constrainteq ] ) .
the relations between them are presented in the previous subsection as the set of initial data ( [ ivqw])-([idgm0 ] ) .
we have integrated numerically the systems of equations ( [ edgqw])-([epsiw ] ) for @xmath2 and for @xmath3 together with ( [ cons_m1])-([cons_m2 ] ) for adiabatic initial conditions ( [ ivqw])-([idgmw ] ) and ( [ ivq0])-([idgm0 ] ) using the publicly available code dverk .
we assumed @xmath213 and integrated up to @xmath214 .
the evolution of perturbations is scale dependent , so we performed calculations for @xmath215 and @xmath216 @xmath217 for the cosmological model with parameters @xmath82 , @xmath83 , @xmath84 , @xmath85 .
the evolution of gauge - invariant variables of matter perturbations @xmath158 , @xmath218 , @xmath219 , @xmath159 for two scales @xmath220 and @xmath221 @xmath217 is shown in top panels of fig.[fig4 ] and fig.[fig5 ] . in the bottom panels
the analogical gauge - invariant variables of quintessence perturbations ( @xmath5 , @xmath6 , @xmath7 , @xmath8 , @xmath222 ) are presented ( for @xmath3 the curves @xmath222 and @xmath7 overlap ) .
the evolution of gauge - invariant gravitational potential @xmath155 is shown in all panels for comparison .
all plots are shown for the following range : @xmath223 .
from the top panels of fig.[fig4 ] , [ fig5 ] it follows that the magnitudes of the adiabatic matter density perturbations grow similarly in both models ( and like in @xmath4cdm - model ) , but time dependences of magnitudes of the adiabatic quintessence energy density perturbations are more varied ( bottom panels of the same figures ) : gauge - invariant variables @xmath5 and @xmath6 decay from initial constant value after the particle horizon entry while @xmath7 and @xmath8 grow at early stage before the horizon entry and decay after that in the quintessence - dominated epoch , when gravitational potential starts to decay . the perturbation shown in fig.[fig4 ]
enters the particle horizon ( @xmath224 ) at @xmath225 for @xmath55-quintessence and at @xmath226 for @xmath57-quintessence .
the perturbation shown in fig.[fig5 ] enters the particle horizon at @xmath227 and @xmath228 for @xmath55- and @xmath57-quintessence respectively .
the particle horizon at current epoch ( @xmath229 ) in the cosmological model with parameters @xmath82 , @xmath83 , @xmath84 , @xmath85 and @xmath55-quintessence equals @xmath230 mpc . in the model with @xmath57-quintessence
it is @xmath231 mpc . at early epoch @xmath232 for both models of quintessence .
after the particle horizon entry the amplitudes start to decay slowly in the matter - dominated epoch and decay fast in the quintessence - dominated one . at asymptotic regime for quintessence model with @xmath3 approximately @xmath233 . in the quintessence model with @xmath2 the transition epoch is extended in time . in fig.[fig6 ] we show the dependences of ratios of quintessence density perturbations to matter density ones in conformal - newtonian gauge ( @xmath234 ) on scale factor for perturbations with the scales @xmath235 , @xmath236 , @xmath221 and @xmath216 @xmath217 .
these curves emphasise the difference of evolution of perturbations in ordinary matter and quintessence as well as the similarity of behaviour of perturbations in two models of quintessence .
the magnitudes of quintessence density perturbations in units of matter ones in both models at current epoch are close although their initial magnitudes differ by order .
the magnitudes of quintessence density perturbations with scale less than particle horizon are lower than corresponding magnitudes of matter density perturbations by factor @xmath237 so , that for scale @xmath238 @xmath239 .
therefore , on subhorizon scales the quintessential scalar fields are practically smoothed out while the matter clusters . on scale factor for linear perturbations with scales @xmath235 , @xmath236 , @xmath221 and @xmath216 @xmath217 ( from top to bottom ) in the models with non - relativistic matter and quintessence ( @xmath2 top panel , @xmath3
bottom).,width=302 ] the tests for choice of the type of dark energy are based on the results of its action on luminous matter and cosmic microwave background .
so , the key question is how these types of quintessence affect the growth of matter density perturbations and the time variation of gravitational potential . from top panels of fig.[fig4 ] and [ fig5 ] we can see that they are more suppressed for @xmath3 than for @xmath2 and for perturbations with smaller scale .
in order to illustrate this effect in fig.[fig7 ] we present ratios @xmath240 and @xmath241 for scales @xmath235 , @xmath236 , @xmath221 and @xmath216 @xmath217 .
we can see that scale dependence of suppression of magnitude of matter density perturbations as well as of gravitational potential is strong for @xmath3 quintessence and weak for @xmath2 one . in the @xmath4cdm - model
it is scale - independent @xcite .
( in the einstein de sitter model both ratios are equal to 1 for all times and scales ) .
these ratios are substantial for calculations of magnitude of the matter density power spectrum at different redshifts and the angular power spectrum of cmb temperature fluctuations in the range of scales of the late integrated sachs - wolfe effect .
evolution of quintessence perturbations depends on scalar field model ( i.e. its lagrangian and potential ) , contents of the universe , coupling of the quintessence to other components , initial conditions and scale of perturbations @xcite .
here we have analysed the evolution of scalar matter and quintessence perturbations for potentials of scalar fields with classical lagrangian constructed to give either @xmath2 or @xmath3 . therefore , obtained here results could be compared to the results of other authors only qualitatively .
evolution of eos parameter in our @xmath3-model ( fig.[fig1 ] ) is similar to that of @xcite . despite the other cosmological model and potential of scalar field ,
the qualitative behaviour of quintessence perturbations is close to obtained here : @xmath5 is @xmath242 when the perturbation is outside the particle horizon and decays when it enters the horizon .
the growth of magnitude of quintessence density perturbations long before the horizon entry in synchronous gauge was shown by @xcite ( models with @xmath2 in fig.3 ) .
our results for evolution of gauge - invariant variable @xmath7 ( density perturbation in synchronous gauge ) shown in fig.[fig4 ] support this conclusion .
( we do not discuss the oscilations at early stage visible in fig.3 of @xcite because of different initial conditions and background . ) the ratios of quintessence ( @xmath2 , @xmath152 ) density perturbations to matter density ones in synchronous gauge are shown in fig.1 of @xcite for @xmath238h@xmath243 @xmath217 .
presented here in fig.[fig6 ] analogical ratios in conformal - newtonian gauge are similar .
the conclusion about anti - correlation between the perturbations of the matter and quintessence has been done by @xcite and @xcite on the base of analysis of their evolution in the matter rest frame . @xcite and
@xcite noted this effect too .
recalculation of the frame - dependent variables to gauge - invariant ones will in our belief remove such variance .
and @xmath241 for linear perturbations with scales @xmath235 , @xmath236 , @xmath221 and @xmath216 @xmath217 ( from top to bottom ) in the models with non - relativistic matter and quintessence ( @xmath3 solid line , @xmath2 dashed line).,width=302 ] the dynamics of expansion of the universe and evolution of scalar perturbations are studied for the quintessential scalar fields @xmath0 with the classical lagrangian @xmath1 satisfying the additional condition @xmath2 or @xmath3 . for both quintessential scalar fields the potential @xmath10 and time dependence of @xmath0
are constructed for the same cosmological model and it is shown that the accelerated expansion of the universe is caused by the effect of rolling down of the potential to minimum ( fig.[fig3 ] ) . in qcdm - model with @xmath2 the ratio @xmath192 when @xmath91 , while in qcdm - model with @xmath3 @xmath193 and @xmath194 when @xmath91 . at the early epoch @xmath55-quintessence is dynamically unsubstantial like cosmological constant while @xmath57-quintessence mimics dust matter ( @xmath65 ) at @xmath203 and cosmological constant ( @xmath244 ) at @xmath245 .
the dependence of acceleration parameter on redshift is a bit different for them ( fig.[fig1 ] ) but close to @xmath4cdm - model one and indistinquishable observationally now .
asymptotic analysis of the systems of evolutionary equations for gauge - invariant perturbations has shown that adiabatic initial conditions for non - relativistic matter and @xmath55- and @xmath57-quintessence are allowed .
the numerical integration of these systems give time dependences of gauge - invariant variables for matter and quintessence scalar perturbations ( fig.[fig4 ] , [ fig5 ] ) .
the main conclusion deduced from them is following : the magnitudes of the adiabatic matter density perturbations grow like in @xmath4cdm - model , while for quintessence @xmath5 , @xmath6 are constant and @xmath7 , @xmath8 grow before the particle horizon entry but all variables decay after that in such way , that at the current epoch they are approximately two orders lower than the corresponding quantities for dust matter on supercluster scales .
therefore , on subhorizon scales the quintessential scalar field is smoothed out while the matter is clustered . the quintessential scalar fields studied here suppress the growth of matter density perturbations and the magnitude of gravitational potential ( fig.[fig7 ] ) .
in these qcdm - models unlike @xmath4cdm ones such suppression is scale dependent and more visible for @xmath57-quintessence . such features of quintessence are important for calculations of the matter density power spectrum at different redshifts and the power spectrum of cmb temperature fluctuations in the range of scales of the late integrated sachs - wolfe effect . that can be used for interpretation of data of current and planned experiments in order to identificate the nature of dark energy .
this work was supported by the project of ministry of education and science of ukraine `` the linear and non - linear stages of evolution of the cosmological perturbations in models of the multicomponent universe with dark energy '' ( state registration number 0107u002062 ) and the research program of national academy of sciences of ukraine `` the exploration of the structure and components of the universe , hidden mass and dark energy ( cosmomicrophysics ) '' ( state registration number 0107u007279 ) . the authors are thankful also to yu .
kulinich for useful discussions .
apunevych s. , venhlovska b. , kulinich yu . & novosyadlyj b. , 2007 , kinematics and physics of celestial bodies , 23 , 45 , arxiv : 0802.0599 [ astro - ph ] bardeen j.m .
, 1980 , phys .
rev.d . , 22 , 1882 bartolo n. , corasaniti p .- s . , liddle a.r .
& malquarti m. , 2004 , phys .
rev.d . , 70 , 043532 bean r. & dore o. , 2004 , phys .
d. 69 , 083503 bennett c.l .
, halpern m. , hinshaw g. , jarosik n. , kogut a. , limon m. , meyer s. s. , page , l. et al . , 2003 , apjs , 148 , 1 bludman s. , 2007 , vi reunion anual sociedad chilena de astronomia ( sochias ) , valparaiso , chile , nov .
2007 , 9 bousso r. , 2008 , gen .
. gravit . , 40 , 607 carroll s.m . , 2001 , living rev .
rel . , 4 , 1 carroll s. m. , press w. h. & turner e. l. , 1992 , ann .
astron . and astrophys . , 30 , 499 chevallier m. & polarski d. , 2001 , int
d , 10 , 213 copeland e.j .
, sami m. , tsujikava s. , 2006 , int . j. mod
d. , 15 , 1753 dave r. , caldwell r.r . & steinhardt p.j .
, 2002 , phys .
d , 66 , 023516 dedeo s. , caldwell r.r . & steinhardt p.j . , 2003 ,
d , 67 , 103509 doran m. , muller c.m . , schafer g. , wetterich c. , 2003 , phys . rev .
d , 68 , 063505 durrer r. , 2001 , j.phys.studies , 5 , 177 durrer r. & maartens r. , 2008 , gen .
, 40 , 301 dutta s. & maor i. , 2007 , phys .
d , 75 , 063507 erickson j.k .
, caldwell r.r . ,
steinhardt p.j .
, armendariz - picon c. , mukhanov v. , 2007 , phys .
lett . , 88 , 121301 frieman j.a . ,
turner m.s .
, huterer d. , 2008 , ann .
astron . and astrophys .
, 46 , 385 hlozek r. , cortes v. , clarkson c. , bassett b. , 2008 , gen .
, 40 , 285 hu w. apj , 1998 , 506 , 485 hu w. , eisenstein d. , phys . rev .
d. , 1999 , 59 , 083509 kodama h. & sasaki m. , 1984 , prog.theor.phys.suppl .
, 78 , 1 komatsu e. , dunkley j. , nolta m.r . ,
bennett c.l . , gold , b. et al .
, 2009 , apjs , 180 , 330 kulinich yu . , novosyadlyj b. , pelykh v. , 2007 , j.phys.studies , 11 , 473 kunz m. , 2007 , astro - ph/0702615 linder e.v . , 2006 , phys .
d , 2006 , 76 , 063010 linder e.v . , 2008 , gen .
gravit . , 40 , 329 liu j. , 2004 , phys .
d , 69 , 083504 mota d.f .
, shaw d.j . & silk j. , 2008 , apj , 675 , 29 ma c .- p . & bertschinger e. , 1995 , apj , 455 , 7 novosyadlyj b. , 2007 , j.phys.studies , 11 , 226 padmanabhan t. , 2003 , phys . rep . , 380 , 235 peebles p.j.e . , 2001 , principles of physical cosmology //
princeton u press , 718 p. peebles p.j.e . & ratra b. , 2003 , rev .
, 75 , 559 sahni v. , saini t. d. , starobinsky a. a. , alam u. , 2003 , j. exp .
lett . , 77 , 201 sahni v. , starobinsky a.a . , 2000 , int . j. mod . phys .
d. , 9 , 373 sahni v. , starobinsky a.a .
, 2006 , int .
d , 15 , 2105 seljak u. , zaldarriaga m. , 1996 , apj , 469 , 437 sergijenko o. & novosyadlyj b. , 2008 , kinematics and physics of celestial bodies , 24 , 345 , arxiv : 0805.3782 [ astro - ph ] sergijenko o. , kulinich yu . , novosyadlyj b. , pelykh v. , 2009 , kinematics and physics of celestial bodies , 25 , 17 , arxiv : 0809.3349 [ astro - ph ] shapiro i.l . & sola j. , 2008 , arxiv : 0808.0315 [ astro - ph ] spergel d.n . , bean r. , dore o. , et al .
, 2007 , astrophys .
, 170 , 377 unnikrishnan s. , jassal h.k . & seshadri t.r . , 2008 , phys .
d , 78 , 123504 weller j. & lewis a.m. 2003 , mnras , 346 , 987 zaldarriaga m. , seljak u. , 1999 , astrophys . j. suppl . ser .
, 29 , 431 zhdanov v.i . , ivashchenko g. , 2008 , kinematics and physics of celestial bodies , 25 , 73 , arxiv : 0806.4327 [ astro - ph ] | the dynamics of expansion of the universe and evolution of scalar perturbations are discussed for the quintessential scalar fields @xmath0 with the classical lagrangian @xmath1 satisfying the additional condition @xmath2 or @xmath3 .
both quintessential fields are studied for the same cosmological model .
it is shown that the accelerated expansion of the universe is caused by the effect of rolling down of the field to minimum . at the early epoch
the contribution to dynamics of the quintessence with @xmath2 is negligible ( like that of cosmological constant ) while quintessence with @xmath3 mimics dust matter . in future
the scalar field with @xmath3 will mimic cosmological constant .
the systems of evolution equations for gauge - invariant perturbations of metric , matter and quintessence have been analysed analyticaly for the early stage of the universe life and numerically up to the present epoch .
it is shown that amplitudes of the adiabatic matter density perturbations grow similarly in both models ( and like in @xmath4cdm - model ) , but time dependences of different amplitudes of the quintessence perturbations are varied : gauge - invariant variables @xmath5 and @xmath6 decay from initial constant value after the particle horizon entry while @xmath7 and @xmath8 grow at the early stage before the horizon entry and decay after that in the quintessence - dominated epoch , when gravitational potential starts to decay
so , that at the current epoch they are approximately two orders lower than matter ones on the supercluster scales .
therefore , on the subhorizon scales the quintessential scalar fields are smoothed out while the matter clusters .
it is also shown that both quintessential scalar fields suppress the growth of matter density perturbations and the amplitude of gravitational potential . in these qcdm - models unlike @xmath4cdm ones such suppression is scale dependent and more visible for the quintessence with @xmath3 . |
the well - known phase structure of the two - dimensional coulomb - gas was investigated in several papers by using the differential rg approach with a smooth cut - off in the momentum space ( or equivalently using a sharp cut - off in the coordinate space ) ( kosterlitz 1973 , kosterlitz 1974 , jose 1977 , huang 1991 , gersdorff 2000 ) .
our goal is in this work to determine the rg flow for the two - dimensional coulomb - gas in the framework of the rg approach of wegner and houghton ( wegner 1973 ) taking the field - independent wave - function renormalization into account .
results of two different methods were compared : renormalization by means of the blocking construction in the coordinate space using the dilute gas approximation ( real space rg ) ( kosterlitz 1973 , kosterlitz 1974 , jose 1977 , huang 1991 ) , and the renormalization of the equivalent sine - gordon scalar field model , performing the blocking transformations in the momentum space in wegner - houghton s framework with a sharp cut - off and usage of the derivative expansion ( wegner - houghton rg method ) ( wegner 1973 , hasenfratz 1986 , polonyi 2000 ) .
it is believed that several different models , like the sine - gordon , thirring , and the x - y planar models belong to the same universality class , namely to that of the two - dimensional coulomb - gas . the x - y model with external field , which is a classical two - component spin - model described by the action
, @xmath0 has topological excitations , called vortices , which interact via coulomb interaction .
therefore , the x - y model can be mapped by means of the villain - transformation to a coulomb - gas ( huang 1991 ) .
such a mapping is , however , only valid up to irrelevant interaction terms .
the other example , the sine - gordon model in dimension @xmath1 is a one - component scalar field theory with periodic self - interaction , which is defined by the euclidean action : @xmath2.\ ] ] the equivalence between the x - y model and the lattice regulated compactified sine - gordon model is shown by expressing in terms of the compact variable @xmath3 ( huang 1991 ) .
this makes the kinetic energy periodic and introduces vortices in the dynamics .
there having been made several efforts during the last two decades to improve the coulomb - gas results obtained in the dilute - gas approximation , where the analogy with the electrodynamics of polarizable media has been exploited ( amit 1980 , minnhagen 1985 , kupferman 1997 ) .
here we also show that the method of wegner and houghton applied to the sine - gordon model provides a systematic way of obtaining higher - order corrections to the dilute - gas results .
the real space rg approach for the coulomb - gas in dimension @xmath1 has been investigated in a great detail in the literature ( kosterlitz 1973 , kosterlitz 1974 , jose 1977 , huang 1991 ) .
the dilute vortex gas approximation assumes that all the vortex pairs ( charges ) are separated with at least to the minimum distance @xmath4 in the coordinate space which is increased with infinitesimal steps during the blocking transformation .
the contributions of the order higher than @xmath5 are neglected .
the real space rg equations for the dilute vortex - gas are well - known and their derivation is given in the literature ( kosterlitz 1973 , kosterlitz 1974 , jose 1977 , huang 1991 ) : @xmath6 with the dimensionless coupling constants @xmath7 and @xmath8 .
the differential rg transformations for the sine - gordon model are realized in the momentum space integrating out the high - frequency modes above the moving sharp cut - off @xmath9 sequentially in infinitesimal steps . in the infinitesimal step , corresponding to moving the cut - off @xmath9 to @xmath10 , it is integrated for the high - frequency fourier - modes of the field variable @xmath11 : @xmath12 } = \int d[\phi ' ] e^{-s_{k}[\phi+\phi']}.\ ] ] performing the path integral in using the saddle point expansion at @xmath13=0 , the wegner - houghton equation is obtained ( wegner 1973 ) : @xmath14 = -{1\over2 } tr ' \log(g_{p p'}^{-1}[\phi])\ ] ] with the inverse propagator @xmath15=\partial^2 s[\phi]/(\partial\phi_p\partial\phi_{p'})$ ] and the trace @xmath16 over the momentum shell @xmath17 $ ] . using the derivative expansion in the form @xmath18 $ ] this reduces to : @xmath19 ^ 2 \left[{4 z^2 k^2\over d{\em a}^4 } - { z\over{\em a}^3}\right],\ ] ] with @xmath20 , the potential @xmath21 , and the field - independent wave - function renormalization @xmath22 .
( [ momentum ] ) are obtained by the method described in ( polonyi 2000 ) .
the flow equations for the dimensionful coupling constants @xmath23 and @xmath24 are obtained by expanding both sides of eqs .
( [ momentum ] ) in fourier series and neglecting the higher harmonics missed on the left hand sides .
it is straightforward to use the derivative of the first equation in with respect to @xmath25 .
thus , we find @xmath26 , \hspace{1 cm } k\partial_k z(k ) = - { 1\over 8\pi } { x^2(k ) \left(1+x^2(k)\right ) \over \left(1-x^2(k)\right)^{5/2 } } , \ ] ] with @xmath27 . in order to compare eqs .
one should introduce the dimensionless couplings via @xmath28 and @xmath29 . choosing the moving distance scale according to @xmath30 , we get @xmath31 + 2{\tilde h } - \pi t { \tilde h}^3 { \left(1 + 8\pi^2{\tilde h}^2\right ) \over \left(1 - 8\pi^2{\tilde h}^2\right)^{5/2 } } , \\ \nonumber a { dt\over da } & = & - \pi t^2 { \tilde h}^2 { \left(1 + 8\pi^2{\tilde h}^2\right )
\over \left(1 - 8\pi^2{\tilde h}^2\right)^{5/2}}.\end{aligned}\ ] ] expanding in series of @xmath32 , one finds : @xmath33 we see that the leading order terms on the right hand sides are those of eqs .
( [ real ] ) .
the terms of higher order in @xmath7 yield the infinite series of corrections to the dilute - gas result and are summed up in a closed form on the right hand sides of eqs .
( [ dimless ] ) .
in a previous paper ( nandori 1999 ) , using the local potential approximation ( @xmath34 ) , it was shown , that the periodicity and the convexity are so strong constraints on the effective potential @xmath35 that it becomes flat .
this flattening was tested numerically , as well . in order to determine the flow of the blocked potential in next - to - leading order of the derivative expansion
, the flow of the wave - function renormalization @xmath36 should also be obtained .
the rg flow of @xmath36 and @xmath37 ( i.e. @xmath38 , @xmath39 ) , obtained by the above described two different blocking transformations ( real space and wegner - houghton rg ) are qualitatively the same .
the real space rg method ( see eqs . ) yielded the well - known phase - structure for the coulomb - gas , ( see fig .
[ coulomb ] ) that has already been obtained in the literature ( kosterlitz 1973 , kosterlitz 1974 , huang 1991 ) using the dilute gas approximation .
there are two phases connected by the kosterlitz - thouless transition . in the molecular phase the vortices and anti - vortices form closely bound pairs
while in the ionised phase they dissociate into a plasma .
similar results were obtained in ( gersodorff 2000 ) , using momentum space rg with smooth cut - off , without the dilute gas approximation .
the flow diagram for the sine - gordon model obtained by the wegner - houghton rg approach ( see eqs . ) using a sharp momentum cut - off , is plotted in fig .
[ sinegor ] . in order to compare the results obtained by the two different rg methods , four rg trajectories calculated by the real space rg method
are plotted in fig .
[ sinegor ] , as well . for small values of @xmath40 and @xmath8 ,
the trajectories for both methods are the same for the same initial conditions . for larger values of @xmath40 the dilute gas approximation , for large values of @xmath8 the villain transformation loose their validity , so that the different rg trajectories belonging to the same initial values start to diverge .
the flow diagram for the sine - gordon model is valid for @xmath41 . at @xmath42
a non - trivial saddle point occurs in the path integral ( [ blocking ] ) and the wegner - houghton equation looses its validity . in order to investigate the flattening of the blocked potential in next - to - leading order of the derivative expansion also the higher harmonics of the periodic potential should be included .
therefore , the generalization to the field dependent , periodic wave - function renormalization should be performed .
since it is unclear how to treat the field - dependent wave - function renormalization in the wegner - houghton approach , such a generalization is in progress using polchinski s method . the authors ( i.n . and k.s . )
thank g. soff , g. plunien and r. schtzhold for the useful discussions .
this work has been supported by the nato grant pst.clg.975722 , the daad - mb project n@xmath43 27/1999 , and the grants otka t023844/97 , t29927/98 .
amit , d.j . , y.y .
goldschmidt , and g. grinstein , j. phys .
* a13 * 585 .
( 1980 ) + gersdorff g.v . and c. wetterich , _ nonperturbative renormalization flow and essential scaling for the kosterlitz - thouless transition _ , hep / th-0008114 + hasenfratz a. and p. hasenfratz , nucl . phys . *
b270 * 687 .
( 1986 ) + huang k. and j. polonyi , int
phys . * a6 * 409 .
( 1991 ) + jose j. v. , l. p. kadanoff , s. kirkpatrick , and d. r. nelson , phys
. rev . * b16 * 1217 .
( 1977 ) + kosterlitz j. m. and d. j. thouless , j. phys .
* c6 * 118 .
( 1973 ) + kosterlitz j. m. , j. phys .
* c7 * 1046 . ( 1974 ) + kupferman r. and a.j .
chorin , _ a numerical study of the kosterlitz - thouless transition in a two - dimensional coulomb or vortex gas _ , lbnl-40368/1997 .
+ minnhagen p. , phys .
* b32 * 3088 . ( 1985 ) + nndori i. , j. polonyi , k. sailer , hep - th/9910167 , accepted to phys .
d + polonyi j. and k. sailer , _ renormalization of composite operators _ , hep - th/0011083 + wegner f.j . and a. houghton , phys . rev . * | the rg flow for the sine - gordon model is determined by means of the method of wegner and houghton in next - to - leading order of the derivative expansion . for small values of the fugacity
this agrees with the well - known rg flow of the two - dimensional coulomb - gas found in the dilute gas approximation and a systematic way of obtaining higher - order corrections to this approximation is given .
# 1([#1 ] ) |
the luminosity function ( lf ) of quasars is fundamental to cosmology , but even after more than 30 years of study , its relation to the intrinsic luminosities of quasars remains poorly understood .
previous work modeling the quasar lf has relied on restrictive assumptions about lifetimes and light curves of quasars , imagining , for example , that quasars either have universal lifetimes or that they evolve exponentially , usually on the galaxy dynamical time or the @xmath0-folding time for eddington - limited black hole growth @xmath1yr for accretion with radiative efficiency @xmath2 @xcite . under these circumstances ,
the distribution of quasars with a given mass or peak luminosity is trivially related to the observed lf ( in the absence of selection effects ) , and the two have essentially identical shape .
recently , simulations of galaxy mergers incorporating black hole growth and feedback @xcite have reproduced the @xmath3 relation between black hole mass and galaxy velocity dispersion @xcite and linked quasar activity @xcite to galaxy evolution @xcite .
the simulations predict more complicated quasar light curves than have been adopted previously .
the peak , exponential black hole growth phase is determined by the gas supply over timescales @xmath4 and shuts down when significant gas is expelled by feedback .
the light curves have been studied by @xcite , who showed that the self - termination process gives observable lifetimes @xmath5yr for bright optical quasars , in good agreement with observations , and yields a large population of obscured sources as a natural stage of quasar evolution . @xcite analyzed simulations over a range of galaxy masses and found that the quasar light curves and lifetimes are always qualitatively similar , with both the intrinsic and observed quasar lifetimes being strongly decreasing functions of luminosity , with longer lifetimes at all luminosities for higher - mass ( higher peak luminosity ) systems . here , we use our quasar lifetimes and light curves to deconvolve the observed quasar lf from the distribution of peak luminosities , and find that they differ qualitatively , unlike for the trivial light curves or even complex , cosmologically evolving light curves or distributions of eddington ratios that have been employed previously , demanding a new interpretation of the quasar lf .
up to now , theoretical studies of the quasar lf have generally employed very simple descriptions of the quasar light curve , namely some variant of a `` feast or famine '' or `` light bulb '' model ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ? * ) , in which quasars have only two states : `` on '' or `` off . '' when `` on , '' quasars have a constant luminosity and radiate at a characteristic eddington ratio @xmath6 , generally in the range @xmath7 .
this state is assumed to last for a universal lifetime @xmath8 , which is the same for all quasars at a given redshift ( although some models allow for redshift evolution of @xmath8 ) .
the value of @xmath8 is an input parameter of the models , generally adopted from observations or assumed to be related to the salpeter time or the dynamical time of the host galaxy .
the time spent in a logarithmic luminosity interval is then just @xmath9 .
although this approach is analytically simple , with the trivial light curve @xmath10 where @xmath11 is the heaviside step function , it has no strong theoretical or observational motivation . for present purposes , models where quasars live arbitrarily long with slowly evolving mean volume emissivity or mean light curve ( e.g. * ?
* ; * ? ? ?
* ; * ? ? ?
* ) are equivalent to the `` light bulb '' scenario , as they still assume that quasars observed at a luminosity @xmath12 radiate at that approximately constant luminosity over some universal lifetime @xmath8 at a particular redshift .
we also consider a variant of the `` light bulb '' scenario , which we term the `` eddington limited '' model , where a black hole accretes at a fixed eddington ratio @xmath13 from an initial mass @xmath14 to a final mass @xmath15 ( or equivalently , a final luminosity @xmath16 ) , and then shuts off .
this gives exponential mass and luminosity growth , with the light curve @xmath17 where @xmath18 for @xmath19 .
the time spent in any logarithmic luminosity bin is constant , @xmath20 for @xmath21 .
this is true for any exponential light curve @xmath22 , such as that of @xcite , with only the normalization @xmath23 changed , and thus any such model will give identical results with correspondingly different normalizations .
while such a light curve allows the black hole mass and luminosity to change , there is no theoretical expectation that black hole growth is well - approximated by a constant eddington ratio .
we compare these scenarios to a physically motivated model using quasar lifetimes derived from simulated galaxy mergers @xcite .
the light curves from the simulations are complex , generally having periods of rapid accretion after `` first passage '' of the galaxies , followed by an extended quiescent period , then a transition to a peak , highly luminous quasar phase , and then a dimming as self - regulated mechanisms expel gas from the remnant center after the black hole reaches a critical mass @xcite .
in addition , the accretion rate at any time can be variable over small timescales @xmath24 , but even with these complexities , the statistical nature of the light curve can be described by simple forms .
@xcite find that the total quasar lifetime @xmath25 above a given luminosity @xmath12 is well - approximated by a truncated power law , with @xmath26 where @xmath27 , over the range @xmath28 for a given quasar .
@xmath29 is , as above , determined by the final black hole mass , @xmath30 . over a wide range of @xmath29 ( from @xmath31 ) ,
@xcite also find that @xmath32 is a function of @xmath15 ( or @xmath29 ) , given by @xmath33 , with @xmath34 ( the approximate slope of the eddington - limited case ) as an upper limit .
this reflects the fact that larger quasars have shallower slopes as they spend more time at higher luminosities up to some larger peak luminosity .
the time spent in any logarithmic luminosity interval in this range is then simply @xmath35 @xcite .
the simulation lifetime ( solid line ) is compared to the expected eddington - limit model lifetime ( dotted ) , the power law approximated lifetime ( dashed ) , and the mixed power law - eddington model ( dash - dot ) , from the formulae in 2 .
[ fig : lf.lightcurve],width=336 ] in our picture , quasars spend far more time at low luminosities than at their peaks .
this is a consequence of both pure @xmath36 eddington - limited growth up to a final black hole mass as well as a `` brightening '' and `` dimming '' out of and into extended quiescent phases of black hole growth around the peak quasar phase , during which the eddington ratio changes with time in the range @xmath37 .
we also consider a more complicated fit described in @xcite , where quasar lifetimes follow the eddington - limited model at the brightest luminosities , down to some @xmath38 , below which they are described by a power law .
in all these cases , the best - fit break luminosity is a nearly constant fraction of the peak luminosity @xmath39 .
these fits have the advantage of being accurate at both low and high luminosities , down to lifetimes @xmath40myr . in figure 1 , we show an example of the light curve from one of our simulations ( top ) and the relation between the lifetime and luminosity and various fits to it ( bottom ) .
our model contrasts sharply with both the `` light bulb '' and `` eddington limited '' cases .
the `` light bulb '' scenario ignores the evolution of the quasar as it accretes , and neglects the vast majority of its life at luminosities below its final peak .
the `` eddington limited '' model may be a reasonable approximation during the bright quasar phase near the final peak luminosity , but it underestimates the time spent at luminosities @xmath41 by as much as two orders of magnitude @xcite .
furthermore , in our model , quasars can evolve significantly on short timescales ( @xmath42myr ) , which means that our interpretation yields a statistical mix of contributions from different peak luminosities and eddington ratios at any observed luminosity at a given redshift , which contrasts strongly with any model in which the quasar light curve or emissivity evolves over cosmic time .
we note that our arguments do not in any way depend on the obscuration model described in @xcite , as we consider only low enough redshifts that quasar lfs should be reasonably complete and therefore only `` intrinsic '' quasar lifetimes are relevant . incorporating such effects , however , does not qualitatively change our conclusions .
moreover , the simulations from which we derive our lifetimes cover the complete range of intrinsic and observed luminosities of the lfs considered here , and show smoothly - changing properties in good agreement with the fits described above for all simulations in this range .
given quasar lifetimes as functions of luminosity and peak luminosity , the lf is a convolution of the lifetime with the intrinsic distribution of sources with a given @xmath29 , @xmath43 here , @xmath44 is the rate at which sources in a logarithmic interval in @xmath29 are created or `` activated '' per unit volume at some redshift , and @xmath45 is the observed number density of sources per logarithmic interval in @xmath12 .
this formulation implicitly accounts for the `` duty cycle '' ( the fraction of active quasars at a given time ) , which is proportional to the lifetime at a given luminosity .
as we do not attempt to model the number of halos containing quasars and wish to consider wide classes of lifetime models ( which may have different normalizations ) , we do not concern ourselves with the absolute normalization of @xmath45 or @xmath44 .
for the `` light bulb '' and `` eddington limited '' scenarios respectively , this convolution and the corresponding deconvolution are trivial , giving @xmath46 @xmath47 for the intrinsic @xmath29 distributions .
@xcite use a similar method to de - convolve the distribution of sources and consider in detail a very wide range of models for the quasar light curve and distribution of eddington ratios , but none of the models they consider include the fundamentally important feature of the lifetime increasing with decreasing luminosity to the lowest observed quasar luminosities , and further they do not allow the eddington ratio distribution to depend on final black hole mass . as a result , every example in the range of models they consider , as well as those in previous works , results in an @xmath44 distribution qualitatively similar to the observed lf , increasing monotonically to lower luminosities ( in the absence of an arbitrary truncation ) .
however , in the case of our luminosity - dependent lifetimes , we must solve numerically for @xmath48 , fitting to a given lf , and find a fundamentally different result .
, from fitting to the @xcite lf ( diamonds ) at redshift @xmath49 .
upper panel shows the fitted lf , lower panel the @xmath50 distribution for `` light bulb '' models ( dashed ) , `` eddington limited '' models ( dotted ) , and luminosity dependent lifetimes fitted to power laws ( thin ) and fitted to combined power laws and eddington growth ( thick ) .
[ fig : lf.normal],width=336 ] figure [ fig : lf.normal ] shows the best - fit @xmath44 distributions for the above cases fitted to the @xcite optical quasar lf at redshift @xmath49 , as well as the best - fit lf . for luminosity - dependent lifetimes , we take a power law dependence on @xmath29 from @xcite , with @xmath51 for pure truncated power - law fits and @xmath52 for fits with a power - law mapped onto an eddington - limited fit at near - peak luminosities . for ease of comparison with black hole mass and eddington ratios ,
we rescale all b - band quantities to the bolometric luminosity using the corrections of @xcite . in all cases ,
it is possible to reproduce the double - power law lf and `` break '' luminosity @xmath53 quite accurately , and the @xmath44 distributions are similar at luminosities above @xmath53 .
however , below @xmath53 , a proper accounting of the luminosity dependence of quasar lifetimes results in a radically different @xmath44 distribution , with the shallow slope of the lf a direct reflection of the slope in the lifetime vs. luminosity relation , dominated by quasars at the peak of the @xmath44 distribution , which then determines the `` break '' luminosity of the observed lf , with @xmath54 . note that the steepness of the @xmath44 cutoff is an artifact of extrapolating the lf to luminosities well below those observed
any strong decline below @xmath53 gives an equivalent fit to the actual data .
it is possible that objects which have already undergone a quasar phase may re - activate , or that black holes may already be very large when agn activity first begins . to determine whether or not this changes our results
, we consider the most extreme possible case , in which we return to the simulations of @xcite and re - fit the quasar lifetimes but ignore all black hole activity prior to the final merger .
this neglects all early accretion activity as black holes build up to a significant mass , as well as weaker , sub - eddington accretion during the merger as gravitational torques from early merger stages funnel gas to the galaxy centers .
therefore , we expect this case to set a strong lower limit to the lifetime at low luminosities , regardless of the mechanism driving quasar activity . however , we still find that these lifetimes are well above the eddington - limited model expectations , albeit with shallower best fit power law slopes @xmath55 and a slightly lower ( constant ) normalization . it has also been argued from observations of stellar black hole binaries that a transition between accretion states occurs at a critical eddington ratio @xmath56 , from radiatively inefficient accretion flows at low accretion rates ( e.g. , * ? ? ?
* ) to radiatively efficient accretion through a standard @xcite disk . although the critical eddington ratio for supermassive black holes is uncertain , observations of black hole binaries @xcite as well as theoretical extensions of accretion models ( e.g. , * ? ? ?
* ) suggest @xmath57 .
we examine whether this can change our conclusions by re - fitting quasar lifetimes from our simulations . because we assumed a constant radiative efficiency @xmath58 with @xmath59 , we account for this effect by multiplying the simulation luminosity at all times by an additional `` efficiency factor '' @xmath60 which depends on the eddington ratio @xmath61 ( given an a priori constant assumed efficiency , @xmath62 always ) , @xmath63 this choice for the efficiency factor follows from adaf models @xcite and ensures that the radiative efficiency is continuous at the critical eddington ratio @xmath64 .
we expect this to set an extreme limit to the impact of this transition , as we do not incorporate this effect dynamically in the simulations , which would slow `` blowout '' after the peak quasar phase and increase lifetimes at low luminosities .
we again find that power - law lifetimes at low luminosities are much larger than the expectation from the eddington - limited model , with best - fit slopes @xmath65 . , lower panel , but with lifetimes recalculated considering only times after the onset of peak quasar activity ( thick ) , or decreasing the radiative efficiency below a critical eddington ratio ( thin ) .
the dot - dashed line uses lifetimes calculated including both effects .
`` light bulb '' ( dashed ) and `` eddington limited '' ( dotted ) models are shown for comparison .
triangles show the distribution of black hole masses expected from the distribution of early - type velocity dispersions @xcite and the @xmath3 relation from @xcite .
[ fig : lf.limits],width=336 ] figure [ fig : lf.limits ] shows the best - fit @xmath44 distributions and lfs for these cases , in the manner of figure [ fig : lf.normal ] .
we also plot as an extreme limit the results obtained applying both the radiative efficiency correction above and considering only times during and after the final merger .
this case gives the shallowest slopes , @xmath66 .
although the shallower slopes broaden the @xmath44 distribution , our results are qualitatively unchanged .
we have further considered extreme cases such as making our original slopes shallower by @xmath67 , a factor @xmath68 , and find nearly identical results .
we consider realistic quasar light curves and lifetimes derived from hydrodynamical simulations of quasar activity , and find that this results in a qualitatively different form and evolution of the distribution in peak quasar luminosities @xmath44 ( equivalently , final black hole masses or host system properties ) than that derived or implied in previous works ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
we describe the resulting new interpretation of the quasar luminosity function ( lf ) in which the steep bright - end consists of quasars radiating near their eddington limits and is directly related to the distribution of intrinsic peak luminosities ( or final black hole masses ) as has been assumed previously .
however , the shallow , faint - end of the lf describes black holes either growing efficiently in early stages of activity or in extended , quiescent states going into or coming out of a peak bright quasar phase , with eddington ratios generally between @xmath69 and 1 .
these are not objects which were active at some earlier cosmic time and have since `` shut down '' , as are invoked in some `` light bulb '' models , but correspond to the evolution of quasars in mergers over short timescales and critically related ( both physically and statistically ) to the observed bright , eddington - limited quasar phase .
the `` break '' luminosity in the lf corresponds directly to the _ peak _ in the distribution of intrinsic quasar properties @xmath44 . as such , observations of this break and
its evolution present a powerful new probe of the quasar population .
we find that the observed quasar lf can be reproduced to high accuracy with a model of quasar lifetimes which depends on both the observed luminosity and peak luminosity of a quasar , for some distribution of peak luminosities @xmath44 .
we demonstrate this for the @xcite lf at @xmath49 , but find identical qualitative results for the @xcite and @xcite soft and hard x - ray lfs , over a wide range of redshifts @xmath70 .
this is not trivial , as it is in the case of simpler models of quasar lifetimes ( in which the @xmath44 distribution can be directly recovered from the lf )
. if the faint - end slope of the lf is too shallow , power - law luminosity dependent lifetimes will be too steep to reproduce it . if the faint - end slope is too steep , our lifetime model can reproduce the lf , but without a characteristic qualitative difference in behavior from `` light bulb '' or eddington - limited models .
in fact , it is a strong argument in favor of our picture that observed faint - end slopes lie just within the expected range , for any peaked population @xmath44 .
our new interpretation of the quasar lf immediately explains several other observations .
the distribution of velocity dispersions in early - type systems ( expected to have undergone merger - driven agn activity ) from @xcite turns over and decreases below @xmath71 , which from the @xmath3 relation implies that the black hole mass distribution for remants of major mergers of massive galaxies should turn over and decrease below @xmath72 ( @xmath73 ) .
moreover , @xcite find that black holes with masses less than @xmath74 must be rare at high redshift to avoid significantly overpredicting the counts of bright radio sources .
these observations are exactly as predicted by our deconvolution of the lf , where at high redshift @xmath44 is shifted to higher peak luminosity ( black hole mass ) and thus fewer low - mass black holes are produced .
the @xmath44 distribution based on traditional models of quasar lifetimes , however , would necessarily predict that the vast majority of quasars , especially at high redshift where the number of quasars is larger , have final black hole masses well below this limit , with the number density increasing towards lower black hole masses .
the shift of @xmath44 with time in our model can explain the entire observed range of the @xmath75 black hole mass function ( e.g. , * ? ? ?
the distribution of observed eddington ratios is also a natural consequence of our model , and does not invoke a particular probability distribution across sources to match observations .
quasars observed around and below the `` break '' in the lf ( which dominate the total population ) may be accreting at the eddington rate , as they build up early in their lives , or may be in quiescent phases going into or out of their peak quasar activity . for these quiescent phases , both comparison of the lifetime power laws in 3 and direct calculation of the lifetime as a function of eddington
ratio @xcite show that they are dominated by relatively large eddington ratios @xmath76 .
below @xmath77 , quasars at the peak of the @xmath44 distribution will be below limits measured in most surveys .
therefore , we expect a distribution of eddington ratios concentrated between @xmath78 , with a small tail down to @xmath69 and mean eddington ratios increasing with luminosity , exactly as observed @xcite
. the sharp contrast between the intrinsic distribution of peak quasar luminosities and the observed lf provides clean and elegant tests of our theory . in our interpretation ,
the bright and faint ends of the lf correspond statistically to similar mixes of galaxies , but in various stages of evolution .
however , in all other competing scenarios , the quasar luminosity is directly related to the mass of the host galaxy .
therefore , any observational probe that differentiates quasars based on their host galaxy properties such as , for example , the dependence of clustering of quasars on luminosity , can be used to discriminate our picture from older models .
early evidence from both quasar - quasar @xcite and quasar - galaxy @xcite correlations support this picture and suggest that quasar hosts have a well - defined characteristic mass , exactly as predicted by our model with a peaked @xmath44 and in stark contrast to previous models in which @xmath44 continues to increase below the break in the luminosity function .
our modeling of realistic quasar lifetimes and the resulting new @xmath44 distribution and interpretation of the quasar luminosity function have a wide range of implications , .
theoretical predictions of the active and relic supermassive black hole density and mass functions , the contribution of quasars to the x - ray and infrared background , the observed eddington ratio distribution as a function of luminosity and redshift , the evolution of characteristic black hole masses and quasar host galaxy masses with redshift , quasar correlation functions and bias , the relation between quasar and starburst or luminous infrared galaxy luminosity functions , the role of quasars in reionization , the population of high - redshift radio sources , means to discriminate between pure density and pure luminosity evolution of quasar populations at very high redshift , and the distribution and evolution of quasar host properies all depend on this quantity and must be revised in this new model .
semi - analytical models and cosmological simulations have focused on reproducing @xmath44 distributions qualitatively similar to the observed lf based on idealized models of the quasar lifetime , but should instead attempt to reproduce the fundamentally different @xmath44 distribution implied by our realistic , physically motived models of the quasar light curve and luminosity - dependent quasar lifetimes .
this work was supported in part by nsf grants aci 96 - 19019 , ast 00 - 71019 , ast 02 - 06299 , and ast 03 - 07690 , and nasa atp grants nag5 - 12140 , nag5 - 13292 , and nag5 - 13381 .
the simulations were performed at the center for parallel astrophysical computing at the harvard - smithsonian center for astrophysics .
adelberger , k. l. & steidel , c. c. 2005 , apj , in press [ astro - ph/0505210 ] boyle , b. j. , shanks , t. , croom , s. m. , smith , r. j. , miller , l. , loaring , n. , & heymans , c. 2000 , , 317 , 1014 croom , s. m. , et al .
2005 , , 356 , 415 di matteo , t. , springel , v. , & hernquist , l. 2005 , nature , 433 , 604 esin , a. a. , mcclintock , j. e. , & narayan , r. 1997 , , 489 , 865 haiman , z. , & loeb , a. 1998 , , 503 , 505 haiman , z. , & menou , k. 2000 , , 531 , 42 haiman , z. , quataert , e. , & bower , g. c. 2004 , , 612 , 698 hartwick , f. d. a. , & schade , d. 1990 , , 28 , 437 hopkins , p. f. , hernquist , l. , martini , p. , cox , t. j. , robertson , b. , di matteo , t. , & springel , v. 2005a , , accepted [ astro - ph/0502241 ] hopkins , p. f. , hernquist , l. , cox , t. j. , di matteo , t. , martini , p. , robertson , b. , & springel , v. 2005b , , submitted [ astro - ph/0504190 ] kauffmann , g. , & haehnelt , m. 2000 , , 311 , 576 maccarone , t. j. 2003 , , 409 , 697 marconi , a. , risaliti , g. , gilli , r. , hunt , l. k. , maiolino , r. , & salvati , m. 2004 , , 351 , 169 meyer , f. , liu , b. f. , & meyer - hofmeister , e. 2000 , , 354 , l67 miyaji , t. , hasinger , g. , & schmidt , m. 2000 , , 353 , 25 narayan , r. & yi , i. 1995 , , 452 , 710 salpeter , e. e. 1964 , , 140 , 796 schmidt , m. 1968 , apj , 151 , 393 schmidt , m. & green , r. f. 1983 , apj , 269 , 352 shakura , n. i. & sunyaev , r. a. 1973 , , 24 , 337 sheth , r. k. , et al .
2003 , , 594 , 225 small , t. a. , & blandford , r. d. 1992 , , 259 , 725 springel , v. , di matteo , t. , & hernquist , l. 2005a , , in press , [ astro - ph/0409436 ] springel , v. , di matteo , t. , & hernquist , l. 2005b , , submitted , [ astro - ph/0411108 ] steed , a. & weinberg , d. h. 2003 , in press [ astro - ph/0311312 ] tremaine , s. , et al .
2002 , , 574 , 740 ueda , y. , akiyama , m. , ohta , k. , & miyaji , t. 2003 , , 598 , 886 vestergaard , m. 2004 , , 601 , 676 volonteri , m. , haardt , f. , & madau , p. 2003
, , 582 , 559 warren , s. j. , hewett , p. c. , & osmer , p. s. 1994 , , 421 , 412 wyithe , j. s. b. , & loeb , a. 2003 , , 595 , 614 | we propose a new interpretation of the quasar luminosity function ( lf ) , derived from physically motivated models of quasar lifetimes and light curves . in our picture , quasars evolve rapidly and their lifetime depends on both their instantaneous _ and _ peak luminosities .
we study this model using simulations of galaxy mergers that successfully reproduce a wide range of observed quasar phenomena . with lifetimes inferred from the simulations , we deconvolve the observed quasar lf from the distribution of peak luminosities , and show that they differ qualitatively , unlike for the simple models of quasar lifetimes used previously .
we find that the bright end of the lf traces the intrinsic peak quasar activity , but that the faint end consists of quasars which are either undergoing exponential growth to much larger masses and higher luminosities , or are in sub - eddington quiescent states going into or coming out of a period of peak activity .
the `` break '' in the lf corresponds directly to the maximum in the intrinsic distribution of peak luminosities , which falls off at both brighter and fainter luminosities .
our interpretation of the quasar lf provides a physical basis for the nature and slope of the faint - end distribution , as well as the location of the break luminosity . |
the implications of neutrino masses in cosmology and astrophysics are numerous .
one fascinating effect , is that of relic neutrino asymmetries , which may be produced if there is mixing between active and sterile neutrino species @xcite .
neutrino asymmetries are interesting since they affect big bang nucleosynthesis ( bbn ) , altering the primordial helium yield .
bbn is particularly sensitive to an asymmetry between the @xmath0 and @xmath1 .
additionally , since the matter potential arising from forward scattering is proportional to the asymmetry , a large asymmetry will suppress active - sterile oscillation modes which could otherwise equilibrate sterile neutrino species around the time of bbn .
first consider two - flavour mixing between an active neutrino , @xmath2 , and a sterile neutrino , @xmath3 . for each neutrino flavour
we define an asymmetry @xmath4 where @xmath5 and @xmath6 are the neutrino and photon number densities , respectively .
the neutrino momentum distributions , @xmath7 , are such that @xmath8 and in thermal equilibrium are just given by fermi - dirac distributions @xmath9 here @xmath10 is the temperature and @xmath11 , with @xmath12 being the neutrino chemical potential . in equilibrium , @xmath13 is related to @xmath12 as per @xmath14.\ ] ] we wish to track @xmath13 as the universe evolves , with the period of interest being the epoch between @xmath15 and @xmath16 .
the evolution of the neutrino ensemble will be controlled by ( i ) oscillations , ( ii ) collisions of the neutrinos with other particles , resulting in decoherence of the oscillations , and ( iii ) the expansion of the universe , which redshifts energies and momenta .
the matter affected mixing angle , @xmath17 , is @xmath18 where @xmath19 is the mixing angle in vacuum .
the term @xmath20 arises from the effective matter potential @xmath21 , where -/+ corresponds to @xmath22 and @xmath23 is the mass squared difference of the two neutrino mass eigenstates .
the functions @xmath24 and @xmath25 are given by @xmath26 where @xmath27 @xcite .
the function @xmath28 is defined as @xmath29 where @xmath30 is due to the baryon asymmetry . note that the mixing angle is dependent on the size of the asymmetry , which makes the evolution of the asymmetry non - linear .
there are two distinct phases in the growth of a neutrino asymmetry .
the first is a collision dominated stage , which occurs at high temperature .
the total collision rate of a @xmath0 ( @xmath31 or @xmath32 ) of momentum @xmath33 with particles in the background plasma is approximately @xmath34 these collisions tend to destroy the coherent superposition of flavour states . in this regime , it can be shown that the growth of the neutrino asymmetry is governed by the equation @xmath35[x+(c - b - a)^2]},\end{aligned}\ ] ] where @xmath36 , @xmath37 , @xmath38 ^ 2 , \;\;\ ; n^{+ } = \frac{1}{2}(n_{\nu}+ n_{\overline{\nu}}).\ ] ] eq.([hightemp ] ) can be derived either from a heuristic picture @xcite that views collisions as measurement - like interactions , or can be extracted from the exact quantum kinetic equations @xcite .
it is valid when the collision rate is large and the evolution is adiabatic - see @xcite for details . note that @xmath39 , which can lead to runaway positive feedback and thus rapid growth of @xmath13 .
the mixing parameters for which this occurs are specified by @xmath40 and @xmath41 , and we require lighter of the two mass eigenstates to approximately coincide with the @xmath3 . typically , the asymmetry will reach values @xmath42 in this phase . note that the sign of the asymmetry , which depends on the initial conditions , can be ambiguous @xcite .
the second phase is that lower temperature .
here the collisions are less important , and coherent msw transitions drive the asymmetry to values @xmath43 @xcite .
if , for example , the asymmetry is positive , the neutrino msw resonance occurs at a much higher momentum value than that of the antineutrinos and may be neglected .
the rate at which the asymmetry grows is then @xmath44 where @xmath45 is the difference between the number of @xmath46 and @xmath47 at the resonance .
eq.([res ] ) quantifies the rate at which the resonance moves through the momentum distribution , and holds when transitions are adiabatic and the resonance width is small .
we wish to determine the impact of such asymmetry production on the bbn light element abundances , and in particular , on the helium yield . in general , the results are somewhat model dependent ( compare , say , @xcite with @xcite . )
the results are particularly sensitive to the size of the @xmath0 asymmetry , which directly affects the neutron / proton ratio through the reaction rates @xmath48 the n / p ratio determines the helium mass fraction @xmath49 via the equation @xmath50 for typical values of @xmath23 , one must compute the growth of the asymmetry and the effect on bbn simultaneously , as the asymmetries will not have reached their final values before the bbn epoch commences .
momentum modes which are depleted due to the oscillations , are refilled by scattering and annihilation processes as per @xmath51 where @xmath52 is the equilibrium distribution in eq.([nic ] ) , and @xmath53 .
the example we consider is a particular `` 2 + 2 '' style 4-neutrino model , consisting of a maximally mixed @xmath54 pair , separated by a mass gap from @xmath55 .
the oscillations modes most important in the evolution of the various flavour neutrino asymmetries will be the ones with the largest @xmath23 .
specifically , the @xmath56 oscillation modes will create large @xmath57 and @xmath58 asymmetries , and the @xmath59 oscillation modes will transfer a small amount of this asymmetry to the electron neutrinos . see @xcite for further details . integrating the bbn reaction rates , eq . (
[ eq : rates ] ) , the direct effect of the @xmath0 asymmetry on the @xmath62 ratio , and hence @xmath49 , is determined .
we obtain @xmath63 , corresponding to an effective change in the number of neutrinos of @xmath64 .
in addition , there is a small direct change to the energy density , leading overall to @xmath65 for @xmath66 .
the value of @xmath67 obtained in this way must then be compared with that calculated from the observed light abundances @xcite .
it is not yet clear whether the observed helium and deuterium abundances can be accommodated within the standard bbn scenario with @xmath68 , and it is possible that a small @xmath0 asymmetry such as that discussed here would help to provide a better fit @xcite .
recent measurement of the cosmic microwave background radiation ( cmbr ) anisotropy also provides a determination of the number of neutrinos in the early universe @xcite , although the constraints are , as yet , not as strong as those arising from bbn .
asymmetry , while the cmbr measures the total energy density residing in neutrinos ( at a much later epoch of the universe . ) ] however , future measurements are expected to pin down the effective neutrino number very precisely @xcite and thus will be of great use in constraining models which feature the mixing of active and sterile neutrinos . n. f. bell , r. r. volkas and y. y. y. wong , phys . rev .
* d59 * , 113001 ( 1999 ) ; r. r. volkas and y. y. y. wong , phys . rev . * d62 * 093025 ( 2000 ) ; k. s. m. lee , r. r. volkas and y. y. y. wong , phys . rev .
* d62 * 093026 ( 2000 ) . | we outline how relic neutrino asymmetries may be generated in the early universe via active - sterile neutrino oscillations .
we discuss possible consequences for big bang nucleosynthesis , within the context of a particular 4-neutrino model . |
the late - time dynamics of astronomical binaries consisting of neutron stars and/or black holes is dominated by relativistic motion and non - linear general relativistic effects .
the component bodies would be accelerated to velocities close to half the speed of light before they plunge towards each other , resulting in a violent event during which the source would be most luminous in the gravitational window .
such events are prime targets of interferometric gravitational wave ( gw ) detectors like ligo / virgo / geo / tama that are currently taking data at unprecedented sensitivity levels and bandwidths @xcite .
binary coalescences are the end state of a long period of adiabatic dynamics in which the orbital frequency of the system changes as a result of gravitational radiation backreaction but the change in frequency per orbit is negligible compared to the orbital frequency itself .
indeed , the adiabatic inspiral phase is well - modelled by the post - newtonian ( pn ) approximation to einstein s equations but this approximation becomes less accurate close to the merger phase .
additionally , there are different ways of casting the gravitational wave phasing formula the formula that gives the phase of the emitted gravitational wave as a function of time and the parameters of the system .
these different approaches make use of the post - newtonian expansions of the binding energy and gravitational wave luminosity of the system .
the standard approach in deriving the phasing formula uses the _
specific _ gravitational binding energy @xmath2 ( i.e. the binding energy per unit mass ) of the system and its luminosity @xmath3 both to the same relative accuracy @xcite . including the radiation reaction at dominant order , however , is not a first order correction to the dynamics of the system , rather it is a correction that arises at @xmath4,$ ] where @xmath5 is the post - newtonian expansion parameter describing the velocity in the system and @xmath6 is the speed of light .. ] thus , the phasing of the waves when translated to the relative motion of the bodies implies that the dynamics is described by the dominant newtonian force and a correction at an order @xmath7 but neglecting conservative force terms that occur at orders @xmath8 and @xmath9 such considerations have led to an approximation scheme in which one constructs the phasing of gravitational waves using the following ordinary , coupled differential equations : @xmath10 where @xmath11 and @xmath12 is the total mass of the binary . the phasing obtained by numerically solving the above set of differential equations
is called the _ taylort1 _
approximant @xcite .
if the detector s motion can be neglected during the period when the wave passes through its bandwidth then the response of the interferometer to arbitrarily polarized waves from an inspiralling binary is given by @xmath13 , \label{eq : waveform1}\ ] ] where @xmath14 is defined so that it is zero when the binary coalesces at time @xmath15 @xmath16 is the phase of the signal at @xmath17 @xmath18 is the symmetric mass ratio , @xmath19 is the distance to the source and @xmath20 is a numerical constant whose value depends on the relative orientations of the interferometer and the binary orbit .
it suffices to say for the present purpose that for an optimally oriented source @xmath21 one can compute the fourier transform @xmath22 of the waveform given in eq .
( [ eq : waveform1 ] ) using the stationary phase approximation : @xmath23 } , \label{eq : waveform2}\ ] ] where the phase of the fourier transform obeys a set of differential equations given by @xmath24 in the above expressions , including the post - newtonian expansions of the energy and flux functions , the parameter @xmath25 .
the waveform eq.([eq : waveform2 ] ) computed by numerically solving the differential equations eq.([eq : phasing2 ] ) is called _ taylorf1 _ @xcite approximant . before we proceed further ,
let us recall the notation used in post - newtonian theory to identify different orders in the expansion . in the conservative dynamics of the binary , wherein there is no dissipation , the energy is expressed as a post - newtonian expansion in @xmath26 with the dominant order termed newtonian or 0pn and a correction at order @xmath27 @xmath28 called @xmath29pn , with the dynamics involving only even powers of @xmath30 when dissipation is added to the dynamics , then the equation of motion will have terms of both odd and even powers of @xmath31 .
thus , a correction of order @xmath32 is termed as @xmath33pn . in the adiabatic approximation of a test - particle orbiting a schwarzschild black hole ,
the energy function @xmath2 is exactly calculable analytically , while the flux function @xmath34 is exactly calculable numerically @xcite .
in addition , @xmath34 has been calculated analytically to 5.5pn order @xcite by black hole perturbation theory @xcite .
in contrast , in the case of a general binary including bodies of comparable masses , the energy function @xmath2 has been calculated recently to 3pn accuracy by a variety of methods @xcite .
the flux function @xmath3 on the other hand , has been calculated to 3.5pn accuracy @xcite up to now only by the multipolar - post - minkowskian method and matching to a post - newtonian source @xcite .
the gravitational wave flux arising from the lowest order quadrupole formula , that is the 0pn order flux , leads to an acceleration of order _
2.5pn _ in the equations - of - motion .
this far - zone computation of the flux requires a control of the dynamics , or acceleration , to only _
newtonian _ accuracy .
the lowest order gw phasing in the adiabatic approximation uses only the leading terms in the energy ( newtonian ) and flux ( quadrupolar ) functions . for higher order phasing ,
the energy and flux functions are retained to the same relative pn orders .
for example , at 3pn phasing , both the energy and flux functions are given to the same _ relative _ 3pn order beyond the leading newtonian order .
we refer to this usual physical treatment of the phasing of gws computed in the adiabatic approximation , and used in the current ligo / virgo / geo / tama searches for the radiation from inspiralling compact binaries , as the _ standard adiabatic _ approximation .
we will denote the @xmath29pn _ standard adiabatic _ approximant as @xmath35},{\cal f}_{n})$ ] , where @xmath36 $ ] denotes the integer part of @xmath37 . as a prelude to go beyond the standard adiabatic approximation ,
let us consider the phasing of the waves in terms of the equations of motion of the system . to this end
, it is natural to order the pn approximation in terms of its dynamics or acceleration . from the viewpoint of the dynamics ,
the leading order standard adiabatic approximation is equivalent to using the 0pn ( corresponding to 0pn conserved energy ) and 2.5pn ( corresponding to the newtonian or 0pn flux ) terms in the acceleration ignoring the intervening 1pn and 2pn terms .
a complete , mathematically consistent treatment of the acceleration , however , should include _ all _ pn terms in the acceleration up to 2.5pn , _ without any gaps_. we refer to the dynamics of the binary , and the resulting waveform , arising from the latter as the 0pn _ complete non - adiabatic _ approximation .
in contrast , the waveform arising from the former choice , with gaps in the acceleration at 1pn and 2pn , is referred to as the 0pn _ standard non - adiabatic _ approximation .
extension to higher - order phasing is obvious . at 1pn
the standard non - adiabatic approximation would involve acceleration terms at orders 0pn , 1pn , 2.5pn and 3.5pn , whereas the complete non - adiabatic approximation would additionally involve the 2pn and 3pn acceleration terms .
finally , we propose a simple extension of the above construction to generate a new class of approximants in the adiabatic regime . to understand the construction let us examine the lowest order case .
given the 0pn flux ( leading to an acceleration at 2.5pn ) , one can choose the energy function at 2pn ( equivalent to 2pn conservative dynamics ) instead of the standard choice 0pn ( equivalent to 0pn or newtonian conservative dynamics ) .
this is the adiabatic analogue of the complete non - adiabatic approximant .
extension to higher pn orders follows naturally .
for instance , corresponding to the flux function at 1pn ( 1.5pn ) , the dissipative force is at order 3.5pn ( 4pn ) , and , therefore , the conservative dynamics , and the associated energy function , should be specified up to order 3pn ( 4pn ) . in general , given the flux at @xmath29pn - order ,
a corresponding complete adiabatic approximant is constructed by choosing the energy function at order @xmath38$]pn , where as mentioned before , @xmath36 $ ] denotes the integer part of @xmath37 .
we refer to the dynamics of the binary and the resulting waveform arising from such considerations , as the _ complete adiabatic _ approximation .
we will denote the @xmath29pn _ complete adiabatic _ approximant as @xmath39},{\cal f}_{n})$ ] . before moving ahead
the following point is worth emphasizing : the standard adiabatic phasing is , by construction , consistent in the _ relative pn _ order of its constituent energy and flux functions , and thus _ unique _ in its ordering of the pn terms .
consequently , one can construct different inequivalent , but consistent , approximations as discussed in ref .
@xcite by choosing to retain the involved functions or re - expand them .
the complete adiabatic phasing , on the other hand , is constructed so that it is consistent in spirit with the underlying dynamics , or acceleration , rather than with the relative pn orders of the energy and flux functions .
consequently , it has a unique meaning only when the associated energy and flux functions are used _ without any further re - expansions _ when working out the phasing formula .
as a result , though the complete non - adiabatic approximant is more consistent than the standard non - adiabatic approximant in treating the pn accelerations , in the adiabatic case there is no rigorous sense in which one can claim that either of the approximants is more consistent than the other .
the important point , as we shall see is that , not only are the two approximants _ not _ the same but the new complete adiabatic approximants are closer to the exact solution than the standard adiabatic approximants . in our view
, these new approximants should be of some interest .
they are simple generalizations of the _ standard adiabatic _ approximants coding information of the pn dynamics beyond the standard approximation without the need for numerical integration of the equations of motion
. they should be appropriate approximants to focus on when one goes beyond the adiabatic picture and investigates the differences stemming from the use of more complete equations of motion ( see section [ sec : non - adiabatic models ] ) . in the case of comparable mass binaries , the energy function
is currently known up to 3pn order and hence it would be possible to compute the complete adiabatic phasing of the waves to only 1pn order .
one is thus obliged in practice to follow the standard adiabatic approximation to obtain the phasing up to 3.5pn order .
consequently , it is a relevant question to ask how ` close ' are the complete and standard adiabatic approximants .
the standard adiabatic approximation would be justified if we can verify that it produces in most cases a good lower bound to the mathematically consistent , but calculationally more demanding , complete adiabatic approximation . in this paper
we compare the standard and complete models by explicitly studying their overlaps with the exact waveform which can be computed in the adiabatic approximation of a test mass motion in a schwarzschild spacetime .
the availability , in this case , of the exact ( numerical ) and approximate ( analytical ) waveforms to as high a pn order as @xmath40 , allows one to investigate the issue exhaustively , and provides the main motivation for the present analysis .
assuming that the comparable mass case is qualitatively similar and a simple @xmath41-distortion of the test mass case would then provide a plausible justification for the standard adiabatic treatment of the gw phasing employed in the literature - distortion of the test mass approximation is not universal .
in particular , blanchet @xcite has argued that the dynamics of a binary consisting of two bodies of comparable masses is very different from , and possibly more accurately described by post - newtonian expansion than , the test mass case . ] .
-derivative of ) energy function @xmath42 ( left ) , and flux function @xmath43 ( right ) in the test mass limit along with the exact functions ( denoted by @xmath44 ) .
also plotted is the amplitude spectral density ( per @xmath45 ) of initial ligo noise in arbitrary units.,width=480 ] the phasing formulas derived under the various adiabatic approximation schemes assume that the orbital frequency changes slowly over each orbital period . in other words , the change in frequency @xmath46 over one orbital period @xmath47
is assumed to be much smaller than the orbital frequency @xmath48 denoting by @xmath49 the time - derivative of the frequency , the adiabatic approximation is equivalent to the assumption that @xmath50 or @xmath51 this assumption becomes somewhat weaker , and it is unjustified to use the approximation @xmath52 when the two bodies are quite close to each other .
buonanno and damour @xcite introduced a non - adiabatic approach to the two - body problem called the _ effective one - body _ ( eob ) approximation . in this approximation one
solves for the relative motion of the two bodies using an effective hamiltonian with a dissipative force put - in by hand .
eob allows to extend the dynamics beyond the adiabatic regime , and the last stable orbit , into the plunge phase of the coalescence of the two bodies @xcite .
recently , buonanno , chen and vallisneri @xcite have studied a variant of the non - adiabatic model but using the effective lagrangian constructed in the post - newtonian approximation .
we shall use both the _ standard _ and _ complete _ non - adiabatic lagrangian models in this study and see how they converge to the exact waveform defined using the adiabatic approximation for a caveat in this approach . ] . in our study we will use the _ effectualness _ and _ faithfulness _ ( see below ) to quantify how good the various approximation schemes are .
there are at least three different contexts in which one can examine the performance of an approximate template family relative to an exact one .
firstly , one can think of a mathematical family of approximants and examine its convergence towards some exact limit .
secondly , one can ask whether this mathematical family of approximants correctly represents the gws from some physical system .
thirdly , how does this family of approximate templates converge to the exact solution in the sensitive bandwidth of a particular gw detector . in the context of gw data analysis
, the third context will be relevant and studied in this paper .
although there is no direct application to gw data analysis , equally interesting is the mathematical question concerning the behavior of different approximations , and the waveforms they predict , in the strongly non - linear regime of the dynamics of the binary , which is also studied in this paper .
the latter obviously does not require the details of the detector - sensitivity and it is enough to study the problem assuming a flat power spectral density ( i.e. a _ white - noise _ background ) for the detector noise . to summarize ,
our approach towards the problem will be two - pronged .
first , we will study the problem as a general mathematical question concerning the nature of templates defined using pn approximation methods .
we shall deal with four families of pn templates the standard adiabatic , complete adiabatic , standard non - adiabatic and complete non - adiabatic ( in particular , lagrangian - based ) approximants and examine their closeness , defined by using effectualness and faithfulness , to the exact waveform defined in the adiabatic approximation .
since this issue is a general question independent of the characteristics of a particular gw detector , we first study the problem assuming the white - noise case . having these results
, we then proceed to see how and which of these results are applicable to a specific detector , namely the initial ligo . by @xcite , @xmath53 $ ] where @xmath54 hz and the psd rises steeply below a lower cut - off @xmath55 hz . ] during the course of this study , we also attempt to assess the relative importance of improving the accuracy of the energy and flux functions by studying the overlaps of the pn templates constructed from different orders of energy and flux functions with the exact waveform .
it should be kept in mind that this work is solely restricted to the inspiral part of the signal and neglects the plunge and quasi - normal mode ringing phases of the binary @xcite . in order to measure the accuracy of our approximate templates
we shall compute their overlap with a fiducial exact signal .
we shall consider two types of overlaps @xcite .
the first one is the _ faithfulness _ which is the overlap of the approximate template with the exact signal computed by keeping the intrinsic parameters ( e.g. the masses of the two bodies ) of both the template and the signal to be the same but maximizing over the extrinsic ( e.g. the time - of - arrival and the phase at that time ) parameters .
the second one is the _ effectualness _ which is the overlap of the approximate template with the exact signal computed by maximizing the overlap over both the intrinsic and extrinsic parameters .
faithfulness is a measure of how good is the template waveform in both detecting a signal and measuring its parameters .
however , effectualness is aimed at finding whether or not an approximate template model is good enough in detecting a signal without reference to its use in estimating the parameters . as in previous studies ,
we take overlaps greater than 96.5% to be indicative of a good approximation . in the next section we study the test - mass waveforms in the adiabatic approximation .
we discuss the construction of the exact energy and flux functions as well as the respective t - approximants .
the overlaps of various standard adiabatic and complete adiabatic approximants are also compared in this section .
section [ sec : non - adiabatic models ] deals with the non - adiabatic approximation .
section [ sec : compmass ] explores the extension of the results in the comparable mass case .
it presents the energy and flux functions which are the crucial inputs for the construction of the fiducial ` exact ' waveform as well as the approximate waveforms followed by a discussion of the results . in the last section we summarize our main conclusions .
one of the main conclusions of this paper is that the effectualness of the test - mass approximants significantly improves in the complete adiabatic approximation at pn orders below 3pn .
however , standard adiabatic approximants of order @xmath1 3pn are nearly as good as the complete adiabatic approximants for the construction of effectual templates . in the comparable mass case
the problem can be only studied at the lowest two pn orders .
no strong conclusions can be drawn as in the test mass case .
still , the trends indicate that the standard adiabatic approximation provides a good lower bound to the complete adiabatic approximation for the construction of both effectual and faithful templates at pn orders @xmath1 1.5pn . from the detailed study of test - mass templates
we also conclude that , provided the comparable mass case is qualitatively similar to the test mass case , neither the improvement of the accuracy of energy function from 3pn to 4pn nor the improvement of the accuracy of flux function from 3.5pn to 4pn will result in a significant improvement in effectualness in the comparable mass case . as far as faithfulness is concerned , it is hard to reach any conclusion . to achieve the target sensitivity of 0.965 in effectualness corresponding to a 10% loss in the event - rate , standard adiabatic approximants of order 2pn and 3pn
are required for the @xmath56 and @xmath57 binaries , respectively , when restricting to only the inspiral phase .
( be warned that this is not a good approximation in the bh - bh case since the approach to the plunge and merger lead to significant differences . )
binary in white - noise .
different lines in the panels correspond to different orders of the flux function .
each line shows how the overlaps are evolving as a function of the accuracy of the energy function .
standard adiabatic approximants @xmath35},{\cal f}_{n})$ ] are marked with thick dots.,width=480 ] our objective is to compare the _ effectualness _ ( i.e larger overlaps with the exact signal ) and _ faithfulness _ ( i.e. smaller bias in the estimation of parameters ) of the standard adiabatic @xmath35},{\cal f}_{n})$ ] and complete adiabatic @xmath39},{\cal f}_{n})$ ] approximants . as a by - product of this study
, we would also like to have an understanding of the the relative importance of improving the accuracy of the energy function and flux function .
thus , what we will do is to take all possible combinations of t - approximants of energy and flux functions , construct pn templates and calculate the overlap of these templates with the exact waveform . in all cases ,
the exact waveform is constructed by numerically integrating the phasing formula in the time - domain [ _ taylort1 _ approximant , cf .
( [ eq : phasing1 ] ) and ( [ eq : waveform1 ] ) ] .
the waveforms ( both the exact and approximate ) are all terminated at @xmath58 which corresponds to @xmath59 hz for the @xmath60 binary , @xmath61 hz for the @xmath62 binary and @xmath63 hz for the @xmath64 binary is the velocity at the last stable circular orbit of schwarzschild geometry having the same mass as the total mass @xmath65 of the binary ( we adopt units in which @xmath66 ) and @xmath67 is the gw frequency corresponding to it . ] .
the lower frequency cut - off of the waveforms is chosen to be @xmath68 hz . in this study
, we restrict to approximants taylort1 and taylorf1 since they do not involve any further re - expansion in the phasing formula and hence there is no ambiguity when we construct the phasing of the waves using approximants with unequal orders of the energy and flux functions . except that the signal corresponds to a @xmath69 binary.,width=480 ] in the case of a test - particle @xmath70 moving in circular orbit in the background of a schwarzschild black hole of mass @xmath71 , where @xmath72 , the energy function @xmath73 in terms of the invariant argument @xmath74 is given by @xmath75 the associated @xmath5-derivative entering the phasing formula is @xmath76 we use the above exact @xmath77 to construct the exact waveform in the test - mass case . in order to construct various approximate pn templates , we taylor - expand @xmath78 and truncate it at the necessary orders .
\label { eq : e_taylor}\end{aligned}\ ] ] different t - approximants of the energy function @xmath80 along with @xmath81 are plotted in fig .
[ fig : tmenflfnsligo]@xmath82 . in the test - particle limit
, the exact gravitational - wave flux has been computed numerically with good accuracy @xcite .
we will use this flux function ( see fig .
[ fig : tmenflfnsligo]@xmath83 ) , along with the energy function given by eq .
( [ eq : eprimeexact ] ) , to construct an exact waveform in the test - mass case . in the test - particle limit
, the gw flux is also known analytically to 5.5pn order from black hole perturbation theory @xcite and given by @xmath84 , \label{eq : exactflux}\ ] ] where the various coefficients @xmath85 and @xmath86 are @xcite , we will use the energy and flux functions given by eq .
( [ eq : e_taylor ] ) eq .
( [ eq : exactfluxcoeffs ] ) to construct various approximate templates by truncating the expansions at the necessary order .
the different t - approximants of the flux function @xmath87 along with the ( numerical ) exact flux @xmath88 are plotted in fig .
[ fig : tmenflfnsligo]@xmath83 . 0.5 true cm except that the signal corresponds to a @xmath89 binary.,title="fig:",width=480 ] we present the results of our study in the test mass limit in four parts . in the first part
we discuss our conclusions on the mathematical problem of the closeness of the standard adiabatic and complete adiabatic template families with the family of exact waveforms in the adiabatic approximation . in the next part we exhibit our results in the case of the initial ligo detector . in the third part
we compare the relative importance of improving the accuracy of the energy and flux functions . finally , in the fourth part we compare the total number of gw cycles and the number of useful cycles accumulated by various standard adiabatic and complete adiabatic approximants .
.effectualness of _ standard adiabatic _
@xmath35},{\cal f}_{n})$ ] and _ complete adiabatic _ @xmath39},{\cal f}_{n})$ ] templates in the test mass limit .
overlaps are calculated assuming a flat spectrum for the detector noise ( white - noise ) . [ cols="^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^,^ " , ]
the _ standard adiabatic _ approximation to the phasing of gravitational waves from inspiralling compact binaries is based on the post - newtonian expansions of the binding energy and gravitational wave flux both truncated at the _ same relative _ post - newtonian order . to go beyond the adiabatic approximation one must view the problem as the dynamics of a binary under conservative relativistic forces and gravitation radiation damping . in this viewpoint
the standard approximation at leading order is equivalent to considering the 0pn and 2.5pn terms in the acceleration and neglecting the intermediate 1pn and 2pn terms . a complete treatment of the acceleration at
leading order should include _ all _ pn terms up to 2.5pn .
these define the _ standard _ and _ complete _ _ non - adiabatic _ approximants respectively . a new post - newtonian _ complete adiabatic _ approximant based on energy and flux functions is proposed . at the leading order it uses the 2pn energy function rather than the 0pn one in the standard approximation so that heuristically
, it does not miss any intermediate post - newtonian terms in the acceleration .
we have evaluated the performance of the standard adiabatic vis - a - vis complete adiabatic approximants , in terms of their _ effectualness _
( i.e. larger overlaps with the exact signal ) and _ faithfulness _
( i.e. smaller bias in estimation of parameters ) .
we restricted our study only to the inspiral part of the signal neglecting the plunge and quasi - normal mode ringing phases of the binary @xcite .
we have studied the problem both for the white - noise spectrum and initial ligo noise spectrum .
the main result of this study is that the conservative corrections to the dynamics of a binary that are usually neglected in the standard treatment of the phasing formula are rather important at low pn orders . at the low pn orders , they lead to significant improvement in the overlaps between the approximate template and the exact waveform . in both the white - noise and initial ligo cases we found that at low ( @xmath90 3pn ) pn orders the effectualness of the approximants significantly improves in the complete adiabatic approximation .
however , standard adiabatic approximants of order @xmath1 3pn are nearly as good as the complete adiabatic approximants for the construction of effectual templates .
in the white - noise case , the _ faithfulness _ of both the approximants fluctuates as we go from one pn order to the next and is generally much smaller than our target value of 0.965 .
the fluctuation continues all the way up to 5pn order probably reflecting the oscillatory approach of the flux function to the exact flux function with increasing pn order .
poor faithfulness also means that the parameters extracted using these approximants will be biased .
it is again interesting to note that complete adiabatic approximants are generally more faithful than the standard adiabatic approximants .
for the initial ligo noise case on the other hand , the faithfulness of the _ complete adiabatic _ approximants is vastly better at all orders . to the extent possible , we also tried to investigate this problem in the case of comparable mass binaries by studying the overlaps of all the approximants with a fiducial ` exact ' waveform using initial ligo noise spectrum .
it is shown that , provided we have a t - approximant of the flux function of order @xmath1 1.5pn , the _ standard adiabatic _ approximation provides a good lower bound to the _ complete adiabatic _
approximation for the construction of both _
effectual _ and _ faithful _ templates .
this result is in contrast with the test mass case where we found that the _ complete adiabatic _ approximation brings about significant improvement in _ effectualness _ up to 2.5pn order and significant improvement in _ faithfulness _ at all orders . to achieve the target sensitivity of 0.965 in effectualness , standard adiabatic approximants of order 2pn and 3pn
are required for the @xmath56 and @xmath57 binaries , respectively . whether the complete adiabatic approximant achieves this at an earlier pn order is an interesting question .
it is worth stressing that this result is relevant only for the family of inspiral waveforms . in the real physical case of bh - bh binaries
the inspiral family would not be adequate and must be supplemented by the plunge part of waveform as first discussed in @xcite and later in @xcite .
a discussion of plunge requires a 3pn description of dynamics so that the 2pn templates are no longer adequate .
this is an example of the second variety of questions one can study in this area referred to in our introduction related to whether a template family indeed represents the gws from a specific astrophysical system .
we have also constructed both standard and complete non - adiabatic approximants using the lagrangian models in ref .
however , we were limited in this investigation because of two reasons : firstly , the lagrangian models are available only up to 3.5pn order , and higher order pn accelerations are as yet unavailable which makes it impossible to calculate the _ complete non - adiabatic _ approximants of order @xmath91 1pn . secondly , the only _ exact _ waveform we have has , however been constructed only in the _ adiabatic _ approximation .
so we are unable to make strong statements of general trends and view this effort only as a first step towards a more thorough investigation . from the non - adiabatic models studied ,
the conclusion one can draw is that while complete non - adiabatic approximation improves the effectualness , it results in a decrease in faithfulness .
there is a limitation to our approach which we should point out : complete adiabatic models can be very well tested in the test mass where both approximate and exact expressions are available for the various quantities .
however , complete models can not be worked out to high orders in the comparable mass case since they need the energy function to be computed to 2.5pn order greater than the flux and currently the energy function is only known to 3pn accuracy .
also , due to the lack of an exact waveform , one is constrained to depend upon some fiducial exact waveform constructed from the approximants themselves .
though , in the present paper we have used the new approximants to construct waveform templates , one can envisage applications to discuss the dynamics of the binary using numerical integration of the equations of motion . during the course of this study , we also attempted to assess the relative importance of improving the accuracy of the energy function and the flux function by systematically studying the approach of the adiabatic pn templates constructed with different orders of the energy and the flux functions to the exact waveforms . from the study of test - mass templates we also conclude that , provided the comparable mass case is qualitatively similar to the test mass case , neither the improvement of the accuracy of energy function from 3pn to 4pn nor the improvement of the accuracy of flux function from 3.5pn to 4pn will result in a significant improvement in effectualness in the comparable mass case
. we are grateful to luc blanchet and alessandra buonanno for discussions and comments on the manuscript .
this research was supported partly by grants from the leverhulme trust and the particle physics and astronomy research council , uk .
pa thanks raman research institute and albert einstein institute for hospitality and support , and cardiff university for hospitality during various stages of this work .
pa also thanks k. g. arun for useful discussions .
bri thanks cardiff university , iap paris and ihes france for hospitality during the final stages of the writing of the paper .
cakr thanks ian taylor and roger philip for useful discussions and pparc for support while bss thanks the raman research institute for supporting his visit in july - august 2004 during which some of this research was carried out .
k. g. arun , b. r. iyer , b. s. sathyaprakash and p. a. sundararajan , parameter estimation of inspiralling compact binaries using 3.5 post - newtonian gravitational wave phasing : the non - spinning case , ( 2004 ) ; gr - qc/0411146 .
l. blanchet , t. damour and b. r. iyer , surface - integral expressions for the multipole moments of post - newtonian sources and the boosted schwarzschild solution , class .
( submitted ) ( 2004 ) ; gr - qc/0410021 . | the _ standard adiabatic _ approximation to phasing of gravitational waves from inspiralling compact binaries uses the post - newtonian expansions of the binding energy and gravitational wave flux both truncated at the _ same relative _ post - newtonian order .
motivated by the eventual need to go beyond the adiabatic approximation we must view the problem as the dynamics of the binary under conservative post - newtonian forces and gravitational radiation damping . from the viewpoint of the dynamics of the binary , the standard approximation at leading order is equivalent to retaining the 0pn and 2.5pn terms in the acceleration and neglecting the intervening 1pn and 2pn terms .
a complete mathematically consistent treatment of the acceleration at leading order should include all pn terms up to 2.5pn _ without any gaps_. these define the _ standard _ and _ complete _ _ non - adiabatic _ approximants respectively .
we propose a new and simple _ complete adiabatic _
approximant constructed from the energy and flux functions . at the leading order it uses the 2pn energy function rather than the 0pn one in the standard approximation so that in spirit it corresponds to the dynamics where there are no missing post - newtonian terms in the acceleration .
we compare the overlaps of the standard and complete adiabatic approximant templates with the exact waveform ( in the adiabatic approximation ) for a test - particle orbiting a schwarzschild black hole .
overlaps are computed using both the white - noise spectrum and the initial ligo noise spectrum .
the complete adiabatic approximants lead to a remarkable improvement in the _ effectualness _
( i.e. larger overlaps with the exact signal ) at lower pn ( @xmath0 3pn ) orders
. however , standard adiabatic approximants of order @xmath1 3pn are nearly as good as the complete adiabatic approximants for the construction of effectual templates . in general , _ faithfulness _
( i.e. smaller biases in the estimation of parameters ) of complete approximants is also better than that of standard approximants .
standard and complete approximants beyond the adiabatic approximation are next studied using the lagrangian models of buonanno , chen and vallisneri in the test mass limit . a limited extension of the results to the case of comparable mass binaries
is provided . in this case ,
standard adiabatic approximants achieve an effectualness of 0.965 at order 3pn .
if the comparable mass case is qualitatively similar to the test mass case then neither the improvement of the accuracy of energy function from 3pn to 4pn nor the improvement of the accuracy of flux function from 3.5pn to 4pn will result in a significant improvement in effectualness in the comparable mass case for terrestrial laser interferometric gravitational wave detectors . |
the global infrared ( ir ) and radio emission are tightly and virtually linearly correlated in a broad variety of star forming ( sf ) systems ( see * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* and references therein ) , thus defining what is known as the ` _ ir - radio relation _ ' .
studies in the nearby universe have shown that not only late - type galaxies @xcite ranging from normal spirals to the most vigorously star forming ultra - luminous infrared galaxies ( ulirgs ; e.g. * ? ? ?
* ; * ? ? ?
* ) follow the relation , but also many early - type galaxies with low - level star formation @xcite , as well as interacting systems of mixed morphological composition @xcite .
+ following first indications of the correlation in ground - based observations at 10@xmath1 m and 1.4ghz @xcite the ubiquity and tightness of the ir - radio relation became fully appreciated during the analysis of the combination of data from the very large array ( vla ) and the _ infrared astronomical satellite _ ( _ iras _ ) which measured the far - ir ( fir ) properties of @xmath320,000 galaxies at @xmath10 0.15 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
ir observations of sources at higher redshifts became available with the advent of the _ infrared space observatory _
( _ iso _ ) and , in recent years , with the _
spitzer space telecope_. they have provided increasing evidence that the locally observed correlation likely holds until @xmath7 1 @xcite and that the linearity of the correlation is maintained as far back as @xmath7 3 although the slope may change , especially for sub - mm galaxies ( @xcite , but see also @xcite ) . the statistical significance of these high-@xmath11 studies , however , is still low because so far the number of sources detected at @xmath12 0.5 is limited . the very tightness of the ir - radio relation ( the intrinsic dispersion in the local galaxy population is less than a factor 1.5 ) ,
combined with the fact that it spans five orders of magnitude in bolometric luminosity has provided a useful tool for numerous astrophysical applications and motivated continued study of the relation .
@xcite , for example , use the measured ir / radio flux ratio to determine the evolutionary stage of systems undergoing a starburst , depending on whether they show excess ir or radio emission with respect to the average locus of star forming galaxies ( sfgs ) .
the consideration of radio - excess outliers to the ir - radio relation also has been used to select galaxies in which an active galactic nucleus ( agn ) rather than star formation is the dominant source of ir and radio emission ( e.g. * ? ? ? * ; * ? ? ?
* ) . at high redshift
the ir - radio relation has been used to compute distance estimates of sub - mm galaxies lacking optical counterparts @xcite , as well as to estimate the contribution of sfgs to the cosmic radio background @xcite .
+ in the absence of agn activity both ( f)ir and radio flux measurements are in principle unbiased tracers of star formation since the radiation in these spectral regions is not attenuated by dust as opposed to the often heavily obscured emission at ultraviolet ( uv ) or optical to near - ir ( nir ) wavelengths .
the observation that a correlation between the ir and radio exists implies that the thermal ir dust emission and the radio continuum flux which is a frequency - dependent mixture of thermal emission and a non - thermal synchrotron component have a common origin . soon after the discovery of the ir - radio relation the birth and demise of massive ( @xmath135@xmath14 ) stars was identified as its likely cause ( see * ? ? ?
* ; * ? ? ?
* and references therein ) .
deviations from exact proportionality of the ir and radio emission as well as changes in the ir - radio relation with cosmic time thus potentially imply changes in the mechanisms steering star formation , or that ir and radio emission are not equally good proxies of star formation under all circumstances . however , several decades after the discovery of the relation many details concerning the physical processes which shape it still need to be settled .
+ a number of models to explain the global correlation between the _ integrated _ ir and radio fluxes of sfgs have been advanced : they include the calorimeter model @xcite , the ` optically thin ' scenario @xcite and the ` linearity by conspiracy ' picture of @xcite .
models focusing on the magnetohydrodynamic state of the interstellar medium ( ism ) are often constructed in order to explain the correlation on both kpc scales _ within _ galaxies as well as globally ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
many of these fare well in reproducing multiple aspects of the correlation but either can not satisfy all observational constraints or make predictions which still await confirmation .
+ on the observational side further insight into the ( astro-)physical processes shaping the ir - radio relation has been gained by studying the ir - radio relation on small scales in resolved nearby galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* dumas et al . , in prep . ) which revealed that ir / radio properties vary inside spiral galaxies or by investigating the impact of environmental effects @xcite .
another important finding was that the linearity of the relation is maintained with both the thermal and non - thermal radio emission @xcite taken separately , as well as for the warm and cold dust components @xcite .
as a quantitative measure of the correlation @xcite introduced the logarithmic flux density ratio @xmath15 here @xmath16 is the rest frame fir flux traditionally computed as a linear combination of measurements in the iras 60 and 100@xmath1 m bands under the assumption of a typical dust temperature of @xmath330k and 3.75@xmath1710@xmath18hz is the ` central ' frequency of the fir - window ( 42.5 - 122.5@xmath1 m ) .
@xmath19 is the monochromatic rest frame 1.4ghz flux density .
+ strong improvements of the sensitivity of ir observatories over the last decade , in particular in the mid - ir ( mir ) with the _ mips _ photometer on _ spitzer _ , have promoted the use of monochromatic flux ratios , e.g. at 24 or 70@xmath1 m ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) : @xmath20 } = { \rm log}\left(\frac{s_{\nu}(24\,[70]~{\rm { \mu}m})}{s_{\nu}(1.4\,{\rm ghz})}\right)~ , \label{eq : monchromqs}\ ] ] where both flux densities are specified in units of @xmath21 .
this approach is convenient for evolutionary studies of the ir - radio relation at intermediate and high redshift as it avoids the computation of ir luminosities from a single ir flux measurement , usually made at 24@xmath1 m due to the high sensitivity of the according _ mips _ filter .
however , it still requires the computation of @xmath22-corrections , if the flux densities in equation ( [ eq : monchromqs ] ) are to be given in the rest frame .
+ since current radio surveys at centimeter wavelengths are generally significantly shallower than the mir photometry several authors have carried out radio stacking experiments @xcite with the aim of determining the ir / radio properties of the faint ir population .
this is of particular interest in view of the potential for detecting changes in the ir - radio relation brought about by relativistic cooling of cosmic ray electrons by inverse compton scattering off photons from the cosmic microwave background ( cmb ) .
this cooling may overwhelm at least at low decimeter frequencies the synchrotron losses in the ism of normal galaxies at @xmath12 0.5 @xcite if the energy density of their magnetic fields is similar in strength to that typically measured in spiral arms ( @xmath23erg@xmath24 ; * ? ? ?
. alas , the stacking results have produced strongly discrepant results , both among different cosmological survey fields and with respect to recent studies of sources directly detected in both the ir and radio .
an additional concern of particular relevance in evolutionary studies is the bias that is introduced by constructing samples from different selection bands in flux limited surveys . in this paper
we focus on the impact of selection biases and apply the statistical technique of survival analysis to our data which permits the inclusion of constraints from flux limits in the study .
this approach is a significant improvement over limiting the sample to those sources which are detected at both ir and radio wavelengths .
these topics are discussed in [ sect : selsurv ] .
our data sets contain an unprecedented amount and quality of information gathered as part of the cosmos survey @xcite .
this is reflected by a sample which contains significantly more sources in the redshift range @xmath12 0.5 , in which the sources in previous studies began to taper out .
moreover , a large fraction ( 33% ) of spectroscopically determined redshifts , which are supplemented by accurate photometric redshift estimates , in combination with flux constraints at both 24 and 70@xmath1 m for each source in the sample make for improved estimates of ir luminosities ( ( * ? ? ?
* e.g. ) ; kartaltepe et al .
, subm . ) .
they are an important prerequisite for placing accurate constraints on the evolution of the ir - radio at high redshift .
we introduce our data and the subsequently analyzed samples in [ sect : data ] .
for those interested , a detailed review of the different catalogs and images used in the paper and the band - merging of this information into a rich multi - wavelength data is provided in appendices [ appsect : data ] and [ appsect : bandmerge ] . [ sect : sfvsagn ] and [ sect : sedfitting ] deal with methodological considerations applying to the identification of star forming sources and agn galaxies as well as to the derivation of ir luminosities . following the section on biasing and our treatment of flux limits in
[ sect : selsurv ] , the bulk of our analysis is then presented in [
sect : results ] .
our most important results are : + ( a ) the constancy of the ir - radio relation as parametrized by the flux ratios @xmath4 ,
@xmath5 and @xmath6 out to @xmath7 1.4 , as well as for a sub - sample of high-@xmath11 sources at 2.5 @xmath25 5 , + ( b ) the identification of selection biases as a potential explanation for discrepant average ir / radio flux ratios measured in previous studies , and + ( c ) the observation that over the last 10 billion years the distribution of ir / radio ratios of optically selected , radio - quiet agn has been very similar to that of star forming galaxies + we discuss and summarize these findings in [ sect : discussion ] and [ sect : conclusions ] , respectively . throughout this article
we adopt the wmap-5 cosmology defined by @xmath26 0.258 , @xmath27 1 and a present - day hubble parameter of 71.9 kms@xmath28mpc@xmath28 @xcite .
magnitudes are given in the ab - system of @xcite unless the opposite is explicitly stated and we henceforth drop the subscript ` ab ' in the text .
in order to study biases arising from the selection of sources at either ir or radio wavelengths we have constructed both a radio- and an ir - selected sample of cosmos galaxies . all observations and associated data sets , as well as the band - merging procedures used to identify counterparts from radio to x - ray wavelengths
are described in detail in appendices [ appsect : radiodata]-[appsect : redshifts ] and appendix [ appsect : bandmerge ] , respectively .
here we only briefly present the primary data sets ( [ sect : irradiodata ] ) and review the most important properties of our _ radio - selected _ and _ ir - selected _ samples ( [ sect : sampreview ] ) .
this section also contains a summary of the ancillary multi - wavelength photometry and the redshift information which is available for our sources ( [ sect : ancilldata ] ) . the vla - cosmos project @xcite has imaged the cosmos field at 1.4ghz with the vla to a mean sensitivity of @xmath290.01 ( @xmath30.04)mjy / beam at the centre ( edge ) of the field .
as the basis of our subsequent analysis we use the vla - cosmos ` joint ' catalog ( schinnerer et al .
2009 , subm . )
which contains @xmath32900 sources detected with @xmath30 5 .
nearly 50% of these are resolved at a resolution ( fwhm of synthesized beam ) of [email protected]@xmath32 .
flux measurements were carried out with aips ( astronomical image processing system ; * ? ? ? * ) and have been corrected for bandwidth smearing in the case of the unresolved radio sources .
their errors are generally @xmath317% of the flux value .
ir data at 24 and 70@xmath1 m were taken by the s - cosmos survey @xcite using _ mips _ on _ spitzer_. + at 24@xmath1 m the fwhm of the psf is 5.8@xmath32 and the average @xmath33 sensitivity @xmath30.018mjy over a large fraction of the imaged area .
flux measurements @xcite in the deep and crowded _ mips _
24@xmath1 m image were performed with the psf - fitting algorithm daophot @xcite which can simultaneously fit and hence de - blend multiple sources . in the following we use a flux limited catalog of cosmos 24@xmath1 m sources which
is restricted to the range @xmath34 0.06mjy .
typical flux uncertainties at 24@xmath1 m are @xmath38% .
+ mips 70@xmath1 m observations of the cosmos field were carried out in parallel with the 24@xmath1 m imaging .
they have an average 1@xmath35 point source noise of 1.7 mjy and a resolution of 18.6@xmath32 .
our 70@xmath1 m source list includes detections down to @xmath36 3 and was compiled as described in @xcite .
fluxes were derived using the apex ( astronomical point source extraction * ? ? ?
* ) peak fitting algorithm ; their average uncertainty is @xmath316% . in fig .
[ fig : fluxlims ] we show the minimum total ir ( tir ; 8 - 1000@xmath1 m ) luminosity that is detectable as a function of redshift given the sensitivity of the 1.4ghz ( converted to an ir measurement assuming the local ir - radio relation ) and 24/70@xmath1 m data .
different ir sed template libraries ( see colour scheme in the lower right corner ) lead to somewhat different predictions but it is clear that the 1.4ghz and 70@xmath1 m surveys have matching depths while the 24@xmath1 m observations are about seven times deeper . a similar sampling of the ir luminosity function is achieved in all three bands if a 24@xmath1 m flux limit of approx . 0.3mjy is assumed . we therefore limit our 24@xmath1 m catalog to the range @xmath34 0.3mjy when we construct our ir - selected sample but allow fainter counterparts of 1.4ghz sources to be included in the radio - selected sample .
the use of these flux limited samples has the immediate consequence that we only detect the brightest ulirgs at @xmath12 1.5 while the average luminosity of our sources is much lower at , e.g. @xmath7 0.5 , where most sources belong to the lirg class .
optical data and photometric redshifts are taken from the cosmos photometry catalog of @xcite which lists more than 600,000 cosmos galaxies with @xmath37 26 detected in a region roughly contiguous with the area covered by the vla - cosmos survey .
the wavelength range covered by these observations ( 30 broad , medium and narrow band filters ) extends all the way from the uv at 1550 to the mir at 8 @xmath1 m . @xcite
provide a complete description of these observations . + spectroscopic data has been gathered for more than 20,000 sources in the cosmos field , e.g. by the zcosmos survey @xcite and sdss @xcite , or in magellan / imacs and keck / deimos follow - up observations dedicated to specific ( classes of ) sources ( ( e.g .
* ) ; @xcite ; kartaltepe et al . , in prep . ;
salvato et al . , in prep ) .
if a reliable spectroscopic redshift is available it is favoured over the photometric redshift estimate .
the choice of the best possible distance measurement for our radio and ir sources is described in detail in [ appsect : redshifts ] .
the _ xmm - newton _ cosmos survey @xcite has detected a total of 1887 bright ( @xmath382@xmath39 erg@xmath40s@xmath28 in the 0.5 - 10kev band ) x - ray sources over 90% ( 1.92 deg@xmath2 ) of the cosmos field .
a large fraction of these are associated with agn and hence provide a means of identifying agn - powered radio and ir sources in our sample which is complementary to our primary classification scheme introduced in [ sect : sfvsagn ] . for our subsequent analysis
we rely on a list of xmm sources @xcite with unique and secure optical counterparts ( see brusa et al .
, in prep . ) and sed fits to the uv to mir photometry performed by @xcite . due to the differing characteristics ( resolution , astrometric accuracy ) of the radio and ir data the determination of counterparts at other wavelengths differed somewhat for the radio- and the ir - selected sample . which candidate counterparts are incorporated in the final sample and which are rejected is determined by the goals of this study : in our case it is more important to select objects with the cleanest possible radio and ir flux measurements rather than having a statistically complete sample .
the details of the band - merging between the ir and the radio catalogs and the subsequent exclusion of ambiguous counterparts are presented in detail in appendix [ appsect : bandmerge ] . here
we summarize the most important properties of the radio- and ir - selected samples .
based on the positions of @xmath415@xmath35 1.4ghz detections in the vla - cosmos deep project image we searched for ir counterparts in the s - cosmos 24 and 70@xmath1 m catalogs which have @xmath42 3 .
counterparts were determined by direct positional matching of radio and ir coordinates with search radii corresponding to approx .
fwhm/3 of the ir psfs of the respective _ mips _ bands .
if no counterpart was found , a 3@xmath35 point - source detection limit was determined based on the corresponding uncertainty images .
radio sources with ambiguous ir counterparts
i.e. in the presence of more than one potential counterpart or if the counterpart had not been uniquely assigned to a single radio source have been excluded from the analysis of the paper .
the match with the cosmos multi - wavelength and spectroscopy catalogs provides distance estimates for 73% of the radio - selected sample as well as photometry from the uv to the mir which is used to separate galaxies dominated by star formation or agn emission ( see [ sect : sfvsagn ] ) . in the upper panel of fig .
[ fig : ctp_dists ] we show histograms of the separation between radio source positions and the location of the optical and ir counterparts .
note that the distance is normalized by the width of the broader psf of the two involved bands .
[ fig : radioctp_flxchar ] shows the 24 and 70@xmath1 m flux distribution of the radio sources , including information on whether the flux constraint is a well - defined measurement or an upper flux limit .
[ fig : flxchar_vs_z ] ( left - hand panel ) explicitly shows how the fraction of sources that have a directly detected counterpart in either or both of the _ mips _ bands or only upper flux limits changes as a function of redshift ( 38% of the redshifts are spectroscopically , 64% photometrically determined ) .
the ir - selected sample is based on sources listed in the s - cosmos 24@xmath1 m catalog that satisfy the criterion @xmath34 0.3mjy .
this criterion ensures that the ir - selected sample is well matched to the 70@xmath1 m and 1.4ghz data as far as the sampling of the ir luminosity function is concerned . to reduce the likelihood of false identifications due to the positional uncertainty of the 24@xmath1
m sources we searched for irac counterparts , the positions of which were used as a prior in the subsequent band - merging with the other wavelengths .
if no irac counterpart was available we also admitted unambiguous matches with optical sources .
70@xmath1 m counterparts to the 24@xmath1 m sources with @xmath30 3 were determined and validated following exactly the same approach as in the radio - selected sample . for those 24@xmath1 m sources which did not already have a known radio counterpart ( determined in the construction of the radio - selected sample ) we checked whether they are associated with a counterpart having @xmath43 3 .
all new detections satisfying this criterion were then added to the list of radio counterparts with @xmath30 5 that were already known from the construction of the radio - selected sample .
24@xmath1 m sources that are undetected at 70@xmath1 m and/or 1.4ghz are assigned 3@xmath35 upper flux bounds .
+ the distributions of separations between all 24@xmath1 m sources and their counterparts in the optical , far - ir and 1.4ghz maps are given in fig .
[ fig : ctp_dists].b .
the contribution of flux limits and well - defined flux measurements as a function of flux and @xmath44 at 70@xmath1 m and 1.4ghz is illustrated in fig .
[ fig : irctp_flxchar ] . finally , in the right - hand panel of fig .
[ fig : flxchar_vs_z ] we show at different redshifts which fraction of the ir - selected sample has direct detections or upper flux density limits at 70@xmath1 m and/or 1.4ghz .
spectroscopic or photometric redshift measurements are available for 80% of the objects in the ir - selected sample .
the remaining sources are either not bright enough for spectroscopy or have flux information in too few bands to derive a photometric redshift based on sed fitting .
the jointly selected sample is the union of the radio- and ir - selected samples presented in [ sect : radiosamp ] and [ sect : irsamp ] . as such it contains 6863 sources : 1560 sources that are only detected at 1.4ghz , 3960 sources that are only detected at 24@xmath1 m and , finally , 1341 sources which are selected at both wavelengths . in table
[ tab : zstats ] we summarize the available redshift information for the jointly radio- and ir - selected sample , as well as separately for the radio- and ir - selected samples . in fig .
[ fig : flxvsflx ] the ir and radio fluxes of our sources are compared .
the colour coding of the data points distinguishes three kinds of sources : in black those which have entered both the 1.4ghz catalog as well as the 24@xmath1 m catalog ( restricted to sources with flux density larger than 0.3mjy ) , in green 1.4ghz sources without counterpart in the 24@xmath1 m catalog and in orange those 24@xmath1m - detected sources which do not have a counterpart in the vla - cosmos joint catalog .
the sources from these three different categories have been added to the plot in random order to prevent that the symbols of the initially plotted category are systematically hidden by the successively overplotted data in common regions of flux space .
[ fig : flxvsflx].c confronts the fluxes in the two selection bands ; the empty rectangle in the lower left corner of this panel reflects the selection criteria at 1.4ghz and 24@xmath1 m . since the 24@xmath1 m catalog is flux limited , essentially all upper 24@xmath1 m flux limits lie at or below the critical flux threshold ; upper 1.4ghz flux limits for undetected 24@xmath1 m sources on the other hand are also encountered at higher 1.4ghz flux values than the sharp cut - off at @xmath45 0.05mjy because the radio - catalog was constructed using a @xmath44 criterion .
note that the region where both the 1.4ghz and 24@xmath1 m flux density clearly exceed the respective selection thresholds contains some sources which are not included in both the catalog of 24@xmath1 m and that of 1.4ghz detections ( cf . orange and green symbols in the area where @xmath46 0.1mjy and @xmath47 0.3mjy ) .
two reasons can be responsible for this : _ ( a ) _ minor incompleteness of the catalogs , or _
( b ) _ spatial variations in the background noise which , at a given flux , lead to certain sources not being detected at the significance level required for inclusion in the original source list .
both star formation and agn activity cause the host galaxy to ( re-)emit at ( mid-)ir and radio wavelengths . to study the ir / radio properties of these two distinct populations separately , information from different regions of the electromagnetic spectrum is thus required .
@xcite devised a method which , in a statistical sense , is capable of selecting star forming and agn galaxies with a simple cut in rest frame optical colour .
it relies on the tight correlation @xcite between the rest frame colours of emission line galaxies and their position in the bpt diagram @xcite and was developed and calibrated with radio sources at @xmath10 1.3 using the principal component colour , @xmath48 , @xmath49 , @xmath50 ; @xcite ) in the wavelength range 3500 - 5800 ; see @xcite for the definitions and additional details .
] henceforth referred to as ` p1 ' .
it can , however , be easily adapted to other rest frame colours because galaxy seds from the near - uv to the nir represent a one - parameter family @xcite .
here we use the combination of the filters @xmath51 and @xmath22 to select agn and sfgs .
this choice is motivated by the desire to apply the classification to both the radio- and ir - selected sample ; the likely presence of dust - obscured star forming systems in the ir - selected sample requires the inclusion of a red band , to prevent , as best possible , dust - reddened star forming sources from being mistaken for red , early - type agn host galaxies .
[ fig : p1equivcol ] we show the correlation of p1 ( computed according to * ? ? ?
* ) and ( @xmath52 ) for @xmath3950 vla - cosmos sources , for which both p1 and ( @xmath52 ) were available .
rest frame ( @xmath52 ) colours were computed with zebra ( _ zurich extragalactic bayesian redshift analyzer _ ; * ? ? ?
* ) which was used to find the best - fitting sed template to the cosmos photometry in the medium and broad band filters @xmath53 , @xmath54 , @xmath55 , @xmath56 , @xmath57 , @xmath58 , @xmath59 , @xmath60 and @xmath61 , as well as in the four irac channels given the known redshift ( cf .
[ appsect : redshifts ] ) .
note that the magnitudes @xmath51 and @xmath22 used here are computed in johnson - kron - cousins filters rather than the cosmos filters .
an ordinary least squares ( ols ) bisector fit @xcite accounts for the fact that both colours are subject to uncertainty and returned a best - fit correlation given by @xmath62 which is indicated in grey in fig . [ fig : p1equivcol ] .
the criterion @xmath63 0.15 of @xcite for the separation of sf ( @xmath63 0.15 ) and agn sources ( @xmath64 0.15 ) thus corresponds to ( @xmath52 ) = 2.42 .
note that due to our treatment of composite sf / agn sources we adopt a slightly different colour threshold for the selection of sfgs ( see following paragraph and fig .
[ fig : uk_sfprob ] ) . from fig .
24 of @xcite ( reproduced in the upper left corner of fig .
[ fig : uk_sfprob ] ) it is obvious that the tails of the distribution of star forming and agn systems in p1 colour space extend into the selection regions for agn and star forming sources , respectively .
moreover , about 10% of the sample on which the classification scheme was developed are ` composite ' systems and found on either side of the divide .
when a source is classified as sf or agn based purely on its rest frame optical colour there thus is a non - negligible probability of assigning it to the false category .
for some purposes , e.g. when estimating which fraction of agn systems have similar ir / radio properties as star formers , it is thus useful to adopt a probabilistic approach . given the distributions @xmath65 , @xmath66 and @xmath67 ( see fig .
[ fig : uk_sfprob].a ) a possible definition for an _ effective _ probability @xmath68 of correctly classifying a source as star forming at a given rest frame optical colour is @xmath69 where @xmath70 in setting up ( [ eq : sfprob ] ) we have assigned composite systems to the sf and agn population according to the relative abundance @xmath71 of sf and agn sources at the particular colour .
@xmath68 is _ priori _ given as a probability as a function of p1 through the distributions @xmath65 , @xmath66 and @xmath67 presented in @xcite .
however , it may be directly converted to the desired dependency on ( @xmath52 ) by convolving the expression in ( [ eq : sfprob ] ) with the distribution of p1 at fixed ( @xmath52 ) colour ( see inset of fig .
[ fig : p1equivcol ] ) , which reflects the range of probabilities @xmath72 that contribute to @xmath73 . in the upper right panel of fig .
[ fig : uk_sfprob ] we show the distributions @xmath74 obtained according to equation ( [ eq : sfprob ] ) and smoothed with a three - point running average ( black curve
star forming sources ; dark grey curve agn systems ) .
its convolution with a standard normal curve leads to the probability distribution @xmath75 shown in the lower panel of fig .
[ fig : uk_sfprob ] which uses the same colour scheme as in panel _
( b)_. the uncertainty in the best - fit correlation between p1 and ( @xmath52 ) has been translated into an error in the probability function which is shown as a light grey area to either side of the black line giving @xmath73 in panel _ ( c)_. due to the small uncertainties in the ols bisector line parameters of equation ( [ eq : p1umk_corr ] ) the dispersion @xmath76 is the most important factor that determines the differences in the shape of @xmath74 and @xmath75 .
+ if one assigns composite objects to the sf and agn population according to equation ( [ eq : sfprob ] ) the point of equal probability of correctly classifying objects as sf or agn , respectively , is reached at ( @xmath52 ) = 2.36 .
this value is only slightly different from the direct translation ( see previous paragraph ) of the original definition in @xcite . in the remainder of the paper
we will use the ( @xmath52 ) = 2.36 threshold to separate sfgs from sources with emission that is dominated by agn activity . ] .
apart from the tails in the colour distribution of agn and sf systems which cross the colour threshold , three additional effects could reduce the accuracy of the classification scheme .
+ first of all , a general evolution of the sf and agn population to bluer colours at high redshift would lead to increasing contamination by agn of the high - z population of sfgs if the colour cut is not adapted .
@xcite have shown that an unchanging threshold is adequate until at least @xmath7 1.3 . in fig .
[ fig : probcol_distr ] we plot the distribution of ( @xmath52 ) colours of our sources and follow the evolution of the median colour for both the ir- ( dark grey histogram ) and the radio - selected sample ( light grey histogram ; sources common to both samples lie within the hatched area ) .
we find no evidence for a strong evolution of average colours in either of the two samples out to @xmath7 2 , and out to @xmath7 3 only by a small amount .
hence we apply the selection criterion uniformly to all sources , regardless of their redshift , except for the objects at the highest redshift where the medians have begun to change appreciably ( see lower right panel of fig .
[ fig : probcol_distr ] ) .
+ secondly , nonperiodic flux variations of active galaxies will affect the choice of the best - fitting sed if photometric measurements are not simultaneously carried out over the whole spectrum .
since the rest frame optical colours are determined using sed templates this can cause misclassifications of agn or sfgs with colours close to the threshold ( @xmath52 ) = 2.36 .
a variability analysis ( m. salvato , private communication ) of our 1.4ghz sources revealed that maximally 20% of these display strong variability ( defined as @xmath77 0.25 ; cf .
equation ( 1 ) in * ? ? ?
the true fraction of affected sources is likely to be smaller because inaccuracies in the photometry can artificially raise the variability parameter .
+ finally , we can not exclude that some unobscured type 1 agn with a blue ( @xmath52 ) colour will be assigned to the sf category in our classification scheme .
it is also possible that a number of dust - reddened starburst galaxies end up being classified as agn , even though we used a red filter to define our rest frame colour on which we base the separation into sfgs and agn . in the calibration sample of @xcite
this kind of contamination amounted to less than @xmath310% ( see their appendix b2 ) .
data from lensed high-@xmath11 galaxies @xcite and from recent deep fir surveys have shown that the seds of local star forming galaxies reproduce the seds of high redshift galaxies well out to @xmath10 1.5 ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
however , it has also been reported that the seds of some ir - selected galaxies at high redshift can differ from local templates both at mir @xcite and fir @xcite wavelengths , conceivably due to intrinsic scatter in the physical properties of these sources which deviate from the median trend that the empirical galaxy templates represent .
+ following the procedure described in @xcite , we derive infrared luminosities ( @xmath78 ) by fitting the 24 and 70@xmath1 m data points to the @xcite sed templates and integrating between 8 and 1000 @xmath1 m .
this wavelength range is in principle also sampled by s - cosmos observations at 8 and 160@xmath1 m @xcite but we restrict ourselves to the two aforementioned bands because ( _ i _ ) at @xmath79 0.6 a 8@xmath1 m measurement would include stellar light ( which starts to dominate at the sed at rest frame 5@xmath1 m ) , while we are fitting pure dust templates ; and ( _ ii _ ) the shallower coverage and broad psf of the 160@xmath1 m observations complicate the identification of unambiguous counterparts .
our choice of the @xcite templates is motivated by the fact that they have been found to exhibit 24/70@xmath1 m flux density ratios that are more representative @xcite of those measured for galaxies at @xmath80 compared to the @xcite or @xcite templates .
+ for the cases where a source is detected firmly at 24 and 70 @xmath1 m , the best - fit seds are determined by a @xmath81 minimization procedure whereby the sed templates are allowed to scale such that they are being fitted for luminosity and temperature separately .
consequently , the amplitude and shape of the seds scale independently to best match the observations .
the input photometry is weighted by the @xmath44 ratio of the detection if it is a well - defined measurement , and the normalization constant is determined by a weighted sum of observed - to - template flux density ratios for all input data used in the fitting . in the cases where only an upper limit is available at 70 @xmath1 m , the latter
is not incorporated into the @xmath81 minimization but used to reject fits which have flux densities above the associated measured limit .
+ errors on the best - fitting value of @xmath78 are determined by a standard monte carlo approach using the photometric uncertainties of the input flux densities which reflect both calibration errors ( @xmath32% at 24 @xcite and @xmath35% at 70@xmath1 m @xcite ) and the uncertainties in the psf - fitting ( generally of order @xmath82 , where @xmath83 is the flux returned by the psf fit ) .
+ if a source is only detected at 24 @xmath1 m , we also fit the photometry using the sed templates of @xcite and define the best estimate of the ir luminosity as the average @xmath78 from the two separate fits .
the selection effects that are the topic of this section arise in flux limited samples when flux information from one of the selection bands is directly used in the computation of the quantity being studied . in the present case the critical quantity is the logarithmic ir / radio flux ratio @xmath84 , but analogous effects need to be considered in the context of studies of the distribution of spectral indices at different radio frequencies ( e.g. * ? ? ?
* ; * ? ? ?
* ) , of x - ray to optical continuum slopes of agn @xcite or of the @xmath85 and @xmath86 relationships @xcite .
[ fig : qselsketch ] we illustrate the origin of the selection effect : consider the left hand panel in which the ir - to - radio seds of three sources with different observed bolometric flux are distributed along the vertical axis .
each of these three seds splits into three branches at the peak of the sed , thereby schematically reflecting the range of observed ir / radio ratios ( from top to bottom : 3@xmath35 radio - excess outlier
dashed line ; average source solid line ; and 3@xmath35 ir - excess outlier dotted line ) .
if we impose the indicated selection threshold at 1.4 ghz ( red line ) the resulting sample will contain ( _ i _ ) all sources of the brightest flux class , regardless of their ir / radio ratio ; ( _ ii _ ) the source with an average ir / radio ratio and the radio - excess source from objects of the intermediate flux class and ; ( _ iii _ ) in the faintest flux bin only the radio - excess sources .
since the fainter sources are more abundant ( as parametrized by the slope of the differential source counts @xmath87 , with @xmath88 0 ) this results in a surplus of radio - excess sources and consequently a _ low _ average ir / radio ratio in a radio - selected sample . the right hand side of fig .
[ fig : qselsketch ] shows that an ir - selected sample is biased in the opposite direction , i.e. towards _ high _ ir / radio ratios . the analytical expression for the difference between the average ir / radio ratio of ir- and radio - selected samples is @xcite : @xmath89 it thus depends on @xmath90 , the power law index of the source counts , and on @xmath91 , which is the dispersion of the ir / radio relation .
note that this offset will occur _ regardless of the relative depth _ of the two involved bands .
an estimate of the ` intrinsic ' ( i.e. unbiased ) ir / radio ratio can be obtained by constructing the sample using an unrelated selection criterion like optical luminosity , mass or morphological type @xcite .
since the recent work on the evolution of the ir - radio relation at intermediate and high redshift was often based on flux limited surveys , we would expect most of the findings to be affected by this selection bias to a certain extent . in table
[ tab : prevwork ] we have collected the selection criteria and average values of @xmath84 ( final column ) that were published in the literature during the last decade . + we see that broadly speaking the various ir / radio diagnostics have values @xmath92 1 , @xmath93 2.1 , @xmath94 2.3 and @xmath95 2.6 .
these different values are not the result of selection effects but reflect if the ir filter covers a wavelength range that is close to the ir sed peak or a part of the sed with lower energy content . in the following paragraph
we will discuss the plausible influence of selection effects on the various measurements of @xmath4 and @xmath96 , in particular .
+ due to the high sensitivity of the _ mips _
24@xmath1 m band many of the papers listed in table [ tab : prevwork ] have studied the ir / radio ratio @xmath4 .
the radio - selected samples of @xcite and @xcite find that @xmath97 [ 0.94 , 1 ] , depending on the choice of the ir template used for the @xmath22-correction .
the local ir - selected sample of @xcite on the other hand has a mean @xmath4 of @xmath31.25 and shows some signs of variations with ir luminosity .
the offset between the means of the radio - selected samples and the ir - selected data set is @xmath30.3 dex , in good agreement with the predicted @xmath98 0.31 of equation ( [ eq : qbias ] ) if we set @xmath99 0.3 in accordance with observations ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and under the simplified assumption of euclidean source counts ( @xmath100 ) .
in the case of the fir / radio flux ratio @xmath96 we can compare the two radio - selected ( sub)samples of @xcite and @xcite that have @xmath94 2 with a jointly radio- and sub - mm selected mean of 2.07 from @xcite and mean values @xmath101 [ 2.2 , 2.4 ] for ir - selected @xcite or essentially volume limited samples in @xcite and @xcite .
as with @xmath4 there is thus evidence of a @xmath30.3 dex shift in @xmath6 between radio - selected and other samples . as far as we know no measurement of @xmath5 in an ir - selected sample exists but the compilation in table [ tab : prevwork ] shows that reassuringly all determinations of @xmath5 based on radio - selected samples @xcite are quite similar .
+ the radio stacking experiments of @xcite , @xcite and @xcite do not fit the picture which is probably due to the different nature of the analysis .
nevertheless , it can not be excluded that part of the variations in the other studies are due to field - to - field variance or different assumptions about ir seds and the radio spectral slope . to this end
we will test in [ sect : results ] whether or not the offset between our ir- and radio - selected samples that have been consistently constructed from the same parent data sets conforms to our expectation .
if so , it would be strong support for selection effects alone being able to reconcile the seemingly discrepant measurements of average ir / radio properties in the literature .
discarding the information from undetected counterparts introduces a second source of bias in addition to the selection effects mentioned in [ sect : selbias ] .
it arises from the unrepresentative sampling of the true distribution function of ir / radio ratios by sources which are directly detected in all involved bands .
we would like to emphasize that the shift @xmath102 in equation ( [ eq : qbias ] ) is the difference between the mean @xmath84 of ir- and radio - selected samples with _ correctly sampled _ distribution functions .
@xmath102 can not be compensated by accounting for upper or lower limits on @xmath84 due to undetected ir or radio counterparts in the two different samples ; as discussed in the previous section , the mean ir / radio ratio measured in an ir- and radio - selected sample only brackets the value one would measure with an unbiased data set which we can best approximate by a sample jointly selected at ir and radio wavelengths ( see [ sect : jointsamp ] ) .
+ the ratio of two flux constraints that could be either a well - defined measurement or an upper limit will render an upper or lower bound , a well - defined value or be indeterminate ( if both numerator and denominator are limits ) . since we use the pooled information from a radio- and ir - selected sample in this study the latter case never occurs . we do expect , however , to encounter upper limits on ir / radio ratios from radio - selected sources that are not detected in the ir or lower limits if the radio counterpart of an ir - selected source was too faint to be detected ( cf .
[ appsect : radiodata ] & [ appsect : irdata ] ) .
+ let @xmath103 ( @xmath104 1 , ... , @xmath105 ) be the actual value of the flux ratio for each of the @xmath105 sources in a suitably defined sample ( e.g. the population in a certain slice of redshift ) . as a consequence of the noise characteristics in the radio and ir images @xmath103
can only be measured if it lies in the interval @xmath106 $ ] , where @xmath107 and @xmath108 are upper and lower limits on the flux ratio , respectively .
these limits may be different for each source .
our knowledge about the distribution of ir / radio ratios after carrying out all our measurements can thus be summarized with two vectors of variables , @xmath109 and @xmath110 : @xmath111\\ \quad 1 & \quad \mbox{if } q_i > q^+_i \end{cases } \nonumber \label{eq : doubcensodef}\end{aligned}\ ] ] in _ survival _ or _ life time analysis _ the action of imposing measurement limits is referred to as ` censoring ' .
a variable @xmath112 is said to be _ left censored _ if @xmath113 and _ right censored _ if @xmath114 .
if both kinds of censoring occur in a data set it is called _ doubly censored _ , otherwise one talks of _ single censoring_. during the remainder of the paper we will use the terms ` limit ' and censored measurement interchangeably . in appendix
[ appsect : survival ] we sketch the steps that are involved in going from the information ( @xmath109 , @xmath110 ) to the distribution function of the ir / radio ratios . inferring
the true distribution of the @xmath103 of a sample is essential for the calculation of its average ir / radio properties . in
[ sect : results ] we will construct distribution functions for data sets that are both singly and doubly censored .
recipes for dealing with the former case are plentiful in texts on survival analysis ( see e.g. @xcite for applications to astronomy ) such that we only include some brief remarks in [ appsect : singlecense ] . since the more general case of double censoring is not as widely used in astronomical applications , the most important formulae and useful computational guidelines are provided in appendix [ appsect : doubcensetheory ] .
+ the methods described in appendix [ appsect : survival ] have been implemented using perl / pdl scripts written by m.t.s . .
their correct functionality was verified with examples in the literature . in particular
, we checked that our implementation of the algorithm for the calculation of the doubly censored distribution function @xcite when applied to the special case of singly censored data gave the same results as the scripts based on the kaplan - meier product limit estimator ( @xcite ; see also appendix [ appsect : singlecense ] ) .
the main focus of this section is the search for changes with redshift of the average ir / radio ratio in the sf population .
we track evolutionary trends in the range @xmath9 1.4 for both monochromatic and tir / radio flux ratios in [ sect : monochromqs ] and [ sect : tirqs ] , and separately consider a sample of high redshift ( @xmath12 2.5 ) sources in
[ sect : highz ] .
[ sect : agnfract ] is dedicated to the ir / radio properties of agn hosts and in [ sect : qvslum ] we study variations of ir / radio ratios with luminosity . + previous studies have carried out similar analyses using a variety of ir / radio diagnostics .
these include _ mips_-based monochromatic flux ratios @xmath4 and @xmath5 ( e.g. @xcite ; see equation ( [ eq : monchromqs ] ) for the definition of @xmath115}$ ] ) which we discuss in [ sect : monochromqs ] .
other studies have used the fir ( 42.5 - 122.5@xmath1 m ) to radio flux ratio @xmath96 ( see equation ( [ eq : firq ] ) ; e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , or the ratio of total ir luminosity ( @xmath78 ) to radio luminosity ( e.g. * ? ? ?
* ) : @xmath116 total infrared luminosities @xmath78 ( in units of [ w ] ) are calculated by integrating the sed between 8 and 1000@xmath1 m .
the rest frame 1.4ghz luminosity ( expressed in [ whz@xmath28 ] ) is @xmath118 where @xmath119 is the integrated radio flux density of the source and @xmath120 the luminosity distance .
the @xmath22-correction @xmath121 depends on the spectral index @xmath122 of the synchrotron power law @xmath123 . for the rest of the analysis
we will assume that @xmath124 0.8 @xcite .
we will return to the tir / radio flux ratios in
[ sect : tirqs ] .
the observed 24@xmath1m/1.4ghz flux ratio @xmath125 is plotted against redshift in fig .
[ fig : uncorrq24 ] for sfgs ( _ top _ ) and agn ( _ bottom _ ) .
sources are assigned to the two categories depending on whether pr(sf ) is larger or smaller than 50% ( cf .
[ sect : sfvsagn ] ) . + while there clearly are many radio - loud sources in our agn sample , fig .
[ fig : uncorrq24].b shows that a majority of the objects assigned to the agn category displays very similar ir / radio ratios as the sfgs . we will discuss this observation in more detail in [ sect : agnfract ] . at the same time
, the sample of sfgs also includes a number of radio - excess sources .
they usually have photometric redshift estimates and mostly lie at 1 @xmath25 3 .
this roughly corresponds to the redshift range in which photometrically determined redshifts are subject to the largest uncertainty because the 4000 break is only sampled by broad and widely spaced photometric bands . as a consequence , absorption features and emission lines from agn and
sf systems often interfere with each other in the same filter . even though we did attempt to remove all unreliable redshifts as described in
[ appsect : redshifts ] it thus seems probable that at least some of these cases are due to wrong distance estimates and hence to the selection of an inappropriate optical sed .
since this results in a faulty ( @xmath52 ) colour , the source in question could then have been assigned to the sf rather than the agn category .
another possibility is that the peak of agn activity at @xmath7 2 ( e.g. * ? ? ?
* ; * ? ? ?
* ) also influences the sf sample due to the statistical nature of the identification of sfgs and because especially agn in composite systems could have been classified as star forming .
lastly , we tried to assess if unobscured type 1 agn represent a significant fraction of the nominally star forming radio excess sources in the pertinent redshift range .
based on the confidence class ( see [ appsect : specz ] ) of those objects for which follow - up spectroscopy was available , we estimate that only @xmath35% are quasars classified as sf due to their blue colour .
+ the mean value of @xmath126 decreases as a function of redshift .
we will show later on ( see fig .
[ fig : q24agnfrac ] ) that this decrease agrees with the variations local lirgs ( detectable only out to @xmath7 1 ; see vertical dotted lines in panel ( _ a _ ) ) and ulirgs would display if redshifted . in fig .
[ fig : uncorrq70 ] we plot the observed 70@xmath1m/1.4ghz flux ratio @xmath127 of our sources as a function of redshift .
all symbols and colours are exactly as in the previous figure .
note that censored measurements due to flux limits at 70@xmath1 m are more frequent than was the case for @xmath126 because the 70@xmath1 m observations are much shallower .
the observed flux ratio before @xmath22-correction shows the same decline at higher redshifts as was seen for the observed 24@xmath1m/1.4ghz flux ratios . as described in [ sect : sedfitting ] all sources classified as star forming were fit with ir seds in order to derive ir luminosities . as a by - product of the template fitting
we can immediately obtain rest frame ( i.e. @xmath22-corrected ) 24 and 70@xmath1 m flux densities by convolving the best - fitting sed with the filter response functions of _ mips_. in the following we define the @xmath22-corrected 24 and 70@xmath1 m fluxes as the average of the values obtained from the libraries of @xcite and @xcite and use them to construct monochromatic rest frame flux density ratios @xmath128 and @xmath129 .
the associated 1.4 ghz flux densities have been @xmath22-corrected according to equation ( [ eq : l20 ] ) .
+ to quantify the evolution of @xmath128 in the joint ir- and radio - selected sample we : 1 .
bin the data such that each redshift slice contains an identical number of objects ( @xmath3250 ) .
the number of bins is kept limited to guarantee that the distribution function of @xmath128 is sampled sufficiently .
2 . run the iterative algorithm outlined in
[ appsect : doubcensetheory ] to find the cumulative distribution function of @xmath128 at each redshift .
the median immediately follows from this computation as does the scatter in the population which we obtain by fitting a gaussian distribution with known mean ( equal to the previously determined median ) to the distribution function .
the choice of the gaussian is motivated by the shape of the local ir - radio relation @xcite .
3 . determine the evolution of the average ir / radio ratio by fitting a linear trend line to the medians .
only measurements at @xmath130 1.4 are considered for this since the scatter at higher redshifts is found to increase abruptly , thus making the determination of the median uncertain .
steps 1 and 2 are are also carried out individually for the sample of ir- and radio - selected galaxies .
the results are shown in fig .
[ fig : sfcorrq24_medevo ] .
since the cumulative distribution function is normalized it lies in the range between zero and unity and can thus be regarded as the probability of obtaining a measurement of @xmath128 which is less than in the case of the radio - selected sample ( light grey curve ) or in excess of for the ir - selected sample ( dark grey curve ) the ordinate .
the distribution function of the doubly censored union of the ir- and radio - selected samples is also parametrized such that it runs from 0 to 1 with increasing @xmath128 .
it is plotted in black together with a dashed red line which shows the corresponding best - fitting gaussian distribution .
the intersection of the black curve with the 50% probability line ( dotted horizontal line ) defines the median value of @xmath128 .
[ fig : sfcorrq24_medevo ] demonstrates that the median of the radio - selected population lies systematically below that of the ir - selected objects .
the shift is approximately 0.35 dex at low redshift and grows to about 0.7 dex beyond @xmath7 1 .
the increase is probably caused by the intrinsically higher scatter in the ir - radio relation at high luminosities @xcite , possibly in combination with the reduced reliability of photometric redshifts and/or some falsely classified agn that begin to affect the sample starting at @xmath7 1.3 .
a shift of @xmath30.35 dex as observed at @xmath9 1 where the accuracy of our measurements is highest agrees fairly well with the prediction of equation ( [ eq : qbias ] ) and is hence a likely explanation for differences between previously reported average ir / radio properties of both local and high-@xmath11 galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) . in the lower panel of fig .
[ fig : sfcorrq24_medevo ] we plot the medians @xmath131 of the jointly ir- and radio - selected sfgs ( black dots ) at different redshift on top of the @xmath22-corrected values @xmath128 ( colours and symbols are identical to those in fig .
[ fig : uncorrq24].a ) .
the error bars mark the 95% confidence interval associated with the median .
table [ tab : q24info ] lists the median and scatter of @xmath128 which were determined with survival analysis in each of the redshift bins of fig .
[ fig : sfcorrq24_medevo ] .
in addition to the measurements carried out on the jointly ir- and radio - selected sample the table also provides the according values for the ir- and radio - selected samples individually . + using the 2@xmath35 errors on the medians as weights we fit them with a model of linear redshift evolution .
the best - fitting trend of @xmath131 vs. @xmath11 is shown in black in the lower window of fig .
[ fig : sfcorrq24_medevo ] ( see table [ tab : medqevo ] for the parametrization of the line ) . because the fit was carried out with respect to linear redshift space while the plot has a logarithmically scaled redshift axis it is curved . within the errors the slope @xmath132 is consistent with no evolution of the @xmath22-corrected 24@xmath1m/1.4ghz ratio at @xmath130 1.4 ( the maximal distance out to which the precision of the photometric redshifts is high ) .
the @xmath133-axis intercept of the trend line at @xmath134 0 is in agreement with the recent analysis of @xcite who find that @xmath135 1.22 with a scatter of 0.24 . in our sample
we find @xmath136 [email protected] ( where the error states the formal 1@xmath35 uncertainty from the linear fit ) . for a comparison between the average ir / radio properties of radio - selected samples we can refer to the studies of @xcite and @xcite ; they report an average @xmath135 0.94 - 1 , depending on the ir - sed adopted to @xmath22-correct to the rest frame .
these values agree well with the range of medians @xmath138 [ 0.8 , 1 ] measured for radio - selected cosmos data at intermediate redshift ( cf .
left - most column of table [ tab : q24info ] ) .
our convention for choosing sfgs states that pr(sf ) must be at least 50% . in fig .
[ fig : sfcorrq24varevol ] we assess how changing this threshold affects the redshift evolution that is inferred from the data .
the variation of the parameters of the best - fitting evolutionary trend line is shown in the upper- and lower - most window of fig .
[ fig : sfcorrq24varevol ] ( @xmath133-axis intercept and slope , respectively ) . a black symbol in the middle of the displayed data range marks the results that were shown in fig .
[ fig : sfcorrq24_medevo ] .
they are fully consistent with the evolution found if a more conservative threshold e.g. at pr(sf ) = 66% for the selection had been chosen .
it is interesting that the inclusion of a significant fraction of sources with a probability of up to 80% of being agn does not alter the results either .
this is a strong indication that our sample of optically selected agn contains many objects with ir / radio properties that closely resemble those of sf systems .
similar observations were made by , e.g. , @xcite and @xcite , who studied local samples of radio - quiet quasars and/or seyfert 1 sources lacking a compact nucleus .
the middle row of fig .
[ fig : sfcorrq24varevol ] shows that while the average values of @xmath128 are similar for many sfgs and agn the latter are subject to a larger scatter as was previously found by , e.g. , @xcite . + as an additional test of the robustness of our findings we checked if the evolutionary trend in sf samples selected through ( @xmath52 ) or p1 differs . for the radio - selected sample where both colours were available we found equivalent results regardless of the chosen approach .
figures [ fig : sfcorrq70_medevo ] and [ fig : sfcorrq70varevol ] ( the results of which are summarized in table [ tab : q70info ] ) repeat the analysis of figs .
[ fig : sfcorrq24_medevo ] and [ fig : sfcorrq24varevol ] for the @xmath22-corrected 70@xmath1m/1.4ghz flux ratio .
note that in comparison with fig .
[ fig : uncorrq70 ] the number of censored measurements is much smaller when we consider rest frame ir / radio ratios @xmath129 rather than observed flux ratios @xmath139 .
the reason is that we do not require a 70@xmath1 m detection for the ir template fitting but also fit objects which have a limit at 70@xmath1 m and a direct detection at 24@xmath1 m . +
the plot of @xmath129 vs. redshift ( lower panel of fig .
[ fig : sfcorrq70_medevo ] ) as well as fig .
[ fig : sfcorrq70varevol ] , which illustrates the stability of the findings with respect to changes in the selection criterion for sfgs , show that @xmath129 behaves in a similar way as was found for @xmath128 .
the extrapolated average 70@xmath1m/1.4ghz flux ratio at @xmath134 0 , @xmath140 equals [email protected] and if agn are included the scatter in the relation increases in analogy to what was found for @xmath128 . as for @xmath4 , the evolutionary slope @xmath141 ( slope and normalization of the evolutionary trend line
are logged in table [ tab : medqevo ] ) is consistent with zero .
+ the average ir / radio properties of the radio - selected samples of @xcite and @xcitem sources the sample of @xcite becomes essentially a radio - selected sample at the stage when sources without a counterpart in the 1.4ghz radio catalog of the fls field @xcite are removed from the sample .
] are [email protected] and [email protected] , respectively .
although the agreement with our findings is not quite as good as in the case of @xmath4 , they are are nevertheless consistent ( within both the formal error and the scatter ) with the range of medians @xmath142 [ 1.7 , 2.1 ] at intermediate redshift in our radio - selected sample ( see table [ tab : q70info ] ) .
to our knowledge there so far has been no comparable study which uses an ir - selected sample to compute an average @xmath5 . in the local universe
the correlation of ir and radio flux is tightest if integrated ( f)ir luminosities rather than monochromatic flux ratios are considered . to complement the analysis of [ sect : monochromqs ] we thus show in this section the correlation of tir ( 8 - 1000@xmath1 m ) and 1.4ghz luminosity as parametrized by the tir / radio ratio @xmath6 for our vla- and s - cosmos data .
the computation of the distribution functions for the parameter @xmath6 is carried out following the same steps described in [ sect : monochromevo ] .
the results are shown for a number of redshift bins in fig .
[ fig : sfqtir_medevo ] where we also compare the median derived for the jointly ir- and radio - selected sfgs with the local value of @xmath1430.02 ( @xcite ; vertical dashed line ) .
our average values @xmath144 in the range @xmath9 1.4 ( see table [ tab : qtirinfo ] ) lie to either side and always remain well within the dispersion of the local measurement of @xcite .
+ the evolution of @xmath6 is shown in the lower panel of fig .
[ fig : sfqtir_medevo ] using the same presentation of the data as for the monochromatic ir / radio flux ratios .
since the latter were derived based on the ir templates which are used here to calculate the integrated ir luminosity , we expect by construction that the evolutionary trend is in good qualitative agreement with the findings of [ sect : monochromqs ] . for the same reason we can not expect to observe a reduced scatter in the values of @xmath6 with respect to those of the monochromatic flux ratios as the spread in the properties of the best - fitting ir seds must manifest itself in @xmath128 and @xmath129 as well .
+ the line parameters for the evolution of @xmath144 are given together with those of @xmath131 and @xmath145 in table [ tab : medqevo ] : in contrast to @xmath128 and @xmath129 the best - fitting evolutionary trend for @xmath144 suggests a decrease of the average tir / radio ratio by 0.35 dex out to @xmath7 1.4 .
however , this slope is detected at the 2@xmath35 significance level and predicts a median @xmath144 at @xmath7 1.4 that still lies within the dispersion measured in our lowest redshift bin .
it thus seems unlikely that the evolutionary signal is real , especially in view of the results of [ sect : highz ] where we measure an average @xmath146 that is in excellent agreement with the local value for a subset of highly redshifted galaxies in the cosmos field .
an examination of the evolutionary slopes for @xmath131 , @xmath145 and @xmath144 in table [ tab : medqevo ] shows that they become more negative along this sequence .
this could be related to a number of radio - excess sources with 1 @xmath147 3 which are part of our optically selected sf sample ( visible as a diffuse cloud of upper limits and detections below the main locus of symbols in all our plots of @xmath84 vs. @xmath11 ; see also our comment in
[ sect : obsmonochrom ] ) and that tend to lower the average ir / radio in this redshift range . if these objects were falsely classified composite sources or agn the increased emission at 24@xmath1 m from their hot dust might be able to compensate the radio - excess , thus leading to zero evolution in @xmath128 as observed .
@xmath129 and @xmath6 on the other hand sample mainly ir light from star formation and hence are lowered in the presence of excess radio emission .
this scenario can also explain why the evolutionary slope of @xmath131 is insensitive to the selection criterion for sfgs ( cf .
[ fig : sfcorrq24_medevo ] ) while it varies in the same sense as described above in the case of @xmath145 and @xmath144 .
the analysis of the previous sections revealed ( cf . figs .
[ fig : sfcorrq24varevol ] , [ fig : sfcorrq70varevol ] and [ fig : sfqtirvarevol ] ; also figs .
[ fig : uncorrq24 ] and [ fig : uncorrq70 ] ) that the ir / radio properties of sfgs are shared by many of the agn - bearing systems in our sample . in this section
we will study this in more detail .
we first test ( [ sect : xmmonsf ] ) if it remains valid for a subsample of sources which are detected in x - rays and have been found to host an agn using a different approach than the classification scheme introduced in
[ sect : sfvsagn ] . in
[ sect : cohabfract ] we then compute ( in different redshift bins at @xmath10 1 - 1.4 ) the relative frequency of agn and sf sources as a function of the ir / radio ratio .
at the sensitivity of the _ xmm - newton _ observations of the cosmos field a large fraction of the detected sources is expected to be powered by agn .
this is confirmed by @xcite who have shown that @xmath370% of the _ xmm - newton _ sources have uv to nir seds which contain an agn component . in fig .
[ fig : xmmqs ] we compare the observed 24@xmath1 m and 70@xmath1 m to radio flux density ratios @xmath148 and @xmath139 of x - ray detected agn hosts at different redshifts with the predicted ir / radio properties of model sfgs ( coloured tracks - corrections between ( _ i _ ) the flux density at the rest frame ( @xmath149 ) and redshifted effective wavelength ( @xmath150 ) of the _ mips _ filter and , analogously , ( _ ii _ ) that applied to the 1.4ghz band .
the @xmath22-correction @xmath151 is defined as the ratio of the rest frame luminosity @xmath152 and the luminosity at wavelength @xmath150 which is sampled by the observer s measurement of the flux @xmath153 : @xmath154 here @xmath120 is the luminosity distance .
the @xmath22-corrections used in the conversion of observed ir flux measurements at 24 and 70@xmath1 m to rest frame quantities depends on the shape of the ir sed of the galaxies . for the radio flux it has the form given after equation ( [ eq : l20 ] ) . ]
; see footnote and the text of [ sect : cohabfract ] for additional details ) .
note that according to the analysis of @xcite the agn contribution to the uv - nir sed exceeds 50% for most of these sources . from fig .
[ fig : xmmqs ] it is obvious that a majority of the _ xmm - newton _ sources have ir / radio ratios that are perfectly consistent with those expected for starbursts .
they are genuine examples of active galaxies in which the agn , although significantly contributing to the sed at optical and x - ray wavelengths , does not cause significant excess radio emission .
[ fig : xmmqs ] therefore is strong evidence that the findings of [ sect : monochromqs ] and [ sect : tirqs ] can not be ascribed to an inadequacy of the method we adopted to distinguish between agn and sfgs .
we first define the ` main ' locus of sfgs in a plot of _ observed _ ir / radio ratio vs. redshift . working with observed flux densities is necessary because we want to avoid imposing a template fit with the ir sed of a sfg on an agn - bearing source even if it quite probably shares similar ir / radio properties . at each redshift
the star forming locus is centred on the average value
@xmath155,\,template}\rangle$ ] of @xmath148 as predicted by the model seds of sources in the observable range of ir luminosities .
we then consider a region between + 2@xmath91 and -2@xmath91 around @xmath155,\,template}\rangle$ ] in which we chart the relative frequency @xmath156 of agn .
the analysis is restricted to this @xmath157 band because beyond it the sparse sampling of the distribution function of the ir / radio ratios leads to unwanted fluctuations of @xmath156 .
the value of@xmath91 is a representative average of the scatter in our data for the sf population at @xmath10 1 , i.e. 0.35 dex .
+ in the upper panel of fig .
[ fig : q24agnfrac ] ( fig .
[ fig : q70agnfrac ] shows the same information for @xmath5 ) we show the expected variations in @xmath158 of high-@xmath11 galaxies assuming that their seds at ir and radio wavelengths are similar to those of local sfgs .
seds from three different template libraries as well as that of the starburst m82 are shown for different @xmath78 .
the tracks are normalized at @xmath134 0 using the best - fit evolutionary trend line displayed in fig .
[ fig : sfcorrq24_medevo ] . in the background we re - plot ( cf . fig .
[ fig : uncorrq24].a ) the observed 24@xmath1m/1.4ghz ratios of our sample of sf galaxies in order to show how they nicely follow the tracks of the local seds .
the solid black lines delineate the @xmath159 band centred on @xmath160 .
the jumps at @xmath7 0.5 and 1 occur because the averaging of the ir templates is performed with a discrete and restricted set of ir templates .
since we merely use these boundaries to define the parameter space for the subsequent analysis the discontinuities are inconsequential . + the expression for the relative agn abundance which accounts for censored measurements and the use of discrete probability bins is ( see derivation in appendix [ appsect : cohabderiv ] ) @xmath161 here the summation with respect to @xmath162 extends over a finite number @xmath105 of probability bins . in the case of @xmath4 we grouped sources into bins of width @xmath163pr(sf ) = 0.1 in order to have a sufficient number of measurements , and thus to ensure a well - behaved estimate of the distribution function @xmath164 ( computed according to equations ( [ eq : schmittprob ] ) and ( [ eq : schmittcoupled ] ) ) in each probability bin . in the lower panel of fig .
[ fig : q24agnfrac ] we present the function @xmath165 in four redshift bins covering @xmath10 1.4 .
the zero - point of the @xmath166-axis has been renormalized to the average @xmath4 of the ir templates at the centre of the redshift slice .
a value of 2 ( 0.5 ) on the @xmath133-axis implies that at a given value of @xmath4 the relative abundance of agn and sf systems is 2:1 ( 1:2 ) . within the errors @xmath165 is consistent with being unity across the whole width of the star forming locus at all redshifts .
deviations from the generally smooth variations of @xmath165 can occur on the edge of the assessed range of @xmath4 due to fluctuations caused by poor statistics .
there is weak evidence for a gradual decrease of the agn fraction from about @xmath167 to roughly @xmath168 as one goes from the region which hosts sources with radio - excess to that populated by sources with excess ir emission .
this trend is barely significant but interestingly enough it tilts in the opposite direction as would be expected if , e.g. , agn activity were to manifest itself by exciting increased hot dust emission in the mir .
( note that in general the radio emission could also be altered by the presence of an agn , thus making the observed slope less easily interpretable .
however , the fact that the distribution of @xmath169 which has a radio contribution that is identical to that in @xmath170 is essentially flat , suggests that the radio emission is not strongly affected by the agn . ) + the calculation of @xmath171 involved slightly wider probability bins of width @xmath172 ( to ensure convergence of the distribution function ) and was limited to @xmath10 1.1 due to the ubiquity of 70@xmath1 m non - detections at higher redshift ( cf .
upper panel of fig .
[ fig : q70agnfrac ] ) .
@xmath171 appears to be a constant function of @xmath148 with no traces of being tilted as detected with marginal significance for @xmath165 , except maybe in the redshift bin @xmath8 [ 0.72 , 1.06 ] .
overall , we can thus deduce that our optically selected agn and sfgs occupy the sf locus in very similar proportions .
a possible explanation for this is that both the ir and radio emission are predominantly powered by star formation rather than agn activity .
it is also conceivable , however , that other ( combinations of ) astrophysical processes conspire to place agn hosts close to the ir - radio relation ( e.g. * ? ? ?
* ; * ? ? ?
* ) . in a recent work on local ir galaxies
@xcite have found evidence of variations in the @xmath22-corrected average 24@xmath1m/1.4ghz flux ratio @xmath128 with ir luminosity @xmath78 . according to their analysis @xmath128 is a constant function of luminosity at @xmath173 and then begins to rise with increasing luminosity . using the nvss- and iras - detected sdss galaxies , mori et al .
( in prep . ) see an opposite trend of decreasing fir / radio ratio when they examine @xmath96 vs. @xmath174 for various types of active galaxies ( both star forming and agn - bearing ) .
we investigate whether or not the star forming sources in our sample show any evidence of variations of @xmath6 with ir or radio luminosity .
since our @xmath22-corrected monochromatic ir / radio ratios are based on the best - fitting tir seds , all luminosity - dependent trends they display will be qualitatively identical to those measured for @xmath6 .
comparisons with previous studies are therefore possible even if these used a different ir / radio parameter .
+ note that the fact that we are plotting @xmath6 against luminosity implies that upper and lower limits can not always be unambiguously placed along the ordinate .
an example are the radio - selected sources in the upper panel of fig .
[ fig : sfqtir_vs_lum ] of which we merely know that they must lie to the lower left of their limits .
they are indicated by an arrow pointing diagonally downward .
the calculation of the median @xmath175 in a given bin of luminosity should correct for measurements that in truth belong to a fainter luminosity range . to account for this
we construct broad luminosity bins ( @xmath176 1 ) and assume that most of the censored measurements would come to lie in the next lower luminosity bin ( anything fainter would imply that they are more than 3@xmath35 outliers to the ir - radio relation ) . we can then ` average ' away the effect of falsely assigned measurements by ( _ i _ ) computing the median @xmath144 in two sets of luminosity bins which are offset by half a bin width and then ( _ ii _ ) averaging the two estimates of the median thus obtained and reporting the new value half way between the centres of the two involved bins along the luminosity axis .
the medians themselves are calculated by applying survival analysis to the jointly ir- and radio - selected data as previously done in [ sect : monochromqs]-[sect : agnfract ] .
the results of this procedure are shown in the larger two windows in fig .
[ fig : sfqtir_vs_lum ] . using the cosmos data we see no evidence of an increase in the ir / radio ratio at @xmath177 as suggested by @xcite .
we do detect a higher value of @xmath144 in the brightest ir luminosity bin but this increase happens around @xmath178 , similar to the results of @xcite .
it should be mentioned , however , that the methodology used by @xcite to derive @xmath128 differs significantly from the one used here in that it involves for example luminosity - dependent ( and template - based ) conversions of _ iras _
25@xmath1 m flux densities to 24@xmath1 m _ mips _ equivalent values .
+ while no universal trend for variations of @xmath84 with ir luminosity are detected in our sample we do find that @xmath6 is a decreasing function of radio luminosity ( see lower - most window in fig .
[ fig : sfqtir_vs_lum ] ) .
the trend is consistent and increases rapidly at @xmath179w / hz .
this could potentially be the effect of contaminating agn at high radio luminosities in our optically selected sample of sfgs . however , the fact that mori et al .
( in prep . ) see a similar trend in local sf , composite and agn - bearing systems which have been classified based on the standard optical line emission ratios @xcite suggests that the trend is genuine .
+ the two narrower windows in fig .
[ fig : sfqtir_vs_lum ] show the variations of the dispersion of @xmath6 with ir ( _ top _ ) and radio luminosity ( _ bottom _ ) . in the low - redshift samples of @xcite and @xcite an increase in scatter with infrared luminosity is detected . in the present data a similar albeit very weak
tendency is seen ; the reduced accuracy of the @xmath78 measurements of the high-@xmath11 galaxies likely masks most of the trend if present . the plot of @xmath180 vs. @xmath174 , on the other hand shows a clear increase in the scatter which starts to manifest itself at the same radio luminosity at which the strong decline of @xmath144 sets in . while in the previous sections we usually tacitly plotted data points from high-@xmath11 sources the fitting of evolutionary trends in [ sect : monochromqs ] and [ sect : tirqs ] was restricted to galaxies at @xmath9 1.4 .
this corresponds to the redshift at which the 4000 break leaves the reddest subaru band with deep coverage @xcite , the @xmath11-band .
after @xmath7 1.4 the break is constrained by the nir data of the @xmath60 , @xmath181 and @xmath61 bands ( mccracken et al .
2009 , subm . ; capak et al . , in prep .
these exposures of the cosmos field , however , are two magnitudes shallower and have gaps between filters , leading to large uncertainties in the photometric redshift estimates . beginning from about @xmath7 2.5 the ly@xmath122 ( 1215 ) break enters the wavelength range covered by the ground based photometry @xcite . as
a consequence the accuracy of the photometric redshift improves to again @xmath182 0.03 .
+ in an assessment of ongoing spectroscopic follow - up observations of high-@xmath11 sources in the cosmos field , capak et al .
( in prep . )
find that photometric redshift estimates of genuine high-@xmath11 sources may be scattered to low redshift due to confusion between the ly@xmath122 and 4000 break .
most of the confusion is due to regions of the ly@xmath122 forest which are not as opaque as expected and/or light from nearby foreground galaxies contaminating the apertures .
conversely , there is little evidence for any upward scattering of galaxies at low and intermediate redshift to @xmath12 2.5 .
this implies that sources with photometric redshift estimates @xmath132.5 represent , with high likelihood , a clean albeit not complete sample of high-@xmath11 objects .
our sample contains more than @xmath3140 sources at redshift @xmath79 2.5 , of which approx .
60% have direct detections at 1.4ghz and in at least one _ mips _ filter .
as far as we are aware , this is the largest sample of high-@xmath11 sources so far , for which it is possible to study the ir - radio relation based on direct detections rather than flux limits .
we must point out , however , that only 2% of the high-@xmath11 sources have a direct detection at 24 _ and _ 70@xmath1 m while the seds of the remaining 98% are only constrained by a direct detection at 24@xmath1 m and an upper flux limit at 70@xmath1 m .
accordingly , the calculated values of @xmath78 luminosities must be regarded as fairly rough estimates of the true ir luminosity of these sources as they are primarily based on measurements made at a rest frame wavelength of @xmath36@xmath1 m .
@xcite caution that the ir luminosities of high luminosity and high redshift sources ( @xmath183 ; @xmath79 1.4 ) are generally overestimated by a factor of @xmath34 even after subtraction of a flux contribution from agn . however , in view of the cosmos study of kartaltepe et al .
( subm . ) who , in the same range of ir luminosities , do not see this trend and instead report that ir luminosities based solely on 24@xmath1 m data tend to be underestimated in general we refrained from applying any corrections to our data .
+ bearing in mind these uncertainties we plot the tir / radio ratios of our high-@xmath11 sources in fig .
[ fig : sfqtir_medevohighz ] ( left panel ) . for illustrative purposes the measurements of @xmath6
are coloured according to their probability pr(sf ) .
we caution , however , that this classification is based on the fiducial ( @xmath52 ) cut used throughout the paper so far and that the evidence presented in fig .
[ fig : probcol_distr ] indicates that this threshold is no longer appropriate at @xmath12 3 . in view of this
we do not distinguish between star forming systems and agn for the high-@xmath11 sources but use this global sample to derive the average ir / radio flux ratio .
the right hand side shows the distribution function of @xmath6 which is broad ( @xmath1840.05 ) and has a median of 2.71@xmath185 .
this value is in good agreement with the local measurement of @xcite ( dashed line ) and is almost identical to the average value of 2.76@xmath186 we find for the cosmos data in our lowest redshift bin in
[ sect : tirqs ] .
the average ir / radio properties of our high redshift sample the most distant sources of which are detected when the universe was only @xmath31.5 gyrs old are thus very similar to those observed in the local universe .
it is important to remember , however , that at @xmath187 2.5 the cosmos data contains mostly extremely ir - luminous hylirgs ( @xmath188 ) which are a very different kind of object than those encountered at @xmath9 0.5 where the majority of our sources have @xmath189 ( cf .
[ fig : fluxlims ] ) .
various parameterizations of the ir - radio relation exist . the flux ratios @xmath190 and @xmath5 predominantly reflect the ir and radio emission of the ism which is caused by two stages in the life cycle of massive stars ; ( _ i _ ) the main sequence phase during which uv light is converted into fir emission by dust grains and ( _ ii _ ) supernovae explosions inducing synchrotron emission when their shock waves accelerate cosmic ray electrons in the galactic magnetic field .
the parameter @xmath4 , on the other hand , is more sensitive to hot dust emission triggered by agn activity .
several recent papers ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) using 1.4ghz data provide consistent evidence that the local ir - radio relation holds out to high redshift .
an identical conclusion has been reached using radio flux density measurements at 610mhz rather than 1.4ghz @xcite .
+ a majority of the samples previously used to study the ir - radio relation in distant galaxies contain only several dozen to roughly a hundred galaxies at @xmath10 0.5 . the cosmos sample used in the present
analysis increases the number of sources at redshifts @xmath12 0.5 by at least a factor of 5 . also , it is probably the first data set in which the number of directly measured ir / radio ratios at high redshift is larger than the number of censored values .
+ we have studied the ir / radio properties of both sfgs and agn .
various complementary indicators of agn activity are present in the cosmos data base . rather than identifying agn based on a combination of multiple parameters
we have chosen to work with a single statistical criterion based on the work by @xcite which classifies radio- and ir - detected sources as sf or agn based on their rest frame optical colour .
this statistical approach allows us to treat our sources in a probabilistic way which is especially advantageous for the assessment of systematics inherent in the selection of sfgs and agn . at @xmath9 1.4 where both photometric and spectroscopic redshifts have a high accuracy we found no compelling evidence of an evolving ir - radio relation .
the mean ir / radio flux ratios in this redshift range are @xmath191 , @xmath192 and @xmath193 , where the first two are computed using @xmath22-corrected ir and radio flux densities . with the exception of the highest redshift bin ,
the median ir / radio ratios in the different redshift bins covering the range @xmath8 [ 0 , 1.4 [ are offset from the averages @xmath194 by at most half the dispersion in the local ir - radio relation .
an alternative to testing the constancy of rest frame ir / radio flux ratios derived using ir seds of local sfgs , is to form ir / radio ratios with observed flux densities ( cf .
[ sect : obsmonochrom ] ) .
variations with redshift of @xmath148 and @xmath139 in the sf population can then be compared with the changes that would be expected for redshifted local galaxies ( e.g. * ? ? ?
* ; * ? ? ?
* ) in order to assess if the ir / radio properties of the latter are compatible with those of high-@xmath11 systems .
our analysis has shown that the decline with redshift of observed and predicted ir / radio ratios are indeed in good agreement .
+ since measurements of @xmath4 and @xmath5 have been carried out with flux - limited data sets which likely are subject to some selection band - related bias ( see [ sect : selbias ] and [ sect : biasconclu ] ) only our value of @xmath6 lends itself to a straightforward comparison with local measurements .
@xcite find @xmath195 for a sample of local sfgs and show that this figure is in excellent agreement with the _ iras_-based fir measurement @xmath196 of @xcite .
the cosmos measurements of the mean ir / radio properties of high redshift galaxies thus are fully consistent with the local average .
furthermore , a subsample of highly redshifted ( @xmath79 2.5 ) cosmos galaxies has a median 2.71@xmath185 which also agrees well with both the cosmos data at low redshift and independent local measurements .
+ due to malmquist bias we detect systems with very differing star formation rates ( sfr ) in the low and high redshift universe where only extreme starbursts of the hylirg class ( sfr @xmath197yr@xmath28 ) are visible .
that starburst galaxies which we observe when the universe was just 10% of its current age follow the same ir - radio correlation as local galaxies runs counter to expectation which would predict higher - than - average ir / radio ratios for such systems .
@xcite have shown that a strong ( and prompt ) enhancement of the magnetic field strength is required lest a deficit of synchrotron emission develop due to high inverse compton losses of the cosmic ray electrons in the strong radiation field generated by the starburst .
+ on the other hand , it could be that precisely this inability to detect sources with sfrs of ` normal ' ( disk ) galaxies at higher redshift is hiding changes in the ir / radio ratios of the sf population . in particular ,
such changes are expected as soon as the energy density of the cosmic microwave background ( cmb ) exceeds that of the galactic magnetic fields , whereupon inverse compton losses off cmb photons begin to dominate synchrotron emission .
based on the typical magnetic field strengths in spiral arms ( a few @xmath1 g ) , @xcite estimate that this could be the case for regular disks as early as @xmath7 0.5 .
+ regardless of the remaining uncertainties surrounding the maintenance of the local ir - radio relation out to high redshift the observational fact _ per se _ is an important confirmation of the central assumption used in studies that have computed the sf history of the universe @xcite using deep radio surveys , namely that of an unchanging proportionality between sfr and radio luminosity .
the constancy of the ir - radio relation not only implies that ir and radio measurements are equally good tracers of star formation out to high redshift .
it also suggests that the physical processes of massive star formation when the universe was only 1 - 2 gyrs old used to be strikingly similar to those at play in the local universe .
+ currently available observations are not yet capable of revealing the exact workings of the astrophysical mechanisms that cause galaxies at intermediate and high redshift to lie on the ir - radio relation .
the improvements in sensitivity and spatial resolution with _
herschel _ and the extended very large array ( evla ) will be a crucial step forward in this respect . a better sampling of the fir emission from distant ( u)lirgs will reveal if their seds indeed match those of their local analogues as this study and previous work generally assume .
similarly , a systematic study of the range of radio spectral indices in sfgs is important to quantify the inaccuracies that are introduced by assuming a single generic value .
clarifying both these issues is essential if , e.g. , trends for a luminosity dependence of @xmath194 ( see * ? ? ?
* and also our fig .
[ fig : sfqtir_vs_lum ] ) are to be put on a solid basis .
apart from sfgs , the ir - radio relation is also observed by many agn - bearing and/or composite systems ( e.g * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , albeit with a larger dispersion @xcite . in the cosmos sample
we also observe that agn and sfgs often have very similar values of @xmath4 and @xmath5 .
this is true not only for optically - selected agn but also for x - ray detected sources , implying that the phenomenon is not merely an artifact of our statistical colour criterion for the separation between agn and sfgs .
it should be emphasized that the finding is _ not _ an artificial consequence of template fitting ; the abundance fraction of agn and sfgs has been derived using observed ir / radio ratios and hence does not involve any assumptions about the value of the radio spectral index or the shape of the ir sed .
+ we find that our optically - selected agn and sfgs populate the locus of the correlation in nearly equal proportions out to at least @xmath7 1 .
it is important to bear this in mind when ir / radio ratios are used possibly in combination with other indicators to distinguish between agn and sf sources ( e.g. * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) : radio - excess outliers can only be used to single out radio - loud agn rather than a complete agn sample and , conversely , selecting only objects that follow the correlation will result in a mixture of radio - quiet agn and sfgs rather than a pure sf sample . the fact that an agn is present does not exclude coeval star formation in the host galaxy ( e.g. * ? ? ?
. however , whether or not it is responsible for the similar in ir / radio properties of sfgs and agn is still debated , with both supporting and contesting evidence being advanced ( e.g. in the case of pg qsos : see * ? ? ?
* ; * ? ? ?
even though the current data set has been selected both in the ir and radio , the fact that the average ir / radio ratios of the jointly selected sample are usually close to those of the ir - selected sample suggests that the average values @xmath198 we quote are not the ` intrinsic ' value one would hope to find in an entirely unbiased sample . by separately studying an ir- and radio - selected sample of sfgs , however , it is at least possible to bracket the unbiased average ir / radio properties .
as expected the jointly selected sample always lies within this region ( with the exception of one case
see table [ tab : qtirinfo ] where the amount by which the median of the jointly selected sample exceeds that of the ir - selected sample is still smaller than the width of the bins used for the calculation of the distribution functions of @xmath84 in the doubly censored data set ) .
the shift between the average ir / radio properties of an ir- and a radio - selected sample are in principle predictable based on the dispersion of the ir - radio relation and the slope of the differential source counts ( cf . equation ( [ eq : qbias ] ) ) . while comparable to the locally measured dispersion at intermediate redshift , the scatter of the relation in our cosmos sample is significant at @xmath79 1 where it is twice as large as at low redshift .
part of the increase could be due to an intrinsically higher scatter @xmath91 at high luminosities @xcite but probably is also a consequence of the limitations that are inherent in the calculation of ir luminosities when the ir sed is constrained by few points . under the simplified assumption of euclidian source counts ( @xmath199 2.5 ) equation ( [ eq : qbias ] ) predicts the offset @xmath200 0.35 dex , found in our lowest redshift bins .
it also makes a fair prediction of a shift of @xmath30.7 dex between the ir- and radio - selected sample at @xmath7 1 if one accounts for the larger scatter and the finding that at faint fluxes ir @xcite and radio @xcite source counts are sub - euclidean tends to increase @xmath102 while the sub - euclidean counts ( with @xmath201 1.5 ; see references in text ) have the opposite effect ( with respect to a euclidean slope ) .
consider the different measurements of @xmath144 at @xmath8 [ 1.14 , 1.51 [ ( see third row from bottom in table [ tab : qtirinfo ] ) as an illustration that the combination of the two factors leads to a prediction of @xmath102 which agrees excellently with the data . using equation ( [ eq : qbias ] ) with @xmath202 1.5 and @xmath203 0.78 one finds @xmath98 0.70 .
the observed shift between the median @xmath144 in the ir- and radio - selected sample , on the other hand , is @xmath30.67 . ] .
[ fig : sfcorrq24_medevo ] , [ fig : sfcorrq70_medevo ] and [ fig : sfqtir_medevo ] show that apart from biasing the average ir / radio ratio , selection effects can also produce spurious evolution . based on the radio - selected sample alone
we would infer a decrease of the mean @xmath198 out to @xmath7 1 ( see also numerical values in tables [ tab : q24info ] , [ tab : q70info ] and [ tab : qtirinfo ] ) . as the only recent study , @xcite measure an average value of @xmath5 for their radio - selected sample that is reduced by 0.25 dex at @xmath7 1 with respect to low redshift .
this is close to the evolution @xmath204 0.2 - 0.3 dex which we see for the radio - selected cosmos sample .
based on the argument outlined above a simple explanation for this could be selection effects .
however , as @xcite adopt the different approach of stacking radio sources that are not detected at 70@xmath1 m rather than including them in the analysis in the form of flux limits as we have done , other explanations can not be ruled out . as stated in [ sect : selbias ] there is ample evidence from the results summarized in table [ tab : prevwork ] that the offset predicted by equation ( [ eq : qbias ] ) not only occurs in our data but also can reconcile most apparently discrepant measurements of mean ir / radio ratios in the literature .
the one exception to this generally encouraging agreement are the highly inconsistent radio stacking results of @xcite , @xcite and @xcite who have all studied the mean @xmath4 as a function of ir flux .
@xcite in particular pointed out that the field - to - field variation of the mean ir / radio ratio can be considerable .
the prospects are good that the issue will soon be resolved with the aid of significantly deeper evla observations at the @xmath3@xmath1jy level that will even directly detect the radio emission of the faintest 24@xmath1 m sources .
on the preceding pages we have discussed the ir / radio properties of star forming galaxies ( sfgs ) and active galactic nuclei ( agn ) in the redshift range 0 @xmath25 5 .
our analysis has benefitted from the extensive multi - wavelength coverage of the cosmos field : each of our sources has a flux constraint at 24@xmath1 m , 70@xmath1 m and 1.4ghz from _ spitzer / mips _ and vla observations , multiwavelength photometry in @xmath330 bands from the uv to the nir , and in some cases x - ray detections .
roughly one in three of our sources has a spectroscopic , the rest an accurate photometric redshift measurement .
+ our primary focus was the evolution of the ir - radio relation with cosmic time . with a total of approx .
4500 sources our sample is the largest one which has so far been used to study the ir - radio relation at intermediate and high redshift .
this is particularly true at redshift @xmath79 2.5 where we have detected nearly 150 sources of which , for the first time , more objects have direct ir and radio detections rather than upper flux limits in one of the two bands . to our knowledge
this is also the first time the evolution of both monochromatic and integrated ir / radio ratios has been consistently derived using objects from the same field .
the computation of ir luminosities using complete sed template libraries represents a further improvement over previous work which has often relied on single starburst templates , e.g. that of m82 .
the average ir / radio ratio is subject to selection biases in that it depends on the band in which a population is selected .
we have shown that the average ir / radio properties of ir- and radio - selected samples of galaxies differ by an amount which is in agreement with theoretical expectations and that studies relying on objects selected in only one of the two bands run the risk of inferring spurious evolutionary trends .
for this reason , we base our analysis on a sample jointly selected at ir _ and _ radio wavelengths in order to eliminate biases as best possible .
furthermore , we make frequent use of the methods of survival analysis which permit us to include all information carried by flux limits from sources that have failed to be detected at either ir or radio wavelengths . to summarize our findings : 1 .
the median ir / radio ratios of sfgs show little variation in the redshift range @xmath9 1.4 and the ir - radio relation remains similarly tight out to @xmath7 1 as it is in the local universe ( see [ sect : monochromqs ] and [ sect : tirqs ] ) . above @xmath7 1
the dispersion in the cosmos population increases , probably due to the reduced accuracy of our measurements and/or an intrinsically larger dispersion of ir / radio ratios at high luminosities .
2 . for a sample of sources with high confidence redshift estimates in the range
@xmath79 2.5 , the average ir / radio ratio is still the same as that found in the local universe ( see [ sect : highz ] ) .
3 . at both 24 and 70@xmath1 m
many of our optically - selected agn have similar ir / radio ratios as sfgs ( see [ sect : agnfract ] ) .
the relative abundance of agn and sfgs in our sample is about 1:1 .
the same applies to x - ray detected agn , of which a large fraction has ir / radio ratios which lie in the range measured for star forming systems .
4 . the median ir / radio ratios of sfgs consistently decrease as a function of radio luminosity . on the other hand , they remain constant over three orders of magnitude in ir luminosity . only at the highest ir luminosities ( @xmath205 )
has a tendency for an increase been detected ( see [ sect : qvslum ] ) .
apparently discrepant measurements of the average ir / radio ratio reported in the literature can be reconciled if one properly accounts for the selection band of the respective samples ( see [ sect : biasconclu ] ) . in this work
we have applied a statistically more sophisticated treatment of ir / radio ratios than has previously been adopted .
it accounts for both detection limits and selection biases inherent in the data , and our results provide firm support for previous reports that the ir - radio relation remains unchanged out to high redshift ( @xmath12 4 ) .
the observed constancy is striking evidence that the interplay between the life cycle of massive stars and the interstellar matter ( ism ) has followed a very similar pattern for more than 10 billion years .
ir and radio measurements apparently represent equally good tracers of star formation over much of the history of the universe . this has been a critical assumption underlying measurements of the cosmic star formation history with deep radio surveys , and fundamental to the estimation of redshifts for optically undetected sub - mm galaxies .
+ the upcoming generation of ir and radio observatories like _ herschel _ and the extended very large array ( evla ) are expected to be able to perform measurements of sufficient sensitivity and spatial resolution to provide clues on the mechanism shaping the ir - radio relation in distant star forming galaxies with moderate star forming rates .
this will be a major advance over the mere observation of the phenomenon as is presently the case .
in the following , additional information on our data sets is provided .
since the determination of source fluxes and positions is central to the accuracy of our ir / radio ratios and the band - merging between the observations at different wavelengths , we describe the derivation of these quantities in particular detail . following the completion of the vla - cosmos large project @xcite which had mapped the entire 2 deg@xmath2 cosmos field at 1.4ghz with the vla in antenna configurations a and c for 275 hours , the central 0.84 square degrees were re - imaged during an additional 62 hours in configuration a in the spring of 2006 .
the resulting vla - cosmos ` deep project ' mosaic has a resolution ( fwhm of synthesized beam ) of 2.5@xmath31 2.5@xmath32 and a pixel scale of 0.35@xmath32/px .
the mean @xmath206 sensitivity is @xmath290.01mjy / beam at the centre of the field and more than 60% ( 80% ) of the field has an @xmath206 level better than 0.02 ( 0.03)mjy ( see fig . 3 in schinnerer et al .
2009 , subm . ) .
the data reduction was carried out with aips ( astronomical image processing system ; * ? ? ?
* ) and followed standard procedures described in @xcite and ( 2009 , subm . ) .
the aips source / component finding task ` sad ` was used to detect radio sources in the deep project mosaic in multiple iterations with cut - off levels of successively lower @xmath44 .
the resulting source list was then combined with the vla - large project catalog @xcite to create a list of 1.4ghz sources ( henceforth referred to as the ` _ joint catalog _ ' ) that are at least of 5@xmath35 above the local background .
of the @xmath32900 sources in the joint catalog 51% are found to be unresolved . in these cases the integrated flux density is set equal to the peak flux density of the sources which was measured with the aips task ` maxfit ` . in order to correct for bandwidth
smearing a position - dependent correction based on a model sensitivity map of the deep project mosaic has been applied to the peak flux values . integrated flux densities ( which are not affected by bandwidth smearing ; e.g. * ? ?
* ) for resolved radio sources were determined by integrating over the size of the best - fitting elliptical gaussian component with the task ` jmfit ` .
for another 131 ( 4.6% ) of the sources in the catalog which were best fit by the sum of more than one gaussian flux component , the task ` tvstat ` was used to measure the flux within a manually defined contour around the source .
the distribution of 1.4ghz flux density values in the joint catalog is shown in fig .
15 of schinnerer et al .
( 2009 ) where a detailed description of the construction of the catalog is given .
the flux measurements carried out on the vla - cosmos radio maps have been compared to those of about 300 sources also observed at 1.4ghz in the context of the nvss and first surveys @xcite .
the agreement between the different data sets is reasonable except for a number of nvss sources where the vla - cosmos observations have probably resolved out a large extended flux component ( cf . * ? ? ?
+ for the multiple component sources which often have a complicated radio morphology due to outflows or continuum emission from star formation activity we set the most likely source position by visual inspection of hst acs images @xcite . in all other cases
the source position is located at the peak of the radio emission as determined by ` maxfit ` .
this definition was also adopted for a small fraction of multiple component sources that could not be reliably associated with an optical source .
the typical accuracy of the radio positions is 0.13@xmath32 as shown by @xcite . while the joint catalog serves as the basis of our radio - selected sample ( introduced in [ sect : radiosamp ] ) we will also use the _ mips _
24@xmath1 m catalog described in [ appsect:24umdata ] to construct a sample of infrared - selected sources .
in addition to counterparts with @xmath30 5 from the joint catalog we also allow for sources having @xmath43 3 in the presence of ir detections . to find additional sources in this @xmath44 range we ran ` maxfit ` at the position of ir detections without a counterpart in the joint catalog .
the ` maxfit ` box size is chosen in accordance with the uncertainty in the position of the ir source ( see appendix [ appsect : bandmerge ] ) .
since the vast majority of sources at these low detection levels are likely unresolved ( radio sources with flux density @xmath290.1mjy typically have sub - arcsecond sizes , see @xcite and references therein ) their integrated flux is equated to the peak flux density .
+ if the extracted peak flux density has a significance level of less than @xmath207 or if ` maxfit ` fails to converge , upper limits on the total flux from the local sky background are used as flux constraints for a potential source at that position .
the noise in the deep project image was estimated with the aips task ` rmsd ` in a box of dimensions 105@xmath31105@xmath32 that was moved across the survey area in steps of 2.45@xmath32 in r.a . and
dec .. we adopt an upper flux limit of 3 times the local @xmath206 noise level for the rest of the analysis for undetected radio sources .
the numeric value of this upper bound is obtained by reading the noise image at the pertinent position . during _ mips
_ cycle-2 and cycle-3 the s - cosmos project @xcite imaged the whole cosmos field at 24@xmath1 m in medium and slow scan mode , respectively .
the data from both _ mips _
cycles was coadded and combined with the mopex package ( ( mosaicker and point source extractor ; * ? ? ? * ) ) after elimination of transient sources such as asteroid .
the resulting mosaic has a pixel scale of 1.2@xmath32/px , while the fwhm of the _ mips _
24@xmath1 m psf is 5.8@xmath32 .
about 90% of the area was mapped with a median integration time of @xmath33400 seconds resulting in a @xmath33 sensitivity of @xmath30.018mjy . over the remaining @xmath310% of the field
the average integration time was roughly 7000 seconds , leading to an equivalent depth of @xmath208jy .
a noise map for the s - cosmos 24@xmath1 m observations was generated using the associated coverage map .
the details of the _ mips _
24@xmath1 m data reduction and source extraction procedures are spelled out in @xcite . in the following
we summarize the points that are most relevant to our work with respect to the 24@xmath1 m source catalog .
source detection in the 24@xmath1 m map was carried out with sextractor @xcite which returned positions that served as input to the psf - fitting algorithm daophot @xcite .
daophot performs simultaneous psf fitting to multiple sources which is an important advantage for the deep and crowded _ mips _
24@xmath1 m images where objects may be blended . by inserting and re - extracting artificial sources in exactly the same manner as true astronomical sources , @xcite found that the approach of psf fitting is reliable down to a flux density of @xmath360@xmath209jy which in most regions of the mosaic corresponds to a @xmath44 of about 3 . at this detection threshold
there are @xmath350,000 sources in the area covered by the 1.4ghz observations .
given the resolution of @xmath36@xmath32 of the _ mips _
images most of these are not resolved . as described in @xcite , the fidelity of the s - cosmos 24@xmath1 m flux calibration
was ascertained by checking that the @xmath210 $ ] colours for stars in the cosmos field listed in the catalog of the 2mass survey @xcite did not deviate from the expected relation .
note also that total flux measurements at 24@xmath1 m account for all the flux in the extended wings of the psf .
the astrometry of the _ mips _ data is usually accurate to within a few tenths of an arcsecond which was confirmed by the cross - correlation with the 2mass catalog , which revealed only a small systematic offset of 0.3@xmath32 in declination that was subsequently corrected in the catalog of 24@xmath1 m sources .
since the 24@xmath1 m catalog extends to a lower detection threshold of @xmath33@xmath35 we do not attempt to extract extra sources in the vicinity of radio detections .
if no 24@xmath1 m counterpart to a radio source is found during the band - merging ( see appendix [ appsect : bandmerge ] ) we directly assign an upper @xmath207 flux limit based on the value of the noise map at the according position .
the conversion from surface brightness noise ( as specified by the noise image ) and point source detection noise ( as required for a flux limit ) was derived by scaling the median value of all pixels with an exposure time of @xmath33400 seconds in the noise image to the average @xmath33 sensitivity level of @xmath30.018mjy .
the adopted value of the conversion factor is 2.3mjy(mjy / sr)@xmath28 .
mips 70@xmath1 m observations of the cosmos field were carried out in parallel with the 24@xmath1 m imaging using the scan mapping mode @xcite .
the fwhm of the data is 18.6@xmath32 and the pixel scale 4@xmath32/px . as in the case of the 24@xmath1 m survey
a limited area has a high coverage by repeated observations with a total exposure time of @xmath32800 seconds .
the median effective exposure time over the remaining 90% of the survey area is 1350 seconds leading to an average point source noise ( @xmath33 ) of 1.7 mjy .
the 70@xmath1 m observations were reduced with the ssc pipeline tools gert and mopex according to standard procedures for _ mips_-ge survey data .
special attention was given to the creation of an accurate noise image which represents the uncertainty owing to both small scale ( scatter in repeated observations of each sky pixel ) and large spatial scale noise properties ( i.e. pixel - to - pixel correlated noise and confusion noise ) .
all steps leading to the final data products are described in @xcite .
source detection and extraction was performed within the mopex package @xcite using the apex ( astronomical point source extraction ) peak algorithm , augmented by additional specialized scripts .
the noise image was employed to detect and then fit peaks with @xmath43 3 using the point source response function ( prf ) .
the final source list has been cleaned of spurious detections along the first airy ring of the bright 70@xmath1 m sources .
we checked all objects down to the flux level at which it would take a @xmath132.5@xmath35 noise fluctuation in order for a spurious source to be flux - boosted to @xmath211 3 in the first airy ring ( the amplitude of which is about 2% of the peak ) .
+ the final list of 70@xmath1 m sources contains almost 3000 entries in the region of the s - cosmos survey that overlaps with the vla - cosmos area .
flux measurements in the catalog have been corrected for the additional flux outside the prf image ( i.e. beyond the first airy ring ) using spitzer tiny tim models @xcite and placed on a constant @xmath212 scale .
this colour - correction is accurate within 2% for a wide range of galaxy seds across the filter bandpass .
the calibration of the 70@xmath1 m data was confirmed to be consistent with the official mips calibration @xcite which is accurate to 7% @xcite .
finally , a small positional offset in declination was found in the comparison of radio and 70@xmath1 m source positions .
however , at @xmath30.2@xmath32 it is significantly smaller than the scatter measured for the positions of individual sources which amounts to about 2@xmath32 . to provide upper @xmath207 flux limits at 70@xmath1 m whenever the band - merging with the radio or 24@xmath1 m fails to identify a 70@xmath1 m counterpart we convert the local surface brightness noise estimate given in the noise image to a point source noise .
after accounting for the flux in the extended wings of the psf the corresponding conversion factor is 14.9 mjy ( mjy / sr)@xmath28 .
the cosmos photometry catalog is an @xmath162-band selected catalog with psf - matched photometry ( fwhm = 0.6@xmath32 ) from 30 broad , medium and narrow band filters extracted at the positions of the subaru @xmath58 band detections .
the wavelength range covered by these observations extends all the way from the uv at 1550 to the mir at 8 @xmath1 m .
@xcite provide a complete description of the observations and data reduction leading to the compilation of the multi - wavelength data set .
here we use the photometry catalog compiled by @xcite which lists more than 600,000 cosmos galaxies with @xmath37 26 detected in a region roughly contiguous with the area covered by the vla - cosmos observations .
the cosmos field has been observed with _
xmm - newton _ for a total of @xmath31.5@xmath213seconds ( @xmath3400 hours ) , resulting in a homogeneous depth of @xmath35@xmath214sec @xcite over 1.92 deg@xmath2 .
the associated catalog includes 1887 point - like sources chosen to have a high probability of being a reliable detection5@xmath215 of being a spurious detection . ] in at least one of the soft ( 0.5 - 2kev ) , hard ( 2 - 10kev ) or ultra - hard ( 5 - 10kev ) bands down to limiting fluxes of @xmath35@xmath216 , @xmath33@xmath39 and @xmath35@xmath39 erg@xmath40s@xmath28 , respectively ( see @xcite for additional details ) .
moreover , the central part of the field has been observed with _
@xcite , providing precise positions for about half of the xmm sources .
brusa et al .
( in prep . ) have determined optical counterparts to more than 90% of the xmm catalog sources . in the present work we use only those with unique / secure optical counterparts and a successful fit to the uv to mir
sed @xcite .
the optical position has been used to associate our radio and ir sources to xmm detections using the same search radii as adopted for the band - merging of our radio and ir sources with the optical photometry ( see appendix [ appsect : bandmerge ] ) .
the catalog of @xcite contains photometric redshift estimates for all tabulated objects with the exception of ( i ) sources lying in a ` masked ' region of the optical imaging due to a bright star in their vicinity , or ( ii ) sources that are detected at x - ray wavelengths , in which case a special set of agn - templates was employed to derive a photometric redshift @xcite .
+ spectroscopic data has already been gathered for more than 20,000 sources in the cosmos field ( ( e.g . * ? ? ?
* ) ; @xcite ; @xcite ; @xcite ; kartaltepe et al . , in prep . ;
salvato et al . , in prep ) . whenever possible , we give precedence to redshift information from spectroscopy . in all other cases
we assign a photometric redshift to the sources in our radio- and ir - selected samples .
the details of the band - merging of the radio and ir catalogs with the optical data are presented in appendix [ appsect : bandmerge ] while the next two paragraphs summarize the properties and screening of the available redshift information itself . at the time of writing
the data base of reduced spectroscopic observations in the cosmos field included more than 14,000 sources .
most of these ( 75% ) were observed by vimos in the context of the zcosmos survey , observations with deimos and imacs(3 ) account for another @xmath312% each and a small fraction of sources has lris , sdss or fors1 spectroscopy . in some cases spectra of the same source have been obtained by more than one of these instruments .
if these measurements do not agree within experimental error we disregard the one with the lower quality flag .
the quality of the redshift measurement for our spectroscopically observed optical counterparts is judged by the spectroscopic confidence classes of the zcosmos survey . in particular
, we accepted only those objects with confidence flag 3 and 4 ( regardless of the decimal place , see below ) , as well as 2.5 , 2.4 , 8.5 , 9.3 , 9.5 and 1.5 .
the decimal place indicates the level of agreement between the spectroscopic and photometric redshift measurements .
our choice of acceptable flag values ensures that the reliability of all measured spectroscopic redshifts is better than 99% @xcite .
+ the classification with decimal places was not available for some of the sources observed by follow - up campaigns other than zcosmos . in such cases
we updated the spectroscopic confidence flag as required by comparison with the photo - z estimates of @xcite and @xcite .
for spectroscopic targets with unsatisfactory confidence flags ( @xmath2173% of the spectroscopic targets ) we checked whether a reliable photometric redshift estimate was available according to the selection criteria described in the next section and were thus able to recover redshift information for all but 4% of these . in table
[ tab : zstats ] we summarize the availability of redshift information for our sources .
sources successfully matched to an entry in the cosmos photometry catalog generally have a precise photometric redshift estimate derived with the code _
le phare _ @xcite .
the average number of bands available for the computation of photometric redshifts for the matched radio and 24@xmath1 m sources was 29 .
the accuracy of the photometric redshifts was calibrated with the help of more than 4000 high - confidence redshifts from zcosmos @xcite sources with @xmath218 22.5 and verified using faint 24@xmath1 m sources ( kartaltepe et al .
, in prep . )
in the range @xmath219 22.5 . at
@xmath9 1.25 @xcite found that the dispersion @xmath220 of photo - z measurements is 0.007 , 0.013 and 0.051 in the subaru @xmath221-band magnitude ranges of @xmath222 22.5 , 22.5 @xmath223 24 and @xmath224 24 , respectively . beyond redshift @xmath7 1.25
the statistical accuracy of the photometric redshifts abruptly decreases by a factor of @xmath33 with respect to the range @xmath9 1.25 because the balmer break is redshifted into the nir filters which have less sensitive photometry and non - contiguous wavelength coverage .
we use these values of the photo - z dispersion to remove sources in our sample with uncertain photometric redshifts ; the criterion which each source must satisfy to remain in the sample is that its photo - z error be smaller than @xmath225 of the dispersion at a given magnitude and redshift .
we also eliminate sources with a @xmath226 above a threshold corresponding to two standard deviations in an ideal @xmath81-distribution ( with the number of degrees of freedom equal to the difference between the number of filters used for deriving the redshift and the number of free parameters , namely three , i.e. , the redshift , the template type and the template normalization ) . in practice
very few sources ( @xmath2170.5% in the combined radio- and ir selected samples ) are rejected due to the latter criterion .
a somewhat larger number has been excluded due to a broad redshift probability distribution but they nevertheless represent only @xmath31% and @xmath32% of the radio- and ir - selected sample , respectively .
the vast majority of sources without spectroscopically constrained redshift have a photometric redshift estimate within the statistically expected accuracy ( 46.4% and 55.5% in the radio- and ir - selected samples , respectively ) .
+ the template library used by @xcite consists of star - forming and passive galaxy seds but includes none that reflect the features expected in spectra of sources with a dominant contribution from an agn .
@xcite have computed photometric redshifts for all @xmath227-detected sources in the cosmos photometry catalog using templates with varying agn contributions .
their redshift measurements account for source variability and achieve an accuracy of @xmath228 0.015 at @xmath229 24.5 for both type 2 and type 1 agn and qsos out to @xmath7 4.5 . as no confidence intervals were available for the best - fitting agn photo - z estimates and
since the statistical dispersion is far smaller than the width of the redshift bins studied below we have kept all sources with redshifts derived from the agn - template library in the sample .
* _ ir - selected sample _ the @xmath36@xmath32 fwhm of the 24@xmath1 m psf and _ spitzer _ pointing uncertainties result in a non - negligible uncertainty on the source centroids of 24@xmath1 m sources . in order to have a more precise reference for the positional matching with other catalogs we searched for irac counterparts to each 24@xmath1 m source ( the fwhm of the irac psf is about @xmath168 of that at 24@xmath1 m ) . since the irac imaging of the cosmos field is shallower than that performed at optical wavelengths but still sufficiently deep to detect a counterpart for most 24@xmath1 m sources this approach simultaneously reduces the likelihood of assigning the wrong optical counterpart in the subsequent band - merging with the optical photometry catalog ( see last paragraph of this section ) .
+ we correlated the 24@xmath1 m catalog with the irac catalog @xcite using a matching radius of 2@xmath32 and found 45,827 unambiguous counterparts ( 92% of all sources in the 24@xmath1 m catalog ) . in 572 cases
more than one irac source was found within the search radius . unless the closest of the potential counterparts was at least twice as close to the 24@xmath1 m position as the other candidates these sources ( 1% of the catalog )
were excluded from the sample . in the case of 3668 24@xmath1 m sources for which no irac counterpart could be determined we searched the photometry catalog for optical counterparts within 0.6@xmath32 , the fwhm of the ground - based imaging . if this match was ambiguous the source was disregarded .
unique optical counterparts were found for another 1386 sources , however , and the final percentage of 24@xmath1 m sources with reference coordinates thus increased to 94% . * _ radio - selected sample _ the 1.4ghz source positions are accurate to within about 0.1@xmath32 , which is sufficiently accurate that they can be used directly as the reference position for the identification of optical and ir counterparts . starting with the radio source position we identified 2233 objects ( 77% of the radio catalog ) that have a single 24@xmath1 m counterpart within 2@xmath32 while 647 ( 22% ) radio sources remain unmatched . based on their radio properties
the 24@xmath1 m non - detections are primarily * tbd * . for @xmath2171% of the radio - selected sample
the match was ambiguous .
this subset includes two pairs of radio sources which have been assigned the same 24@xmath1 m counterpart and 17 for which more than one ir - detection was present within the search radius ( mostly due to a complicated 24@xmath1m - morphology or the occasional over - deblended source ) . since the identification of 24@xmath1 m counterparts to 1.4ghz sources in the immediately preceding paragraph
was based on the precise radio positions these pairs of objects can be taken over directly into the ir - selected samplem source has no trustworthy irac or optical counterpart , the 24@xmath1m - radio association in question is not included in the sample .
this affects only a very small fraction of the ir - selected sample since the fraction of sources for which no reference coordinate could be determined is @xmath2171% . ] .
note that subsequently we search for additional radio detections with @xmath230 3 in the vicinity of all 24@xmath1 m sources without a radio counterpart in the radio catalog described in
[ appsect : radiodata ] .
the box in which we check for these additional faint radio sources has dimensions [email protected]@xmath32 ( i.e. 5@xmath175 px@xmath2 at the pixel scale of the deep project image ) which corresponds to the fwhm of the psf of the first two irac bands .
radio fluxes for an additional 639 24@xmath1 m sources could be measured in this way .
we allow for a maximal separation of 6@xmath32 between reference coordinates and an accepted 70@xmath1 m source . due to the broad fwhm ( 18.6@xmath32 ) of the _ mips _
70@xmath1 m psf there is a relatively large probability that the emission from more than one 24@xmath1 m or 1.4ghz source is blended into a single 70@xmath1 m source with no distinguishable secondary peaks by which the individual components could be separated .
this occurred for @xmath37% of the 1.4ghz sources with a potential counterpart at 70@xmath1 m . in the ir - selected sample 21% of the matched sources had a potential 70@xmath1 m counterpart that lay within less than 6@xmath32 of at least one more 24@xmath1 m source .
unless these ambiguous 70@xmath1 m associations are at least twice as close to the nearest 24@xmath1 m or radio source as to the next best candidate they have been excluded for the analysis of the paper . ultimately , this was the fate of 70% ( 50% ) of the radio ( 24@xmath1 m ) sources with blended 70@xmath1 m photometry . in the interest of achieving a high accuracy of the measured ir / radio properties we prefer this approach to , e.g. , attempting to fit the 70@xmath1 m flux distribution with two components with peak positions fixed to the reference position .
+ to summarize , in we have found @xmath31500 reliable 70@xmath1 m counterparts to 24@xmath1 m sources .
a further 60 sources in the ir - selected catalog had more than one counterpart within 6@xmath32 and were removed together with the 24@xmath1 m objects that are associated with a blended 70@xmath1 m detection .
the majority of the 24@xmath1 m sources ( 67% ) has no directly detected counterpart . in the radio - selected sample
we were able to identify @xmath3820 unambiguous 70@xmath1 m counterparts while excluding @xmath360 sources because they either were associated with a blended 70@xmath1 m source or had an ambiguous match .
for some 70% of the radio sources no 70@xmath1 m counterpart could be found .
a search radius of 0.6@xmath32 is used for the cross - correlation of the radio - selected sample with the cosmos photometry catalog and the spectroscopy catalog .
this figure corresponds to the fwhm of the psf - matched ground - based photometry . whenever more than one optical counterpart is found within this distance the according radio source
is removed from the catalog .
this choice was made because any photometric redshift derived for the source would necessarily be unreliable as both sources contribute to the measured flux .
the same applies to spectroscopic observations which have a slit width of @xmath31@xmath32 and are thus treated analogously because sources within 0.5@xmath32 could potentially lead to ambiguous spectral features .
+ to find optical counterparts and spectroscopic information for sources in the ir - selected catalog we use a search radius of 1@xmath32 which accounts for the somewhat larger uncertainty in the predominantly irac - based reference positions .
also , we ease the rejection criterion for ambiguous matches somewhat in that we accept optical or spectroscopic counterparts which are more than twice as close as the other candidate(s ) , provided the next best candidate lies beyond 0.6@xmath32 .
to derive the underlying distribution @xmath164 of ir / radio ratios @xmath84 for a sample in which measurements of @xmath84 can be limited both from above and below due to non - detections either in the ir or radio band we use the method of @xcite .
their approach is based on maximum likelihood estimation and requires no assumptions about the form of the true distribution of @xmath84 ( it is thus said to be non - parametric ) .
if represented in the form of a cumulative distribution , @xmath231 , the estimator is equivalent to the probability of measuring a value of @xmath84 smaller than a given ratio ( which we will denote by a capital q ) : @xmath232 in the following we summarize the steps involved in the construction of @xmath231 . to begin with ,
the @xmath105 data points @xmath112 are arranged in @xmath233 bins with ascending bin centres @xmath234 , @xmath235 1 , ... , @xmath233 . let @xmath236 be the number of detections in the @xmath237-th bin which spans the range @xmath238 . for the @xmath237-th bin
we also define @xmath239 , the number of upper limits smaller than @xmath240 and , in analogy , @xmath241 as the number of lower limits exceeding @xmath240 .
note that in view of the definition of @xmath239 and @xmath241 it is useful to choose bin boundaries offset by @xmath242 and @xmath243 for the upper and lower limits , respectively ( cf .
this eases the computation of the vectors @xmath244 , @xmath245 and @xmath246 which can be easily constructed as slightly shifted histograms of the detections as well as of the upper and lower limits .
and @xmath245 is necessary if the first uncensored value , @xmath247 , occurs at a higher value of @xmath240 than the smallest upper limit in the data set . by construction
, the algorithm of @xcite is not sensitive to upper limits that lie below the smallest uncensored data point . as a consequence the normalized cumulative distribution function defined in equation ( [ eq : schmittcumul ] ) rises above 0 for the first time in the bin containing the smallest direct detection .
the correct asymptotic value of the distribution function for @xmath248 , however , should not be 0 in this case but a number larger than or equal to @xmath249 . here
@xmath250 is the number of upper limits smaller than @xmath247 and @xmath105 is defined immediately following equation ( [ eq : schmittcoupled ] ) .
the correct behaviour of the cumulative distribution function can be achieved by setting @xmath251 and @xmath252 0 for all bins @xmath237 with bin centres @xmath253 .
+ note that even when there are lower limits that exceed the largest uncensored value , no analogous fixes are needed to correct the doubly censored distribution function in the limit @xmath254 . ] the value @xmath255 of the distribution function @xmath164 in the @xmath237-th bin is constrained by the probability @xmath256 of obtaining the triplet ( @xmath236 , @xmath239 , @xmath241 ) .
intuitively equation ( [ eq : schmittprob ] ) can be understood as the product of the _ priori _ probabilities of a measurement in the @xmath237-th bin being a detection ( first term ) , an upper limit which if detectable would in truth be located in a lower bin ( second term ) or an upper limit in actually located in a higher bin ( third term ) . @xcite and
@xcite have shown that by introducing the likelihood and by using the constraint that @xmath257 1 the probability in equation ( [ eq : schmittprob ] ) can be used to derive a system of @xmath233 coupled fixed point equations for each of the @xmath255 : @xmath258 here @xmath259 is the total number of measurements in the sample
. equations ( [ eq : schmittcoupled ] ) can be solved by iteration starting with @xmath260 and @xmath261 0 and the cumulative distribution constructed according to @xmath262 once convergence has been achieved .
this was typically the case after approximately 30 - 70 iterations .
various approaches to calculate the uncertainty associated with the doubly censored cumulative distribution function @xmath231 have been proposed ( see , e.g. * ? ? ?
here we use the analytical formalism of @xcite which is based on the fisher information matrix @xmath263 with elements 0 ( and hence @xmath264 1 ) the error on @xmath265 is zero by definition . in this case is useful for computational reasons to reduce the dimensionality of the problem by one by setting @xmath266 and thereafter only considering the ( @xmath267)-dimensional vectors @xmath244 , @xmath245 , @xmath246 and @xmath268 . ]
@xmath269 j_{mm } & = \frac{d_m}{(f_m - f_{m-1})^2 } + \frac{l_m}{(1-f_m)^2}-\frac{u_m}{f_m^2}~,\nonumber\\*[2ex ] j_{ij } & = 0~ , \qquad & { \rm for}\quad \left|i - j\right| > 1 .
\nonumber \label{eq : covarelem}\end{aligned}\ ] ] @xmath263 is a symmetric jacobi matrix in which only the diagonal and off - diagonal elements are non - zero . in the limit of large samples its inverse , @xmath270 ,
is an asymptotically unbiased estimate of the covariance matrix of the @xmath271 ( @xmath272 ) such that one has @xmath273 following @xcite and references therein it is then possible to define 2@xmath35 confidence intervals of the form @xmath274 which always lie in the permissible range [ 0 , 1 ] . + by applying the algorithm of equation ( [ eq : schmittcoupled ] ) to singly censored data sets it was possible to check whether its output agrees with that obtained using the formalism of appendix [ appsect : singlecense ] .
we found that the variance in equation ( [ eq : doubcensevar ] ) needed to be doubled in order to achieve consistency of the confidence intervals in equation ( [ eq : doubcenseerr ] ) with those computed with the algorithm specifically designed to deal with singly censored data .
for the sake of consistency the uncertainties on the doubly censored distribution functions stated throughout the paper have thus been computed using twice the nominal value of equation ( [ eq : doubcensevar ] ) .
it can be useful to convey information on the shape of a ( cumulative ) distribution in form of percentiles . given @xmath271 and the associated confidence intervals , the @xmath275-th percentile is defined as that @xmath84 for which @xmath276 .
similarly , the points where the curves defined by the upper and lower confidence intervals given by equation ( [ eq : doubcenseerr ] ) of @xmath231 are equal to @xmath277 are used as a measure of the uncertainty on the sought percentile . a formalism which is identical to or simplified with respect to that of the previous section
can be used when dealing with single censoring .
if , for instance , we consider a radio - selected sample its distribution of @xmath84 is censored from the left and can be represented as @xmath278 \end{cases } \nonumber \label{eq : singcensodef}\end{aligned}\ ] ] in analogy to equations ( [ eq : doubcensodef ] ) .
the treatment of singly censored data has been discussed in detail in @xcite and @xcite with an emphasis on astronomical applications .
the former paper also describes how to transform left censored into right censored data , the treatment of which has been documented more frequently in the literature for historical reasons ( see references in the two aforementioned articles ) . here
we simply would like to point out that equations ( [ eq : schmittcoupled ] ) , which constrain the distribution function of doubly censored data , remain valid in the case of single censoring .
however , the fact that either all of the @xmath279 or @xmath280 as defined in [ appsect : doubcensetheory ] equal zero allows the coupled system of equations to be solved analytically in terms of the cumulative distribution function @xmath271 through the kaplan - meier product limit estimator @xcite . to do
so we follow exactly the prescriptions of @xcite .
the relative frequency of agn at a given value of @xmath115}$ ] ( and in any of the considered redshift slices ) is computed as @xmath281 where the summation is carried out over all objects @xmath162 in the given redshift bin that lie in the pertinent range of ir / radio ratios @xmath115}$ ] .
furthermore , we have by definition that pr(agn ) = [ 1-pr(sf ) ] , as illustrated in fig .
[ fig : uk_sfprob ] .
note that if the probabilities of sf- and agn-`hood ' are discretized ( e.g. drawn from a set of integer probability values in per cent ) the numerator and denominator in equation ( [ eq : agnfrac ] ) could be rewritten as a summation over probabilities . in this case
the denominator , for instance , takes the form : @xmath282 here @xmath283 stands for the number of objects with a probability of @xmath162 per cent of being a sfg .
if a data set with only uncensored measurements is being studied the computation of @xmath156 is straightforward since the numbers @xmath283 can be directly determined by counting the suitable objects in each bin of @xmath84 and @xmath11 . in the presence of censored data ,
however , @xmath283 should be regarded as an _ effective _ number of objects which accounts for the fact that ( _ i _ ) the true value of @xmath84 associated with a limit found in a specific bin , @xmath237 , might lie outside that bin ( if its actual value were determined with the help of better data ) , or , ( _ ii _ ) limits detected in neighbouring bins have a finite probability of ending up in the @xmath237-th bin ( e.g. lower limits populating a bin with rank smaller than @xmath237 or upper limits in a bin centred on a larger value of @xmath84 than that of the @xmath237-th bin ) .
+ to compute this effective number of objects in the @xmath237-th bin , @xmath284 , we derived the distribution function of objects with the desired probability of being sfgs following the techniques presented in appendix [ appsect : doubcensetheory ] .
due to the finite number of objects in each redshift bin it was necessary to work with a limited number of probability bins to ensure a well - behaved estimate of the distribution function @xmath285 . since @xmath285 specifies the fractional contribution of a bin in @xmath84 to the total number of measurements we have that @xmath286 where @xmath287 is the total number of objects used to compute the distribution function
. with these modifications the new expression for @xmath156 in the @xmath237-th bin ( to be compared with equation ( [ eq : agnfrac ] ) ) becomes : @xmath288 note that because of the coarse binning in probability space @xmath162 no longer defines a number in per cent in this case but a range of probabilities instead .
consequently , we also have introduced a new weighting factor , namely the median value of the probabilities in the grouped data , @xmath289 .
it replaces the probability of the individual objects , @xmath290 , used in equation ( [ eq : altsum ] ) . in practice
the median usually does not differ much from the mean , i.e. it might take on a value of @xmath30.16 for objects with probabilities in the range @xmath291 .
we gratefully acknowledges the contribution of the whole cosmos collaboration and its more than 100 scientists around the globe . m.t.s
thanks ute lisenfeld , chris carilli and george helou for useful discussions , as well as marco scodeggio for comments on the manuscript .
+ the german dfg supported this research with grant schi 536/3 - 2 .
gz acknowledges partial support from an inaf contract ( prin-2007/1.06.10.08 ) and an asi grant ( asi / cofis / wp3110 i/026/07/0 ) .
this work is partly based on observations made with the _
spitzer space telescope _ , which is operated by nasa / jpl / caltech .
the national radio astronomy observatory is a facility of the national science foundation operated under cooperative agreement by associated universities , inc .
+ _ facilities : _ nrao ( vla ) , spitzer ( irac , mips ) , eso ( vlt ) , subaru ( suprimecam ) appleton , p. n. , et al .
2004 , , 154 , 147 avni , y. , soltan , a. , tananbaum , h. , & zamorani , g. 1980 , , 238 , 800 baldwin , j. a. , phillips , m. m. , & terlevich , r. 1981 , , 93 , 5 bally , j. , & thronson , h. a. , jr .
1989 , , 97 , 69 barthel , p. d. 2006 , , 458 , 107 beck , r. , & golla , g. 1988 , , 191 , l9 beck , r. 2005 , in : the magnetized plasma in galaxy evolution , 193 beelen , a. , cox , p. , benford , d. j. , dowell , c. d. , kovcs , a. , bertoldi , f. , omont , a. , & carilli , c. l. 2006 , , 642 , 694 bell , e. f. 2003 , , 586 , 794 bertin , e. , & arnouts , s. 1996 , a&as , 117 , 393 beswick , r. j. , muxlow , t. w. b. , thrall , h. , richards , a. m. s. , & garrington , s. t. 2008 , , 385 , 1143 bettens , r. p. a. , brown , r. d. , cragg , d. m. , dickinson , c. j. , & godfrey , p. d. 1993 , , 263 , 93 bicay , m. d. , & helou , g. 1990 , , 362 , 59 bondi , m. , ciliegi , p. , schinnerer , e. , smoli , v. , jahnke , k. , carilli , c. , & zamorani , g. 2008 , , 681 , 1129 boyle , b. j. , cornwell , t. j. , middelberg , e. , norris , r. p. , appleton , p. n. , & smail , i. 2007 , , 376 , 1182 bressan , a. , silva , l. , & granato , g. l. 2002 , , 392 , 377 campbell , g. 1981 , biometrika , 68 , 417 capak , p. , et al .
2007 , , 172 , 99 capak , p. , et al .
2008 , vizier online data catalog , 2284 , 0 cappelluti , n. , et al .
2007 , , 172 , 341 cappelluti , n. , et al .
2009 , , 497 , 635 carilli , c. l. , & yun , m. s. 1999 , , 513 , l13 carilli , c. l. , et al .
2008 , , 689 , 883 chary , r. , & elbaz , d. 2001 , , 556 , 562 chary , r. , et al .
2004 , , 154 , 80 colina , l. , & perez - olea , d. e. 1995 , , 277 , 845 condon , j. j. , condon , m. a. , gisler , g. , & puschell , j. j. 1982 , , 252 , 102 condon , j. j. 1984 , , 287 , 461 condon , j. j. 1992 , , 30 , 575 condon , j. j. 1997 , , 109 , 166 condon , j. j. , cotton , w. d. , greisen , e. w. , yin , q. f. , perley , r. a. , taylor , g. b. , & broderick , j. j. 1998 , , 115 , 1693 condon , j. j. , cotton , w. d. , yin , q. f. , shupe , d. l. , storrie - lombardi , l. j. , helou , g. , soifer , b. t. , & werner , m. w. 2003 , , 125 , 2411 dale , d. a. , & helou , g. 2002 , , 576 , 159 de jong , t. , klein , u. , wielebinski , r. , & wunderlich , e. 1985 , , 147 , l6 dickey , j. m. , & salpeter , e. e. 1984 , , 284 , 461 domingue , d. l. , sulentic , j. w. , & durbala , a. 2005 , , 129 , 2579 donley , j. l. , rieke , g. h. , rigby , j. r. , & prez - gonzlez , p. g. 2005 , , 634 , 169 dunkley , j. , et al . 2009 , , 180 , 306 dunne , l. , clements , d. l. , & eales , s. a. 2000 , , 319 , 813 dunne , l. , et al . 2009 , , 394 , 3 elbaz , d. , cesarsky , c. j. , chanial , p. , aussel , h. , franceschini , a. , fadda , d. , & chary , r. r. 2002 , , 384 , 848 elvis , m. , et al .
2009 , , 184 , 158 engelbracht , c. w. , et al .
2007 , , 119 , 994 feigelson , e. d. , & nelson , p. i. 1985 , , 293 , 192 feldmann , r. , et al .
2006 , , 372 , 565 fomalont , e. b. , kellermann , k. i. , cowie , l. l. , capak , p. , barger , a. j. , partridge , r. b. , windhorst , r. a. , & richards , e. a. 2006 , , 167 , 103 francis , p. j. 1993 , , 407 , 519 frayer , d. t. , et al .
2006 , , 131 , 250 frayer , d. t. , et al .
2009 , arxiv:0902.3273 garn , t. , green , d. a. , riley , j. m. , & alexander , p. 2009
, arxiv:0905.1218 garn , t. , & alexander , p. 2009
, , 394 , 105 garrett , m. a. 2002 , , 384 , l19 gordon , k. d. , et al . 2007 , , 119 , 1019 glazebrook , k. & economou , f. 1997 , the perl journal , 5 , 5 gonzalez , a. h. , clowe , d. , brada , m. , zaritsky , d. , jones , c. , & markevitch , m. 2009 , , 691 , 525 greisen , e. w. 2003 , astrophysics and space science library , 285 , 109 groves , b. a. , cho , j. , dopita , m. , & lazarian , a. 2003 , publications of the astronomical society of australia , 20 , 252 haarsma , d. b. , & partridge , r. b. 1998 , , 503 , l5 haarsma , d. b. , partridge , r. b. , windhorst , r. a. , & richards , e. a. 2000 , , 544 , 641 harwit , m. , & pacini , f. 1975 , , 200 , l127 hasinger , g. , et al . 2007 , , 172 , 29 helou , g. , soifer , b. t. , & rowan - robinson , m. 1985 , , 298 , l7 helou , g. 1991 , the interpretation of modern synthesis observations of spiral galaxies , 18 , 125 helou , g. , & bicay , m. d. 1993 , , 415 , 93 hummel , e. , davies , r. d. , pedlar , a. , wolstencroft , r. d. , & van der hulst , j. m. 1988 , , 199 , 91 ibar , e. , et al .
2008 , , 386 , 953 ilbert , o. , et al .
2009 , , 690 , 1236 ilbert , o. , et al .
2009 , arxiv:0903.0102 isobe , t. , feigelson , e. d. , akritas , m. g. , & babu , g. j. 1990 , , 364 , 104 jarrett , t. h. , chester , t. , cutri , r. , schneider , s. , skrutskie , m. , & huchra , j. p. 2000 , , 119
, 2498 kaplan , e. l. & meier , p. 1958
, j. am . statist .
assoc . , 53 , 457 kauffmann , g. , et al .
2003 , , 346 , 1055 kellermann , k. i. 1964 , , 140 , 969 kennicutt , r. c. , jr .
1998 , , 36 , 189 kewley , l. j. , groves , b. , kauffmann , g. , & heckman , t. 2006 , , 372 , 961 koekemoer , a. m. , et al .
2007 , , 172 , 196 kovcs , a. , chapman , s. c. , dowell , c. d. , blain , a. w. , ivison , r. j. , smail , i. , & phillips , t. g. 2006 , , 650 , 592 krist , j. 2002 , tiny tim/_sirtf _ user s guide ( pasadena : ssc ) lagache , g. , dole , h. , & puget , j .- l .
2003 , , 338 , 555 lefloch , e. , et al .
2009 , , 703 , 222 lauer , t. r. , tremaine , s. , richstone , d. , & faber , s. m. 2007 , , 670 , 249 lilly , s. j. , et al .
2007 , , 172 , 70 lilly , s. j. , et al .
2009 , , 184 , 218 lisenfeld , u. , vlk , h. j. , & xu , c. 1996 , , 306 , 677 magnelli , b. , elbaz , d. , chary , r. r. , dickinson , m. , le borgne , d. , frayer , d. t. , & willmer , c. n. a. 2009 , , 496 , 57 makovoz , d. , & khan , i. 2005 , astronomical data analysis software and systems xiv , 347 , 81 makovoz , d. , & marleau , f. r. 2005 , , 117 , 1113 marx , m. , kruegel , e. , klein , u. , & wielebinski , r. 1994 , , 281 , 718 mauch , t. , & sadler , e. m. 2007 , , 375 , 931 miller , n. a. , & owen , f. n. 2001 , , 121 , 1903 murgia , m. , helfer , t. t. , ekers , r. , blitz , l. , moscadelli , l. , wong , t. , & paladino , r. 2005 , , 437 , 389 murphy , e. j. , helou , g. , kenney , j. d. p. , armus , l. , & braun , r. 2008 , , 678 , 828 murphy , e. j. , kenney , j. d. p. , helou , g. , chung , a. , & howell , j. h. 2009 , , 694 , 1435 murphy , e. j. , chary , r .-
alexander , d. m. , dickinson , m. , magnelli , b. , morrison , g. , pope , a. , & teplitz , h. i. 2009 , , 698 , 1380 niklas , s. , & beck , r. 1997 , , 320 , 54 obri , m. , et al .
2006 , , 370 , 1677 odell , a. p. , schombert , j. , & rakos , k. 2002 , , 124 , 3061 oke , j. b. 1974 , , 27 , 21 papovich , c. , et al .
2004 , , 154 , 70 park , s. q. , et al .
2008 , , 678 , 744 pierini , d. , popescu , c. c. , tuffs , r. j. , vlk , h. j. 2003 , , 409 , 907 pohl , m. 1994 , , 287 , 453 prentice , r. l. , & marek , p. 1979 , biometrics , 35 , 861 price , r. , & duric , n. 1992 , , 401 , 81 reddy , n. a. , & yun , m. s. 2004 , , 600 , 695 richards , e. a. 2000 , , 533 , 611 richards , g. t. , et al .
2006 , , 131 , 2766 rieke , g. h. , alonso - herrero , a. , weiner , b. j. , prez - gonzlez , p. g. , blaylock , m. , donley , j. l. , & marcillac , d. 2009 , , 692 , 556 rigby , j. r. , et al .
2008 , , 675 , 262 roy , a. l. , norris , r. p. , kesteven , m. j. , troup , e. r. , & reynolds , j. e. 1998 , , 301 , 1019 sajina , a. , et al .
2008 , , 683 , 659 salvato , m. , et al .
2009 , , 690 , 1250 sanders , d. b. , phinney , e. s. , neugebauer , g. , soifer , b. t. , & matthews , k. 1989 , , 347 , 29 sanders , d. b. , & mirabel , i. f. 1996 , , 34 , 749 sanders , d. b. , et al .
2007 , , 172 , 86 schinnerer , e. , et al .
2004 , , 128 , 1974 schinnerer , e. , et al .
2007 , , 172 , 46 schmitt , j. h. m. m. 1985 , , 293 , 178 schmitt , j. h. m. m. , kahabka , p. , stauffer , j. , & piters , a. j. m. 1993 , , 277 , 114 scoville , n. , et al .
2007 , , 172 , 1 seymour , n. , et al .
2008 , , 386 , 1695 seymour , n. , huynh , m. , dwelly , t. , symeonidis , m. , hopkins , a. , mchardy , i. m. , page , m. j. , & rieke , g. 2009 , , 1077 siana , b. , teplitz , h. i. , chary , r .-
r . , colbert , j. , & frayer , d. t. 2008 , , 689 , 59 silverman , j. d. , et al .
2009 , , 696 , 396 smoli , v. , et al .
2006 , , 371 , 121 smoli , v. , et al .
2008 , , 177 , 14 smoli , v. , et al .
2009 , , 690 , 610 sopp , h. m. , & alexander , p. 1991
, , 251 , 14p stetson , p. b. 1987 , , 99 , 191 symeonidis , m. , willner , s. p. , rigopoulou , d. , huang , j .- s . , fazio , g. g. , & jarvis , m. j. 2008 , , 385 , 1015 taniguchi , y. , et al .
2007 , , 172 , 9 thompson , t. a. , quataert , e. , waxman , e. , murray , n. , & martin , c. l. 2006 , , 645 , 186 thompson , t. a. , quataert , e. , & waxman , e. 2007 , , 654 , 219 trump , j. r. , et al .
2007 , , 172 , 383 trump , j. r. , et al .
2009 , , 696 , 1195 turnbull , b. w. 1974 , j. am . statist .
, 69 , 169 van der kruit , p. c. 1973 , , 29 , 263 vlahakis , c. , eales , s. , & dunne , l. 2007 , , 379 , 1042 vlk , h. j. 1989 , , 218 , 67 white , r. l. , becker , r. h. , helfand , d. j. , & gregg , m. d. 1997 , , 475 , 479 wolf , c. , wisotzki , l. , borch , a. , dye , s. , kleinheinrich , m. , & meisenheimer , k. 2003 , , 408 , 499 wrobel , j. m. , & heeschen , d. s. 1988 , , 335 , 677 wunderlich , e. , wielebinski , r. , & klein , u. 1987 , , 69 , 487 york , d. g. , et al .
2000 , , 120 , 1579 younger , j. d. , et al .
2009 , , 394 , 1685 yun , m. s. , reddy , n. a. , & condon , j. j. 2001 , , 554 , 80 zhu , c. , & sun , j. 2007 , in : advances in statistical methods for the health sciences , 225 lccl & & & + [ 1ex ] + all & 2901 & 100 & initial 1.4ghz sample + no redshift & 748 & 25.8 & includes 21 1.4ghz sources outside area covered by multi-@xmath149 photometry + spec - z & 769 & 26.5 & + photo - z & 1232 + 114 & 46.4 & regular + agn photo - zs + unreliable redshift & 38 & 1.3 & + & & & + useable & 2020 & 69.6 & no ambiguous counterparts , reliable redshift & agn / sf classification available + & & & + pr(sf ) @xmath41 0.5 & 766 & 26.4 & optically selected sfgs with @xmath292 2.36 ( 38% of ` useable ' sample ) + pr(sf ) @xmath217 0.5 & 1254 & 43.2 & optically selected agn with @xmath293 2.36 ( 62% of ` useable ' sample ) + [ 1ex ] + * ir - selected sample * & & & + [ 1ex ] + all & 5301 & 100 & initial 24@xmath1 m sample restricted to @xmath294(24@xmath1m)@xmath41 0.3mjy + no redshift & 1003 & 18.9 & includes 304 24@xmath1 m sources outside area covered by multi-@xmath149 photometry + spec - z & 1254 & 23.7 & + photo - z & 2712 + 231 & 55.5 & regular + agn photo - zs + unreliable redshift & 101 & 1.9 & + & & & + useable & 3259 & 61.5 & no ambiguous counterparts , reliable redshift & agn / sf classification available + & & & + pr(sf ) @xmath41 0.5 & 1822 & 34.4 & optically selected sfgs with @xmath292 2.36 ( 56% of ` useable ' sample ) + pr(sf ) @xmath217 0.5 & 1437 & 27.1 & optically selected agn with @xmath293 2.36 ( 44% of ` useable ' sample ) + [ 1ex ] + & & + [ 1ex ] + all & 6863 & 100 & 1560/3960 sources selected only at 1.4ghz/24@xmath1 m ; 1341 in both catalogs + no redshift & 1620 & 23.6 & includes 321 sources outside area covered by multi-@xmath149 photometry + spec - z & 1486 & 21.7 & + photo - z & 3362 + 275 & 53.0 & regular + agn photo - zs + unreliable redshift & 120 & 1.7 & + & & & + useable & 4454 & 64.9 & no ambiguous counterparts , reliable redshift & agn / sf classification available + & & & + pr(sf ) @xmath41 0.5 & 2215 & 32.3 & optically selected sfgs with @xmath292 2.36 ( 49.7% of ` useable ' sample ) + pr(sf ) @xmath217 0.5 & 2239 & 32.6 & optically selected agn with @xmath293 2.36 ( 50.3% of ` useable ' sample ) + [ 1ex ] lcrccc hdf - n & & @xcite & ir ; @xmath47 0.08mjy & @xmath295 0.004mjy / bm & @xmath4 = 0.52 - 0.7 ( @xmath296 1mjy ) + [ 1 ex ] swire & & @xcite & ir ; @xmath47 0.1mjy & @xmath295 0.03mjy / bm & @xmath297 1.39 + [ 1 ex ] xfls / swire & & @xcite & ir ; @xmath47 0.15mjy & & @xmath298 0.92 - 1.02 ( @xmath296 1mjy ) + [ 1 ex ] & & & & & @xmath299 + & & & & & ( log@xmath300 11 ) , + & & & & & @xmath301 + & & & & & ( log@xmath302 11 ) + [ 1 ex ] subaru_xmm - new- _ & & & & & + _ ton _ deep field + [ 1 ex ] & & & & @xmath47 0.5mjy , & @xmath4 = 0.94 - [email protected] , + & & & & @xmath303 30 mjy & @xmath5 = [email protected] + [ 1ex ] xfls & @xmath9 1 & @xcite & radio : @xmath304 0.09mjy & @xmath303 15mjy & @xmath305 ( @xmath306 0.2 ) + [ 1 ex ] 13@xmath307_xmm - newton_/ & & & & & + _ chandra _ deep field + [ 1 ex ] _ iras _ & @xmath10 0.16 & @xcite & radio - matched w/ @xmath308 2 jy & & @xmath96 = [email protected] + [ 1ex ] hdf - n & @xmath10 1.4 & @xcite & radio ; 5@xmath35 wsrt srcs . &
& @xmath309 2 + [ 1 ex ] var .
survey fields & @xmath8 [ 1,3 ] & @xcite & sub - mm / radio & & @xmath310 [email protected] + [ 1 ex ] goods - n & @xmath8 [ 0.6,2.6 ] & @xcite & & @xmath47 0.2mjy & @xmath311 [email protected] + [ 1 ex ] xfls & @xmath8 [ 0.5,3 ] & @xcite & ir ; @xmath47 0.9mjy & @xmath304 0.09mjy & @xmath310 2.07/2.21 + [ 1 ex ] egs / fidel & @xmath7 2 & @xcite & ir ; @xmath47 0.5mjy & & @xmath310 [email protected] + [ 1 ex ] & & & & & @xmath96 = [email protected] , + & & & & & @xmath6 = [email protected] + [ 1 ex ] lc|c|cc & * radio - selected sample * & * ir - selected sample * & + + [ -2ex ] + [ -2ex ] 0.08 @xmath312 0.23 & 1.02@xmath313 & 1.35@xmath314 & 1.31@xmath315 & [email protected] + [ 1ex ] 0.23 @xmath312 0.33 & 0.88@xmath316 & 1.28@xmath317 & 1.26@xmath318 & [email protected] + [ 1ex ] 0.33 @xmath312 0.45 & 0.94@xmath319 & 1.29@xmath320 & 1.27@xmath321 & [email protected] + [ 1ex ] 0.45 @xmath312 0.67 & 0.97@xmath322 & 1.26@xmath323 & 1.26@xmath186 & [email protected] + [ 1ex ] 0.67 @xmath312 0.82 & 0.83@xmath324 & 1.29@xmath325 & 1.23@xmath326 & [email protected] + [ 1ex ] 0.82 @xmath312 0.94 & 0.78@xmath327 & 1.32@xmath328 & 1.31@xmath329 & [email protected] + [ 1ex ] 0.94 @xmath312 1.14 & 0.77@xmath330 & 1.36@xmath331 & 1.29@xmath332 & [email protected] + [ 1ex ] 1.14 @xmath312 1.51 & 0.75@xmath333 & 1.47@xmath334 & 1.16@xmath335 & [email protected] + [ 1ex ] 1.51 @xmath312 2.00 & 0.96@xmath336 & 1.63@xmath337 & 1.35@xmath338 & [email protected] + [ 1ex ] 2.00 @xmath312 4.50 & 1.06@xmath339 & 1.93@xmath340 & 1.62@xmath341 & [email protected] + [ 1ex ] lccc @xmath128 & [email protected] & [email protected] & [email protected] + @xmath129 & [email protected] & [email protected] & [email protected] + @xmath6 & [email protected] & [email protected] & [email protected] + lc|c|cc & * radio - selected sample * & * ir - selected sample * & + + [ -2ex ] + [ -2ex ] 0.08 @xmath312 0.23 & 2.07@xmath342 & 2.41@xmath343 & 2.39@xmath344 & [email protected] + [ 1ex ] 0.23 @xmath312 0.33 & 1.91@xmath316 & 2.27@xmath345 & 2.26@xmath346 & [email protected] + [ 1ex ] 0.33 @xmath312 0.45 & 1.99@xmath347 & 2.30@xmath348 & 2.28@xmath318 & [email protected] + [ 1ex ] 0.45 @xmath312 0.67 & 1.94@xmath349 & 2.23@xmath350 & 2.24@xmath351 & [email protected] + [ 1ex ] 0.67 @xmath312 0.82 & 1.77@xmath352 & 2.23@xmath320 & 2.20@xmath353 & [email protected] + [ 1ex ] 0.82 @xmath312 0.94 & 1.79@xmath354 & 2.29@xmath355 & 2.22@xmath356 & [email protected] + [ 1ex ] 0.94 @xmath312 1.14 & 1.67@xmath357 & 2.32@xmath358 & 2.25@xmath358 & [email protected] + [ 1ex ] 1.14 @xmath312 1.51 & 1.67@xmath359 & 2.33@xmath360 & 2.02@xmath361 & [email protected] + [ 1ex ] 1.51 @xmath312 2.00 & 1.78@xmath362 & 2.45@xmath363 & 2.17@xmath364 & [email protected] + [ 1ex ] 2.00 @xmath312 4.50 & 1.85@xmath365 & 2.71@xmath366 & 2.38@xmath367 & [email protected] + [ 1ex ] lc|c|cc & * radio - selected sample * & * ir - selected sample * & + + [ -2ex ] + [ -2ex ] 0.08 @xmath312 0.23 & 2.41@xmath341 & 2.75@xmath368 & 2.76@xmath186 & [email protected] + [ 1ex ] 0.23 @xmath312 0.33 & 2.29@xmath369 & 2.63@xmath351 & 2.64@xmath370 & [email protected] + [ 1ex ] 0.33 @xmath312 0.45 & 2.37@xmath371 & 2.71@xmath372 & 2.70@xmath373 & [email protected] + [ 1ex ] 0.45 @xmath312 0.67 & 2.28@xmath374 & 2.61@xmath375 & 2.56@xmath348 & [email protected] + [ 1ex ] 0.67 @xmath312 0.82 & 2.09@xmath359 & 2.54@xmath320 & 2.51@xmath345 & [email protected] + [ 1ex ] 0.82 @xmath312 0.94 & 2.07@xmath376 & 2.56@xmath372 & 2.51@xmath356 & [email protected] + [ 1ex ] 0.94 @xmath312 1.14 & 1.98@xmath377 & 2.61@xmath378 & 2.56@xmath379 & [email protected] + [ 1ex ] 1.14 @xmath312 1.51 & 1.97@xmath349 & 2.64@xmath343 & 2.29@xmath380 & [email protected] + [ 1ex ] 1.51 @xmath312 2.00 & 2.10@xmath381 & 2.77@xmath382 & 2.49@xmath364 & [email protected] + [ 1ex ] 2.00 @xmath312 4.50 & 2.17@xmath365 & 3.04@xmath383 & 2.72@xmath384 & [email protected] + [ 1ex ] m and 70@xmath1 m catalog ( black histogram ) , as well as the fractional contribution of sources with flux limits in either one of the two bands ( light grey detected 24@xmath1 m counterpart , upper flux limit at 70@xmath1 m ; dark grey upper flux limit at 24 @xmath1 m , detection at 70@xmath1 m ) or in both ( white area ) . as for the first panel but for the ir - selected sample ( restricted to sources with @xmath385 ) .
counterpart detected at both 70@xmath1 m and 1.4ghz ; dark grey detected counterpart at 70@xmath1 m , upper flux limit at 1.4ghz ; light grey detected counterpart at 1.4ghz , upper flux limit at 70@xmath1 m ; white upper flux limits at both 70@xmath1 m and 1.4ghz.[fig : flxchar_vs_z ] ] m ( orange symbols ) or 1.4ghz ( green ) or present in both catalogs ( black ) in plots comparing fluxes that will be used for the computation of their ir / radio properties : _ ( a ) _ 70@xmath1 m flux constraints @xmath386 as a function of 1.4ghz flux density @xmath19 ; _ ( b ) _
@xmath386 vs. @xmath387 ; _ ( c ) _
@xmath387 vs. @xmath19.the ir - selected sample comprises sources plotted in orange and black .
the union of green and black symbols defines the radio - selected sample ( see [ sect : sampreview ] ) . in panel _
c ) _ the straight cut - offs confining the distribution of the green points on the left and that of the orange points on the bottom reflect the lower flux limits of the radio- and ir - selected sample , respectively . [
fig : flxvsflx ] ] ) colours of galaxies in the radio - selected sample ( light grey ) and the ir - selected sample ( limited to sources with @xmath388 0.3 mjy ; dark grey ) in six different redshift slices .
the hatched areas indicate those sources which are common to both samples . the median ( @xmath52 )
colours in the radio- and ir - selected samples are marked by the vertical light and dark grey lines , respectively .
along the upper edge of the plot the probability @xmath389 of ` sf - hood ' associated with a given value of ( @xmath52 ) colour is indicated.[fig : probcol_distr ] ] | vla 1.4ghz ( @xmath0 0.012mjy ) and _ mips _ 24 and 70@xmath1 m ( @xmath0 0.02 and 1.7mjy , respectively ) observations covering the 2 deg@xmath2 cosmos field are combined with an extensive multi - wavelength data set to study the evolution of the ir - radio relation at intermediate and high redshift . with @xmath34500 sources
of which @xmath330% have spectroscopic redshifts
the current sample is significantly larger than previous ones used for the same purpose .
both monochromatic ir / radio flux ratios ( @xmath4 & @xmath5 ) , as well as the ratio of the total ir and the 1.4ghz luminosity ( @xmath6 ) are used as indicators for the ir / radio properties of star forming galaxies and agn .
+ using a sample jointly selected at ir and radio wavelengths in order to reduce selection biases , we provide firm support for previous findings that the ir - radio relation remains unchanged out to at least @xmath7 1.4 .
moreover , based on data from @xmath3150 objects we also find that the local relation likely still holds at @xmath8 [ 2.5 , 5 ] . at redshift @xmath9 1.4
we observe that radio - quiet agn populate the locus of the ir - radio relation in similar numbers as star forming sources . in our analysis
we employ the methods of survival analysis in order to ensure a statistically sound treatment of flux limits arising from non - detections .
we determine the observed shift in average ir / radio properties of ir- and radio - selected populations and show that it can reconcile apparently discrepant measurements presented in the literature .
finally , we also investigate variations of the ir / radio ratio with ir and radio luminosity and find that it hardly varies with ir luminosity but is a decreasing function of radio luminosity . |
the peccei - quinn ( pq ) mechanism and concomitant axion is arguably the most plausible explanation for the smallness of the @xmath7 parameter of qcd @xcite .
this case has become stronger in recent years because the experimental precision exploration of the quark mixing matrix reveals that the kobayashi
maskawa mechanism of explicit cp violation accounts perfectly well for all observations @xcite , so one can hardly argue that cp was a fundamental symmetry that is only spontaneously broken .
likewise , the possibility of a massless up - quark is now convincingly excluded @xcite , leaving few credible alternatives to the pq mechanism .
perhaps the greatest advantage of the pq mechanism is that it can be verified by the detection of axions .
the cp - violating term in the qcd lagrangian is of the form @xmath8 , with @xmath9 the strong fine structure constant and @xmath10 the colour field - strength tensor . with the identification
@xmath11 the axion field @xmath12 inherits its defining interaction with the axial qcd anomaly .
the axion decay constant @xmath5 is the primary parameter for axion physics . in particular , all couplings are inversely proportional to @xmath5 and the axion mass is determined by @xmath13 here @xmath14 mev is the pion mass , @xmath15 mev its decay constant , and @xmath16 the up- and down - quark masses with @xmath170.6 @xcite .
a series of overlapping experimental and astrophysical constraints suggests @xmath18 gev @xcite , so axions are very light and `` invisible , '' in any case far more elusive than neutrinos
. however , during the qcd epoch of the early universe , a non - thermal mechanism produces axions as nonrelativistic coherent field oscillations that can play the role of cold dark matter @xcite . in terms of the initial `` misalignment angle '' @xmath19 relative to the cp - conserving minimum of the axion potential ,
the cosmic axion density is @xcite @xmath20 if @xmath21 is of order unity , axions provide the dark matter of the universe if @xmath22 gev ( @xmath23 ) . as such
they can be detected by experiments that exploit the generic axion
two photon interaction , allowing for the coherent conversion into excitations of a high - quality microwave cavity placed in a strong @xmath24-field @xcite .
if axions are in the mass range 1@xmath25ev and form the dark matter of our galaxy , the admx and new carrack experiments are poised to find them within a decade . such a discovery
would both reveal the nature of dark matter and verify the pq mechanism for cp symmetry restoration .
meanwhile one may study other `` windows of opportunity '' where axions could leave a detectable trace .
ongoing efforts include helioscope searches that are sensitive to solar axions if @xmath26 gev , corresponding to sub - ev masses @xcite . of course
, if solar axions were found something would have to be wrong with the supernova 1987a energy - loss limit that requires @xmath27 gev .
for @xmath28 gev axions thermalise in the early universe after the qcd epoch , so sub - ev mass axions would provide a hot dark matter component @xcite .
if future surveys were to reveal a hot dark matter fraction beyond the minimal neutrino component , an axion interpretation would be conceivable .
we consider here axions in another range beyond the classical cosmological window . in a scenario where the pq symmetry is not restored during or after inflation
, a single value @xmath29 determines the axion density .
it is possible that @xmath30 , allowing for @xmath0 gev .
this case is motivated because the pq mechanism presumably is embedded in a greater framework .
in particular , the pq symmetry emerges naturally in many string scenarios , but then it is hard to obtain @xmath5 much below @xmath31 gev @xcite .
the axion is the nambu
goldstone boson of the global @xmath32 symmetry that is broken at a scale @xmath33 where @xmath34 is an integer .
( both @xmath35 and @xmath36 must be @xmath37-periodic because @xmath35 is the phase of a new higgs field responsible for spontaneous symmetry breaking and the qcd vacuum is periodic in @xmath38 . ) since @xmath39 encounters @xmath34 minima of the axion potential , there is a formidable cosmological domain - wall problem unless @xmath40 or inflation occurs after pq symmetry breaking , further motivating a `` late - inflation scenario '' @xcite .
postulating @xmath30 may look like shuffling the qcd problem of a small @xmath7 parameter into an unnatural initial condition for our universe .
however , a gut - scale value of @xmath41 gev would only require a modest fine - tuning of @xmath42 @xcite .
more importantly , anthropic selection suggests that the cosmic baryon / dark matter ratio can not be too small , so for any @xmath43 the most probable value for @xmath44 is near the one required for the observed dark - matter density @xcite . following these papers
we emphasise that in our late - inflation scenario @xmath44 , and therefore the axion dark matter density , is an environmental parameter of our universe , not a fundamental one , and that its prior probability distribution is known to be flat on the interval @xmath29 .
this contrasts with typical other cases of anthropic reasoning where usually it is not known if a given cosmological or particle - physics parameter is fundamental or environmental . in the axion case anthropic reasoning
is unavoidable to quantify if a given value of @xmath44 is natural or not .
therefore , the `` anthropic axion window '' with @xmath0 gev is well motivated . in this case @xmath45ev and axion dark matter is difficult to detect with the cavity technique because of the large required cavity size and magnetic field region .
an alternative signature is provided by primordial isocurvature fluctuations that can show up in future data . in our late - inflation scenario ,
the massless axion field is present during inflation and thus acquires the usual amplitude fluctuations imprinted by the de sitter expansion .
when axions acquire a mass during the qcd epoch , these fluctuations become dynamically relevant in the form of isocurvature fluctuations that are uncorrelated with the adiabatic fluctuations inherited by all other matter and radiation from the inflaton field .
the isocurvature amplitude depends on both @xmath5 and @xmath46 , the hubble parameter during inflation , so observational limits on isocurvature fluctuations exclude certain regions in this parameter space @xcite .
since there is no trace of isocurvature fluctuations in existing data , perhaps a more interesting question is the remaining window for axions to show up in future .
one forecast was recently provided by the cmbpol study team collaboration @xcite .
the purpose of our paper is to provide a more detailed sensitivity forecast on the amplitude of isocurvature fluctuations .
we explore possible degeneracies that may exist between the isocurvature fluctuations and other cosmological parameters in both present and future data , and determine the region in the underlying axion parameter space that is realistically within the reach of future cosmological probes . to this end
we review in sec .
[ sec : observables ] the axion dark matter and isocurvature contributions in terms of underlying physical parameters . in sec .
[ sec : constraints ] we derive our constraints and sensitivity forecasts on the axion - type isocurvature fraction .
we interpret these results in terms of underlying physical parameters in sec .
[ sec : axionparameters ] and conclude in sec .
[ sec : conclusions ] .
to specify the cosmological imprint of axions we begin with their contribution to the matter density of the present - day universe .
when the pq symmetry breaks at some large temperature @xmath47 , the relevant higgs field will settle in a minimum corresponding to @xmath19 , where @xmath48 .
we assume that this happens before cosmic inflation , so throughout our observable universe we have the same initial condition except for fluctuations imprinted by inflation itself .
when the universe cools and evolves towards the qcd epoch , instanton effects become important and produce a potential for the axion field .
the corresponding temperature - dependent axion mass is @xcite @xmath49 where @xmath50 , @xmath51 , @xmath52 , and @xmath53 , assuming @xmath54 .
the axion field begins oscillating at @xmath55 .
the hubble parameter during radiation domination is @xmath56 where @xmath57 is the planck mass . for @xmath58
the effective number of thermally excited degrees of freedom is @xmath59 , whereas @xmath60 for @xmath61 .
together equations ( [ eq : mass ] ) and ( [ eq : hubble ] ) give @xmath62 we see that the limit @xmath63 is only relevant for @xmath64 .
we will usually focus on @xmath5 values around or below @xmath65 and therefore concentrate on the case where the axion field begins oscillating during the qcd epoch .
the traditional calculation of the present - day axion density @xcite was recently revisited in detail , taking into account modern values for the relevant qcd parameters and anharmonic corrections for the axion potential @xcite . however , since we are interested in the anthropic axion window where the initial misalignment angle is small , anharmonic effects are not important .
we also ignore the possibility of axion dilution by late entropy production , so @xmath66 here , the initial misalignment angle @xmath44 is to be interpreted as the average in our hubble volume , @xmath67 , while @xmath68 is the inflation - induced variance , with @xmath6 the hubble parameter during inflation . in other words ,
the relevant initial parameter is @xmath69 .
we will usually assume that all of the cold dark matter consists of axions , so according to current cosmological data @xmath70 @xcite .
assuming @xmath71 is small , we then find @xmath72 as a unique relationship between the initial misalignment angle and the axion decay constant .
since the axion is essentially massless during inflation , quantum fluctuations are imprinted on the axion field in the same manner that the inflaton field acquires its perturbations . because of the axion s negligible contribution to the total energy budget of the universe during inflation , these fluctuations do not perturb the total energy density and manifest themselves only as perturbations in the ratio of axion number density to entropy density , i.e. , @xmath73 , and @xmath74 . for this reason they are known as entropy or isocurvature perturbations . as the axion field provides some or all of the cold dark matter of the universe , these perturbations contribute to the temperature and polarisation fluctuations in the cmb .
axion - induced isocurvature fluctuations are uncorrelated with the adiabatic fluctuations inherited by other matter and radiation components from the inflaton .
the spectrum of entropy perturbations is @xmath75 assuming @xmath76 has a gaussian distribution , the perturbation power spectrum can be written as @xmath77 where @xmath44 and @xmath78 are the mean and the variance of the initial @xmath7 distribution .
as we shall see later , for all of the viable parameter space , @xmath79 , thus @xmath80 in the second equality we have approximated the power spectrum by a power law with spectral index @xmath81 , where @xmath82 is the first inflationary slow - row roll parameter and @xmath83 the pivot scale .
likewise , adiabatic perturbations sourced by the inflaton field are encoded in the curvature power spectrum , @xmath84 where @xmath85 is the adiabatic spectral index . because fluctuations of the axion field are uncorrelated with those of the inflaton field ,
the primordial scalar power spectrum is simply the incoherent sum of the adiabatic and isocurvature perturbation spectra , @xmath86 the resulting cmb temperature and polarisation anisotropy spectra are given by a similar sum , but with the appropriate transfer functions @xmath87 folded in ( @xmath88 ) , @xmath89 here it is useful to define the isocurvature fraction @xmath90 , @xmath91 where @xmath92 is the amplitude of the total primordial scalar power spectrum at the pivot scale @xmath93 , and we have assumed @xmath94 .
given @xmath95 , the solution @xmath96 can not be realised . in the opposite limit @xmath97
we find @xmath98 where we have used equation ( [ eq : omc ] ) . the misalignment mechanism _ per se _
does not produce gravity waves .
however , all inflationary scenarios produce some amount of tensor perturbations , which is in principle manifest in the cmb temperature and polarisation anisotropies .
the quantity of interest is the tensor - to - scalar ratio @xmath99 , defined as the ratio of the primordial tensor to the scalar perturbation spectrum at @xmath100 .
the amplitude of tensor perturbations from inflation is @xmath101 .
thus , the hubble expansion rate during inflation @xmath102 is directly probed by a measurement of @xmath103 .
to discuss the observational imprint in the cmb we recall that the axion field causes scale - invariant cdm isocurvature fluctuations that are uncorrelated with the dominant inflaton - induced adiabatic fluctuations .
compared to adiabatic initial conditions , the growth of isocurvature cdm density perturbations with wavenumber @xmath104 is suppressed by a factor of @xmath105 on sub - horizon scales @xcite . hence ,
if we are looking for traces of a subdominant isocurvature perturbation with a similar primordial scale dependence as the dominant adiabatic mode , the signal will be strongest at large scales .
we illustrate this point in fig .
[ fig : isocls ] , where we plot the cmb angular power spectra for a model with purely adiabatic perturbations ( @xmath106 ) and @xmath107 , and for a corresponding model with purely cdm isocurvature perturbations ( @xmath108 ) with @xmath109 . for uncorrelated
isocurvature perturbations , cases with @xmath110 simply correspond to weighted sums of these two spectra , so even for largish values of @xmath90 , the spectrum at multipoles @xmath111 will be completely dominated by the adiabatic contribution .
it is also useful to contrast the isocurvature signal with the cosmic variance , which poses a fundamental limit to the amount of information to be extracted from cmb anisotropies . in fig .
[ fig : isoclcv ] we compare the contribution @xmath112 of a subdominant isocurvature part to the total @xmath113 with the intrinsic uncertainty due to cosmic variance , @xmath114 , with @xmath115 $ ] .
this shows that data of small - scale perturbations by themselves carry hardly any information about axion - type isocurvature . , dark red lines ) with @xmath116 and for purely cdm isocurvature initial conditions ( @xmath108 , light red lines ) with a scale invariant spectrum @xmath109 .
all other cosmological parameters are kept fixed .
[ fig : isocls ] ] ( blue lines ) , and @xmath117 ( pale blue lines ) .
plotted is the distortion @xmath118 of the cmb angular power spectra caused by adding a subdominant scale - invariant uncorrelated cdm isocurvature component , given in units of the cosmic variance @xmath119 .
[ fig : isoclcv ] ] based on existing measurements of the cmb anisotropies , the main signature of scale - independent uncorrelated cdm isocurvature , i.e. , an enhancement of the sachs wolfe plateau , can be mimicked to a certain extent by changing the values of other cosmological parameters . in particular
, one can expect @xmath90 to be degenerate with the total matter density @xmath120 and the adiabatic spectral index @xmath85 .
adding data of the large - scale matter power spectrum and of distance probes such as supernovae will however allow us to break these degeneracies and thus improve constraints on @xmath90 indirectly . for inferring constraints from present data and predicting the discovery potential of future cmb experiments
, we consider a cosmological model with seven free parameters : the isocurvature fraction @xmath90 plus the usual six parameters of the `` vanilla '' @xmath2cdm model .
these parameters and their benchmark values to generate mock data for our forecasts are the baryon density @xmath121 , the dark matter density @xmath122 , the hubble parameter @xmath123 , the redshift to reionisation @xmath124 , the scalar spectral index @xmath125 , the normalisation @xmath126 , and the isocurvature fraction @xmath127 .
the spectral index of the primordial isocurvature power spectrum is fixed to @xmath128 . for the other parameters we adopt flat priors in the data analysis .
we do not vary the tensor - to - scalar ratio @xmath103 , since in the anthropic parameter region any possibly observable amount of tensor perturbations would imply a value of @xmath6 high enough to make the isocurvature fraction close to unity , which is clearly inconsistent with observations @xcite .
parameters are estimated using standard markov chain monte carlo techniques , as implemented in the publicly available package ` cosmomc ` @xcite . as input
we use four different data sets : * wmap : the five - year wmap data @xcite .
* wmap+sdss - lrg+sn : wmap plus the galaxy power spectrum from the luminous red galaxy subsample of the sloan digital sky survey @xcite and the luminosity distances of supernovae from the union compilation @xcite .
* planck : simulated @xmath129 , @xmath130 and @xmath131 spectra up to @xmath132 from the planck satellite @xcite , assuming isotropic white noise , a sky coverage of 70% , and 14 months of observation in the 70 , 100 , and 143 ghz channels .
* cvl : simulated , noiseless ( i.e. , cosmic variance limited ) @xmath129 , @xmath130 and @xmath131 spectra up to @xmath132 .
the mock cmb data sets 3 and 4 are generated using the method discussed in ref .
the likelihood function @xmath133 is defined as @xmath134,\ ] ] where @xmath135 with @xmath115 $ ] denotes the mock data covariance matrix , @xmath136 is the total covariance matrix comprising theoretical predictions of the cmb anisotropy spectrum @xmath137 and the noise power spectrum @xmath138 . the quantity @xmath139 counts the number of observable modes , where @xmath140 for observations in temperature and @xmath4-type polarisation . using data set 2 , we find 95% credible intervals for the observables discussed in sec .
[ sec : observables ] , @xmath141 < 3.260 , \\
\quad \alpha & < & 0.09,\end{aligned}\ ] ] consistent with the findings of komatsu et al .
@xcite . as noted by these authors ,
adding large - scale structure and supernova data leads to a significantly tighter bound on @xmath90 because these data sets contain additional information on @xmath120 , and thus help alleviate the degeneracy problem somewhat .
this is illustrated in fig .
[ fig:2d ] , where we show the two - dimensional joint constraints on @xmath90 with the other parameters of the vanilla model for data sets 14 . with the other parameters of the standard @xmath2cdm - model for different data sets .
plotted are the marginalised two - dimensional 95% credible contours for wmap data only ( pale red ) , wmap+sdss - lrg+sn ( red ) , and for simulated data from planck ( thin pale blue dotted ) and a cosmic variance limited experiment ( thin blue dotted lines ) .
[ fig:2d ] ] for the future experiments and if no isocurvature signal shows up , we forecast 95%-credible upper limits of @xmath142 in rough agreement with the cmbpol white paper @xcite.% improvement in the parameter sensitivities compared with numbers derived from a 70% sky coverage ) and the inclusion of @xmath24-polarisation ( which has no bearing on isocurvature detection ) .
a major difference between ref .
@xcite and our analysis is the forecast method : we have analysed mock future data , while the forecasts of ref .
@xcite are based on the fisher matrix which assumes gaussian posterior distributions .
] we emphasise that upcoming cmb experiments will break essentially all remaining degeneracies without the need for additional data .
the improvement one can expect from planck is partly due to this effect , but the improved sensitivity to @xmath4-polarisation will also play a role ( planck @xmath4-polarisation data will essentially be cosmic variance limited for @xmath143 ) .
going from planck to cvl will increase the sensitivity to @xmath90 by another factor of 2.5 , mostly because cvl covers the `` sweet spot '' in the @xmath4-polarisation signal - to - noise , where the first acoustic peak of the isocurvature signal coincides with the first acoustic trough of the adiabatic signal , around @xmath144 ( see fig . [
fig : isoclcv ] ) .
because the isocurvature signal affects predominantly the largest scales , any significant further increase in the sensitivity to @xmath90 from cmb observations beyond the cvl benchmark value is most likely unrealistic .
it is now easy to translate our constraints and sensitivity forecasts on the isocurvature fraction @xmath90 into axion parameters by virtue of equation ( [ eq : axionalpha ] ) .
it can be written as @xmath145 where we have used our present - day @xmath90-constraint from equation ( [ eq : currentconstraint ] ) as a benchmark value .
assuming axions are the dark matter , we show this constraint in fig .
[ fig : tegmark ] with a line marked `` current data . '' in this plot , which is inspired by a similar figure in ref .
@xcite , we also show the relationship between @xmath5 and @xmath44 as dashed lines marked with the relevant value of @xmath44 .
we have also marked the region where the pq scale is smaller than @xmath6 and where therefore our `` late inflation scenario '' would not apply .
also displayed is the region of @xmath6 values excluded by excessive tensor modes , corresponding to @xmath146 .
future sensitivities to @xmath90 from planck and cvl based on equation ( [ eq : sensitivities ] ) are shown labelled `` planck '' and `` cvl '' respectively .
( hubble rate during inflation ) and @xmath5 ( axion decay constant ) , assuming axions are all of the dark matter .
the isocurvature exclusion region based on current data is shown in light blue with a boundary given by equation ( [ eq : currentconstraint ] ) .
the sensitivity forecasts for planck and cvl of equation ( [ eq : sensitivities ] ) are also indicated .
the dashed lines indicate the required @xmath44 for a given @xmath5 to obtain the full amount of axion dark matter .
we also show the region of excessive tensor modes and the region @xmath147 where our late - inflation scenario is not applicable.[fig : tegmark ] ]
we have studied current cosmological limits and future sensitivities for axion - type isocurvature fluctuations .
our present - day limit on the isocurvature fraction @xmath3 ( 95% c.l . ) agrees with similar recent results by other authors .
we have used a @xmath2cdm cosmological model with six standard parameters and in addition a scale - invariant isocurvature fluctuation spectrum that is uncorrelated with the dominant adiabatic fluctuations .
the main effect of axion isocurvature fluctuations is to modify the sachs
wolfe plateau at small cmb multipole orders .
however , breaking degeneracies with other parameters , future cmb probes can significantly improve the sensitivity to @xmath90 and thus to the presence of axions in the universe .
a non - detection of isocurvature modes by planck would improve the limit to @xmath148 , while an ultimate cmb probe limited only by cosmic variance in both temperature and @xmath4-polarisation up to @xmath149 could reach @xmath150 , about a factor of five more restrictive than current limits .
these forecasts agree roughly with previous estimates , notably by the cmbpol collaboration @xcite .
the small differences are probably due to their using fisher - matrix techniques in contrast to our full - fledged analysis of mock data sets .
we emphasise that the cvl sensitivity to @xmath90 is essentially `` as good as it gets '' , since all relevant degeneracies have been lifted and no other cosmological probe besides the cmb is directly sensitive to isocurvature fluctuations . in terms of the axion decay constant @xmath5 , the cosmic - variance limited probe shifts the sensitivity region by approximately a factor of five for some fixed hubble parameter during inflation @xmath6 , and thus opens a sliver of parameter space where axions can still leave a detectable imprint in the microwave sky .
if a tentative signal were found , an experimental search for axion dark matter in this parameter range would be crucial , yet technologically challenging .
the idea of dark - matter axions with @xmath5 at the gut scale or larger has received relatively little attention , at least from the perspective of experimental searches , because the initial misalignment angle would have to be much smaller than unity . however , in our late - inflation scenario the axion dark matter density is fixed by the random number @xmath29 with a flat prior distribution . in this case
anthropic selection is practically unavoidable and implies that the required small @xmath44 value is not unnatural .
therefore , as stressed by previous authors , the anthropic axion window is a plausible parameter range . in the absence of direct experimental searches in this window ,
the appearance of isocurvature fluctuations is the only opportunity for axions to become visible in this scenario .
however , in view of the fundmental limitations posed by cosmic variance on all cosmological probes , a large chunk of parameter space in @xmath5 and @xmath6 will likely remain unreachable by this method .
we acknowledge use of computing resources from the danish center for scientific computing ( dcsc ) . in munich ,
partial support by the deutsche forschungsgemeinschaft under the grant tr 27 `` neutrinos and beyond '' and the cluster of excellence `` origin and structure of the universe '' is acknowledged .
99 r. d. peccei , `` the strong cp problem and axions , '' lect .
notes phys .
* 741 * ( 2008 ) 3 [ hep - ph/0607268 ] .
j. e. kim and g. carosi , `` axions and the strong cp problem , '' rev .
, to be published [ arxiv:0807.3125 ] .
c. amsler _ et al . _
[ particle data group ] , `` review of particle physics , '' phys .
b * 667 * ( 2008 ) 1 .
g. g. raffelt , lect .
notes phys .
* 741 * ( 2008 ) 51 [ hep - ph/0611350 ] .
p. sikivie , `` axion cosmology , '' lect .
notes phys .
* 741 * ( 2008 ) 19 [ astro - ph/0610440 ] .
k. j. bae , j. h. huh and j. e. kim , `` update of axion cdm energy , '' jcap * 0809 * ( 2008 ) 005 [ arxiv:0806.0497 ] .
s. j. asztalos , l. j. rosenberg , k. van bibber , p. sikivie and k. zioutas , `` searches for astrophysical and cosmological axions , '' ann .
nucl . part .
* 56 * ( 2006 ) 293 .
y. inoue _ et al .
_ , `` search for solar axions with mass around 1 ev using coherent conversion of axions into photons , '' phys . lett .
b * 668 * ( 2008 ) 93 [ arxiv:0806.2230 ] .
e. arik _ et al .
_ [ cast collaboration ] , `` probing ev - scale axions with cast , '' jcap * 0902 * ( 2009 ) 008 [ arxiv:0810.4482 ] .
s. hannestad , a. mirizzi and g. raffelt , `` new cosmological mass limit on thermal relic axions , '' jcap * 0507 * ( 2005 ) 002 [ hep - ph/0504059 ] .
a. melchiorri , o. mena and a. slosar , `` an improved cosmological bound on the thermal axion mass , '' phys .
d * 76 * ( 2007 ) 041303 [ arxiv:0705.2695 ] .
s. hannestad , a. mirizzi , g. g. raffelt and y. y. y. wong , `` cosmological constraints on neutrino plus axion hot dark matter : update after wmap-5 , '' jcap * 0804 * ( 2008 ) 019 [ arxiv:0803.1585 ] .
p. svrcek and e. witten , `` axions in string theory , '' jhep * 0606 * ( 2006 ) 051 [ hep - th/0605206 ] .
so - young pi , `` inflation without tears , '' phys .
* 52 * ( 1984 ) 1725 .
a. d. linde , `` inflation and axion cosmology , '' phys .
b * 201 * ( 1988 ) 437 .
m. tegmark , a. aguirre , m. rees and f. wilczek , `` dimensionless constants , cosmology and other dark matters , '' phys .
d * 73 * ( 2006 ) 023505 [ astro - ph/0511774 ] .
d. h. lyth , `` a limit on the inflationary energy density from axion isocurvature fluctuations , '' phys .
b * 236 * ( 1990 ) 408 .
m. beltrn , j. garca - bellido and j. lesgourgues , `` isocurvature bounds on axions revisited , '' phys .
d * 75 * ( 2007 ) 103507 [ arxiv : hep - ph/0606107 ] .
m. p. hertzberg , m. tegmark and f. wilczek , `` axion cosmology and the energy scale of inflation , '' phys .
d * 78 * ( 2008 ) 083507 [ arxiv:0807.1726 ] .
e. komatsu _ et al .
_ [ wmap collaboration ] , `` five - year wilkinson microwave anisotropy probe ( wmap ) observations : cosmological interpretation , '' astrophys .
j. suppl .
* 180 * ( 2009 ) 330 [ arxiv:0803.0547 ] .
b. kaplan and a. e. nelson , arxiv:0809.1206 [ astro - ph ] .
d. baumann _ et al .
_ [ cmbpol study team collaboration ] , `` cmbpol mission concept study : probing inflation with cmb polarization , '' arxiv:0811.3919 [ astro - ph ] .
m. bucher , k. moodley and n. turok , phys .
d * 62 * ( 2000 ) 083508 [ astro - ph/9904231 ]
. m. r. nolta _ et al . _
[ wmap collaboration ] , `` five - year wilkinson microwave anisotropy probe ( wmap ) observations : angular power spectra , '' astrophys .
j. suppl .
* 180 * ( 2009 ) 296 [ arxiv:0803.0593 ] .
m. tegmark _ et al . _
[ sdss collaboration ] , `` cosmological constraints from the sdss luminous red galaxies , '' phys .
d * 74 * ( 2006 ) 123507 [ astro - ph/0608632 ] .
m. kowalski _ et al .
_ , `` improved cosmological constraints from new , old and combined supernova datasets , '' astrophys .
j. * 686 * ( 2008 ) 749 [ arxiv:0804.4142 ] .
m. bersanelli _ et al .
_ , eds . , [ planck science team ]
, `` planck : the scientific programme , '' arxiv : astro - ph/0604069 . l. perotto , j. lesgourgues , s. hannestad , h. tu and y. y. y. wong , `` probing cosmological parameters with the cmb : forecasts from full monte carlo simulations , '' jcap * 0610 * ( 2006 ) 013 [ astro - ph/0606227 ] .
p. fox , a. pierce and s. d. thomas , arxiv : hep - th/0409059 .
a. lewis and s. bridle , `` cosmological parameters from cmb and other data : a monte - carlo approach , '' phys . rev .
d * 66 * ( 2002 ) 103511 [ astro - ph/0205436 ] .
s. kasuya and m. kawasaki , arxiv:0904.3800 [ astro-ph.co ] . | we explore the cosmological sensitivity to the amplitude of isocurvature fluctuations that would be caused by axions in the `` anthropic window '' where the axion decay constant @xmath0 gev and the initial misalignment angle @xmath1 . in a minimal @xmath2cdm cosmology extended with subdominant scale - invariant isocurvature fluctuations ,
existing data constrain the isocurvature fraction to @xmath3 at 95% c.l .
if no signal shows up , planck can improve this constraint to 0.042 while an ultimate cmb probe limited only by cosmic variance in both temperature and @xmath4-polarisation can reach 0.017 , about a factor of five better than the current limit . in the parameter space of @xmath5 and @xmath6 ( hubble parameter during inflation )
we identify a small region where axion detection remains within the reach of realistic cosmological probes .
cern - ph - th/2009 - 044 + lapth-1323/09 + mpp-2009 - 37 |
99 j. dittrich , p. exner , and p. seba , * 30 ( 1989 ) 2875 .
m. loewe and m. sanhueza , * 23 ( 1990 ) 553 .
d. villarroel , _
european j. phys . * 19 ( 1998 ) 85 .
b. mller , h. peitz , j. rafelski , and w. greiner , * 28 ( 1972 ) 1235 .
w. greiner , b. mller , and j. rafelski , _ quantum electrodynamics of strong fields 1985 ( berlin : springer ) .
e. c. svendsen , _ j. of math . analysis and applications
* 80 ( 1981 ) 551 .
f. coutinho and y. nogami , a * 42 ( 1990 ) 5716 . * * _ _ * * _ * * | an elementary treatment of the dirac equation in the presence of a three - dimensional spherically symmetric @xmath0-potential is presented .
we show how to handle the matching conditions in the configuration space , and discuss the occurrence of supercritical effects .
the problem of solving the dirac equation in the presence of a @xmath1 potential represents a curious and non simple situation , in contrast with the equivalent problem in non - relativistic quantum mechanics , i.e. , the schrdinger equation , which is discussed in any course on quantum mechanics .
this is basically related to the fact that being the dirac equation of first order , a singular potential , like the @xmath1 one , induces discontinuities at the level of the wave function themselves instead of the usual discontinuities that appear in the first derivative in the schrdinger scenario .
this puzzling situation has been discussed previously in the literature by many authors .
rigorous construction of self adjoint extensions for the dirac operator , allowing the handling of matching conditions at the support of the @xmath1 potential were discussed in @xcite . using their results , in @xcite a discussion of this problem
was presented .
however , we have realized that the proposed solution corresponds to a different self adjoint extension , associated to also a different singular potential . at present , a simple and elementary discussion , without the necessity of invoking sophisticated mathematical constructions , is still not available in the literature . in this note
we want to overcome this situation , showing in elementary terms how to handle the problem in the configuration space .
other authors have shown how to handle the problem in momentum space @xcite .
using our results , we discuss the occurrence of supercritical effects , i.e. , the possibility that the ground state starts to dive into the depths of the dirac sea , implying positron emission @xcite .
the dirac equation with an external potential can be written as @xmath2 where @xmath3 here @xmath4 and @xmath5 are the usual @xmath6 dirac matrices and @xmath7 is the four - component dirac spinor . in what follows
we will focus on the attractive spherically symmetric ( vector ) potential given by @xmath8 with @xmath9 .
it is convenient to recall some general properties of the solution of the dirac equation in a central potential . for more details
the reader may consult the book by greiner , mller , and rafelski @xcite and reference therein . in this case , the complete set of commuting operators is given by @xmath10 , @xmath11 , @xmath12 and @xmath13 , where @xmath14 is the total angular momentum ( i.e. , @xmath15 ) , and @xmath16 is defined by @xmath17 in this expression @xmath18 where the @xmath19 s are the pauli matrices . in terms of @xmath20 , @xmath21 .
on the other hand @xmath22 is the orbital angular momentum .
the eigenvalues of the operator @xmath13 are given by @xmath23 where @xmath24 is the four - component spinor solution of the dirac equation written in terms of the two - component upper and lower @xmath25 and @xmath26 .
in the equation , @xmath27 are the usual eigenvalues of @xmath11 according to : @xmath28 we note that the four - component dirac spinor is not an eigenfunction of @xmath29 .
however the upper and lower components , taken separately , satisfy @xmath30\psi_u(x ) \equiv l_u(l_u+1)\hbar^2 \psi_u(x),\\\ ] ] and @xmath31\psi_l(x ) \equiv
l_l(l_l+1)\hbar^2 \psi_l(x).\ ] ] note that the orbital parities of the upper and lower components have opposite signs .
it is convenient to parametrise the four - component spinor by separating the radial and angular dependence according to @xmath32 = \left [ \begin{array}{c } g(r ) \omega_{j l_u m}(\theta,\phi)\\ i f(r ) \omega_{j l_l m}(\theta,\phi)\end{array } \right],\ ] ] where @xmath33 are the spherical spinors that carry out the angular part .
here we have coupled , through appropriate clebsh - gordan coefficients , the scalar spherical harmonics @xmath34 with the eigenfunctions of the spin given by the two - components spinors @xmath35 . for the present discussion we do not need the explicit form of the angular part . if we now consider the dirac equation for a spinor parametrised in this way , it is not difficult to show that the radial components satisfy the following set of coupled differential equations : @xmath36 and @xmath37 where @xmath38 and @xmath39 . because of the linear discontinuity of the spinor function , given by the delta potential , we need to fix the boundary conditions in the neighbourhood of the shell @xmath40 .
if we multiply equation by @xmath41 , equation by @xmath42 , and then , sum both expressions to remove the singular delta function , we obtain : @xmath43 in the previous expression , the primes denote radial derivatives . by integrating between @xmath44 and @xmath45 and taken then the limit when @xmath46 we get : @xmath47 assuming that the discontinuities of these functions are well behaved we find that : @xmath48 we may consider @xmath41 and @xmath42 as the real and imaginary parts of a function in @xmath49 . in this context , equation expresses the fact that the absolute value of this function is constant when crossing the support of the @xmath50-potential .
this is in agreement with the condition established by j. dittrich , p. exner , and p. seba ( see equation ( 3.5a ) in @xcite ; see also the remarks after equation ( 21 ) in @xcite ) .
in fact , the absolute value of this function is continuous for all @xmath51 .
if we set @xmath52 and @xmath53 , becomes @xmath54 now , as a second step , we multiply the differential equations by @xmath42 and @xmath41 , respectively , and substract them to get @xmath55 dividing by @xmath56 which is continuous for all values of @xmath51 , we can integrate in the neighbourhood of the shell radius @xmath57 by using @xmath58 and since @xmath59}=\arctan ( h(x))$ ] we have : @xmath60 in this way , our second boundary condition can be written as @xmath61 expressing the coupling constant in units of @xmath62 , we define the dimensionless parameter @xmath63 . our second boundary condition can then be expressed as @xmath64 except for an arbitrary phase , the last expression can be written as a matricial relation between the radial functions at both sides of the potential , @xmath65 = \left [ \begin{array}{cc } \cos ( a/\hbar c ) & -\sin ( a/\hbar c ) \\
\sin ( a/\hbar c ) & \cos ( a/\hbar c ) \end{array } \right ] \left[\begin{array}{cc } f_{- } \\
g_{- } \end{array } \right ] \equiv a \left[\begin{array}{cc } f_{- } \\
g_{- } \end{array } \right]\ ] ] this matrix @xmath66 is unitary ( actually orthogonal ) , @xmath67 , and contains the information for finding the eigenvalue equation for the bound states . returning to our complex valued function with real and imaginary part given by @xmath41 and @xmath42 , respectively ,
it is interesting to remark that the @xmath1 manifests itself through a change of phase of this function , given by @xmath68 . for the solutions of equations and corresponding to the free case
, we may separate the space into two regions : region i , @xmath69 , @xmath70 region ii , @xmath71 , @xmath72 the relations we are looking for reduce to @xmath73 taking into account that for the ground state , for @xmath74 , we have @xmath75 and @xmath76 we can write @xmath77 in these equations @xmath78 denotes a wave number , @xmath79 . in order to find the eigenvalues of the hamiltonian
, we evaluate and of @xmath80 and use the boundary condition . in this way we are led to solving the following transcendental equation @xmath81 without solving explicitly this equation , we can analyse the behaviour of the ground state energy @xmath82 , as a function of the parameter @xmath83 , related to the coupling constant @xmath84 .
we start by introducing the following dimensionless variables @xmath85 in terms of these variables , our eigenvalue equation can be written as @xmath86 for a fixed value of the radius of the @xmath50 -shell , @xmath80 ( with @xmath87 ) , we are interested in determining the existence of the ground state energy in the interval @xmath88 ( i.e. , @xmath89 )
. the existence of an @xmath90 in this range will depend on the values of the coupling constant @xmath84 through the dimensionless parameter @xmath91 .
since @xmath92 , we get from that @xmath93 on the other hand , using we get @xmath94 using the limits and in eqn .
we see that as @xmath82 approaches the free state ( i.e. , @xmath95 ) , @xmath96 , which agrees with the fact that the bound states disappears for a vanishing potential .
proceeding as before , with @xmath97 ( i.e. , as the energy approaches the dirac sea ) , we infer from equation that @xmath91 approaches the value @xmath98 , which gives @xmath99 where @xmath100 , the minimum positive solution of , is the value of the coupling constant for which the ground state energy sinks into the dirac sea ( e.g. , for @xmath101 , @xmath102 ) . the numerical solution of equation for the dimensionless ground state energy @xmath90 as a function of @xmath103 and @xmath84 is plotted in figure 1 .
notice that for a fixed value of @xmath103 , @xmath90 is a decreasing function of @xmath84
. for all finite values of @xmath103 there are supercritical effects .
clearly , the value of @xmath100 for which @xmath90 sinks into the dirac sea , is increasing with @xmath103 .
however , the limit @xmath104 is not well defined in and thus we can not find solutions for bound sates in this limit .
we would like to remark that a general theorem by svendsen , @xcite , tells us that supercritical effects are absent in this limit . |
we fix an o - minimal expansion @xmath0 of the real field . in this paper , we study @xmath1-limits over @xmath0 as defined in section [ def ] below ; they generalize the pfaffian limits over @xmath0 introduced in ( * ? ? ?
* section 4 ) .
pfaffian limits over @xmath0 are definable in the pfaffian closure @xmath3 of @xmath0 @xcite , by the variant of marker and steinhorn s definability - of - types theorem @xcite found in van den dries ( * ? ? ?
* theorem 3.1 ) and ( * ? ? ?
* theorem 1 ) .
the @xmath1-limits over @xmath0 considered here do not seem to fall under the scope of these theorems , as explained in section [ def ] below .
nevertheless , @xmath1-limits were used by lion and rolin @xcite to establish the o - minimality of the expansion of @xmath4 by all rolle leaves over @xmath4 of codimension one . to state our results
, we work in the setting of ( * ? ? ?
* introduction ) ; in particular , recall that a set @xmath5 is a * rolle leaf over @xmath0 * if there exists a nested rolle leaf @xmath6 over @xmath0 such that @xmath7 .
first , we obtain the following generalization of ( * ? ? ?
* thorme 1 ) .
let @xmath8 be the expansion of @xmath0 by all rolle leaves over @xmath0 . 1 .
there is an o - minimal expansion @xmath9 of @xmath8 in which every @xmath1-limit over @xmath0 is definable . 2 .
let @xmath10 be a bounded , definable @xmath11-manifold and @xmath12 be a definable and integrable nested distribution on @xmath13 .
let @xmath14 be a @xmath1-limit obtained from @xmath12
. then @xmath15 .
the question then arises how @xmath9 relates to the pfaffian closure @xmath3 of @xmath0 .
indeed , we do not know in general if @xmath9 is interdefinable with @xmath8 or @xmath3 , or if @xmath16 is interdefinable with @xmath9 .
based on @xcite , we can answer such questions under an additional hypothesis : assume that @xmath0 admits analytic cell decomposition . 1 .
every @xmath1-limit over @xmath3 is definable in @xmath3 .
the structures @xmath9 and @xmath3 are interdefinable ; in particular , @xmath9 and @xmath16 are interdefinable .
we view the combination of theorems a(2 ) and b(1 ) as a non - first order extension of ( * ? ? ?
* theorem 3.1 ) and ( * ? ? ? * theorem 1 ) .
our proofs of these theorems rely heavily on terminology and notation introduced in ( * ? ? ? * introduction and section 2 ) ; we do not repeat the respective definitions here .
we prove theorem a in section [ omin ] below using the approach of @xcite , but based on a straightforward adaptation of some results of ( * ? ? ?
* section 4 ) to @xmath1-limits carried out in section [ limits ] below .
theorem b then follows by adapting ( * ? ? ?
* proposition 7.1 ) to @xmath1-limits and using ( * ? ? ?
* proposition 10.4 ) ; the details are given in section [ theorem_b ] .
let @xmath10 be a bounded , definable @xmath11-manifold of dimension @xmath17 .
we adopt the terminology and results found in ( * ? ? ?
* introduction and section 2 ) , and we let @xmath18 be a definable and integrable nested distribution on @xmath13 . a sequence @xmath19 of integral manifolds of @xmath20 is a * @xmath1-sequence of integral manifolds of @xmath12 * if there are a core distribution @xmath21 of @xmath12 , a sequence @xmath22 of rolle leaves of @xmath23 and a definable family @xmath24 of closed integral manifolds of @xmath25 such that each @xmath26 is an admissible integral manifold of @xmath12 with core @xmath27 corresponding to @xmath23 and definable part in @xmath24 corresponding to @xmath27 , as defined in ( * ? ? ?
* definition 4.1 ) .
in this situation , we call @xmath22 the * core sequence * of the sequence @xmath28 * corresponding to @xmath23 * and @xmath24 a * definable part * of the sequence @xmath28 * corresponding to @xmath22*. 1 . we think of the core sequence of @xmath28 as representing the `` non - definable part '' of @xmath28 . if @xmath29 for all @xmath30 , then @xmath28 is an admissible sequence of integral manifolds of @xmath12 as defined in ( * ? ? ?
* definition 4.3 ) .
2 . let @xmath28 be a @xmath1-sequence of integral manifolds of @xmath12 .
then there is a @xmath1-sequence @xmath31 of integral manifolds of @xmath32 such that @xmath33 for @xmath34 .
let @xmath28 be a @xmath1-sequence of integral manifolds of @xmath12 .
if @xmath28 converges to @xmath35 in @xmath36 ( the space of all compact subsets of @xmath37 equipped with the hausdorff metric ) , we call @xmath35 a * @xmath1-limit over @xmath0*. in this situation , we say that @xmath35 * is obtained from * @xmath12 , and we put @xmath38 2 it is unknown whether the family of all rolle leaves of @xmath23 is definable in @xmath3 .
definable in @xmath3 would imply the second part of hilbert s 16th problem .
] as a consequence , contrary to the situation described by ( * ? ?
* lemma 4.5 ) for pfaffian limits over @xmath0 , the variant of marker and steinhorn s definability - of - types theorem @xcite found in ( * ? ? ?
* theorem 3.1 ) and ( * ? ? ?
* theorem 1 ) does not apply ; in particular , we do not know in general whether a @xmath1-limit over @xmath0 is definable in @xmath3 .
if @xmath29 for all @xmath30 , then @xmath35 is a pfaffian limit over @xmath0 as introduced in ( * ? ? ?
* definition 4.4 ) .
let @xmath10 be a definable @xmath11-manifold of dimension @xmath17 .
we fix a finite family @xmath39 of definable nested distributions on @xmath13 ; we write @xmath40 for @xmath41 . as in (
* section 3 ) , we associate to @xmath42 the following set of distributions on @xmath13 : @xmath43 where we put @xmath44 . if @xmath45 is a @xmath11-submanifold of @xmath13 compatible with @xmath46 , we let @xmath47 be the nested distribution on @xmath45 obtained by listing the set @xmath48 in order of decreasing dimension . in this situation , if @xmath49 is an integral manifold of @xmath50 , for @xmath51 , then the set @xmath52 is an integral manifold of @xmath53 .
let @xmath54 be definable .
for @xmath55 we put @xmath56 .
[ cell_cutting ] let @xmath55 .
then there is a finite partition @xmath57 of definable @xmath11-cells contained in @xmath58 such that @xmath57 is compatible with @xmath59 for every @xmath60 and @xmath61 for every @xmath62 ; whenever @xmath49 is a rolle leaf of @xmath63 for @xmath64 , every component of @xmath65 intersects some cell in @xmath57 . by induction on @xmath66 ; if @xmath67 , there is nothing to do , so we assume @xmath68 and the corollary is true for lower values of @xmath66 . by (
* proposition 2.2 ) and the inductive hypothesis , we may assume that @xmath58 is a @xmath11-cell compatible with @xmath59 for @xmath69 .
thus , if @xmath70 , we are done ; otherwise , we let @xmath71 and @xmath72 be as in ( * ? ? ?
* lemma 3.1 ) with @xmath73 in place of @xmath42 .
let @xmath49 be a rolle leaf of @xmath63 for each @xmath74 ; it suffices to show that every component of @xmath75 intersects @xmath72 .
however , since @xmath76 has dimension , @xmath77 is a closed , embedded submanifold of @xmath58 .
thus , @xmath71 attains a maximum on every component of @xmath77 , and any point in @xmath77 where @xmath71 attains a local maximum belongs to @xmath72 .
[ cell_fiber_cutting ] let @xmath12 be a definable nested distribution on @xmath13 and @xmath78 .
then there is a finite partition @xmath57 of @xmath11-cells contained in @xmath58 such that for every rolle leaf @xmath79 of @xmath12 , we have @xmath80 and for every @xmath81 , the set @xmath82 is a submanifold of @xmath83 , @xmath84 is an immersion and for every @xmath85 and every strictly increasing @xmath86 , the projection @xmath87 has constant rank .
apply lemma [ cell_cutting ] with @xmath88 , @xmath89 for @xmath90 , @xmath91 and @xmath92 .
we assume that @xmath13 has a definable @xmath11-carpeting function @xmath71 , and we let @xmath93 be a definable distribution on @xmath13 with core distribution @xmath21 .
first , we reformulate ( * ? ? ?
* proposition 4.7 ) .
we adopt the notation introduced before ( * ? ? ?
* proposition 4.6 ) and note that the @xmath94 in ( * ? ? ?
* remark 4.2 ) can be chosen independent of the particular @xmath95 .
[ frontier_limit ] let @xmath28 be a @xmath1-sequence of integral manifolds of @xmath12 with core sequence @xmath22 , and assume that @xmath96 exists .
then @xmath97 is a finite union of @xmath1-limits obtained from @xmath98 with core sequences among @xmath99 .
exactly as for ( * ? ? ?
* proposition 4.7 ) , except for replacing `` core @xmath95 '' by `` core sequence @xmath22 '' and `` core @xmath100 '' by `` core sequence @xmath101 '' .
second , as we do not know yet whether @xmath1-limits are definable in an structure , we work with the following notion of dimension ( see also van den dries and speissegger ( * ? ? ? * section 8.2 ) ) : we call @xmath102 a * @xmath103-manifold of dimension @xmath74 * if @xmath104 and each point of @xmath45 has an open neighbourhood in @xmath45 homeomorphic to @xmath105 ; in this case @xmath74 is uniquely determined ( by a theorem of brouwer ) , and we write @xmath106 . correspondingly , a set @xmath107 * has dimension * if @xmath108 is a countable union of @xmath103-manifolds , and in this case put @xmath109 it follows ( by a baire category argument ) that , if @xmath110 and each @xmath111 has dimension , then @xmath108 has dimension and @xmath112 .
thus , if @xmath45 is a @xmath113-manifold of dimension @xmath74 , then @xmath45 has dimension in the sense of this definition and the two dimensions of @xmath45 agree .
[ limit_dimension ] in the situation of ( * ? ? ?
* lemma 1.5 ) , the set @xmath114 is either empty or has dimension @xmath74 . therefore , we replace ( * ? ?
* lemma 4.5 ) by [ limit_dim ] let @xmath35 be a @xmath1-limit obtained from @xmath12
. then @xmath35 has dimension and satisfies @xmath15 .
let @xmath28 be a @xmath1-sequence of integral manifolds of @xmath12 such that @xmath115 .
we proceed by induction on @xmath116 .
if @xmath117 , then ( * ? ? ?
* corollary 3.3(2 ) ) gives a uniform bound on the cardinality of @xmath26 , so @xmath35 is finite .
so assume @xmath118 and the corollary holds for lower values of @xmath116 . by ( * ?
* proposition 2.2 and remark 4.2 ) , we may assume that @xmath13 is a definable @xmath11-cell ; in particular , there is a definable @xmath11-carpeting function @xmath71 on @xmath13 .
for each @xmath119 , let @xmath120 be as before ( * ? ? ?
* lemma 1.3 ) with @xmath20 in place of @xmath12 .
then by that lemma , @xmath121 and each @xmath120 is an open subset of @xmath13 . hence @xmath12 is compatible with each @xmath120 , and after passing to a subsequence if necessary , we may assume that @xmath122 exists for each @xmath123 .
it follows that @xmath124 , so by ( * ? ? ?
* lemma 1.3(2 ) ) , after replacing @xmath13 with each @xmath125 , we may assume that @xmath20 is @xmath126-bounded .
passing to a subsequence again , we may assume that @xmath127 exists as well .
then by corollary [ limit_dimension ] , the set @xmath128 is either empty or has dimension @xmath116 . by proposition [ frontier_limit ] and the discussion before ( *
* proposition 4.6 ) , the set @xmath97 is a finite union of @xmath1-limits obtained from a definable nested distribution @xmath129 on a definable manifold @xmath130 that satisfies @xmath131 and @xmath132 .
so @xmath97 has dimension with @xmath133 by the inductive hypothesis , and the proposition is proved .
[ proper_npl ] a @xmath1-limit @xmath14 obtained from @xmath12 is * proper * if @xmath134 .
[ proper_limit ] let @xmath14 be a @xmath1-limit obtained from @xmath12 .
then @xmath35 is a finite union of proper @xmath1-limits over @xmath0 of degree at most @xmath135 .
we proceed by induction on @xmath116 ; as in the previous proof , we assume @xmath118 and the corollary holds for lower values of @xmath116 .
if @xmath134 , we are done , so assume that @xmath136 .
also as in the previous proof , we now reduce to the case where @xmath20 is @xmath126-bounded and @xmath127 exists .
then corollary [ limit_dimension ] implies that @xmath137 , so the corollary follows from proposition [ frontier_limit ] and the inductive hypothesis .
finally , @xmath1-limits over @xmath0 are well behaved with respect to intersecting with closed definable sets . to see this , define @xmath138 and write @xmath139 for the typical element of @xmath140 with @xmath141 and @xmath142 .
we consider the components of @xmath12 as distributions on @xmath140 in the obvious way , and we set @xmath143 , @xmath144 and @xmath145 for @xmath146 and put @xmath147 .
moreover , whenever @xmath23 is a core distribution of @xmath12 , we similarly define a corresponding core distribution @xmath148 of @xmath149 .
in this situation , for every rolle leaf @xmath95 of @xmath23 and every @xmath142 , the set @xmath150 is a rolle leaf of @xmath151 .
[ intersection_of_npl ] let @xmath35 be a @xmath1-limit obtained from @xmath12 , and let @xmath152 be a definable closed set .
then there is a definable open subset @xmath153 of @xmath154 and there are @xmath155 and @xmath1-limits @xmath156 obtained from @xmath157 such that @xmath158
. for @xmath159 put @xmath160 .
note first that @xmath161 , and the latter is equal to @xmath162 in the sense of ( * ? ? ?
* definition 1.7 ) .
next , let @xmath28 be a @xmath1-sequence of integral manifolds of @xmath12 such that @xmath163 .
then for every @xmath159 , there is a subsequence @xmath164 of @xmath165 such that the sequence @xmath166 converges to some compact set @xmath167 .
note that @xmath168 , since @xmath169 is an open set . fix a sequence @xmath170 of positive real numbers approaching @xmath171 , and for each @xmath172 ,
choose @xmath173 such that @xmath174 . passing to a subsequence if necessary
, we may assume that @xmath175 and @xmath176 exist ; note that these limits are then equal .
hence by the above , @xmath177 the reverse inclusion is obvious , so @xmath178 .
therefore , put @xmath179 ; then @xmath153 is an open , definable subset of @xmath140 and by the above @xmath180 , where @xmath181 . hence @xmath182 . since @xmath183
, it follows that @xmath184 .
since the sequence @xmath185 is a @xmath1-sequence of integral manifolds of @xmath149 , the proposition now follows from ( * ? ? ?
* remark 4.2 ) .
[ uniform_intersection ] let @xmath24 and @xmath186 be two definable families of closed subsets of @xmath37 .
then the @xmath1-limits in the previous proposition depend uniformly on @xmath187 , for all @xmath1-limits obtained from @xmath12 with definable part @xmath24 .
that is , there are @xmath188 , a bounded , definable manifold @xmath189 , a definable nested distribution @xmath149 on @xmath140 and a definable family @xmath190 of subsets of @xmath191 such that whenever @xmath35 is a @xmath1-limit obtained from @xmath12 with definable part @xmath24 and @xmath187 , there are @xmath1-limits @xmath192 obtained from @xmath149 with definable part @xmath190 such that @xmath158 .
similar to @xcite , we show that all sets definable in @xmath9 are of the following form : [ lambda - set ] a set @xmath193 is a * basic @xmath1-set * if there exist @xmath194 , a definable , bounded @xmath11-manifold @xmath195 , a definable nested distribution @xmath12 on @xmath13 with core distribution @xmath23 and , for @xmath196 , a @xmath1-sequence @xmath197 of integral manifolds of @xmath12 with core sequence @xmath198 corresponding to @xmath23 and definable part @xmath24 independent of @xmath172 , such that : for each @xmath172 , the limit @xmath199 exists in @xmath36 ; the sequence @xmath200 is increasing and has union @xmath77 .
in this situation , we say that @xmath77 is * obtained from @xmath12 * with * core distribution * @xmath23 and * definable part * @xmath24 .
a * @xmath1-set * is a finite union of basic @xmath1-sets .
we denote by @xmath201 the collection of all @xmath1-sets in @xmath202 and put @xmath203 .
[ finite_comp ] in the situation of definition [ lambda - set ] , there is an @xmath204 such that every basic @xmath1-set obtained from @xmath12 with core distribution @xmath23 and definable part @xmath24 has at most @xmath45 components . in particular ,
if @xmath193 is a @xmath1-set and @xmath205 , there is an @xmath206 such that for every @xmath207 the fiber @xmath208 has at most @xmath45 components .
let @xmath45 be a bound on the number of components of the sets @xmath209 as @xmath95 ranges over all rolle leaves of @xmath23 and @xmath72 ranges over @xmath24 .
let @xmath77 be a basic @xmath1-set as in definition [ lambda - set ] .
then each @xmath210 has at most @xmath45 components , so each @xmath211 has at most @xmath45 components , and hence @xmath77 has at most @xmath45 components . combining this observation with remark [ uniform_intersection ] yields , for every @xmath1-set @xmath193 , a uniform bound on the number of connected components of the fibers of @xmath77 .
[ definable_sets ] 1 .
any coordinate projection of a @xmath1-limit over @xmath0 is a @xmath1-set .
every bounded definable set is a @xmath1-set .
3 . let @xmath12 be a definable nested distribution on @xmath212 and @xmath213 be a rolle leaf of @xmath12 .
then @xmath213 is a @xmath1-set .
\(1 ) is obvious . for ( 2 ) , let @xmath214 be a bounded , definable cell . by cell decomposition , it suffices to show that @xmath215 is a @xmath1-set . let @xmath71 be a definable carpeting function on @xmath215 . then @xmath216 , so let @xmath217 and put @xmath218 and @xmath219 .
then for @xmath220 , the set @xmath221 is an admissible integral manifold of @xmath149 with core @xmath222 and definable part @xmath223 , so @xmath224 is a @xmath1-limit obtained from @xmath149 .
\(3 ) let @xmath71 be a carpeting function on @xmath13 . then @xmath225 so we let @xmath226 and put @xmath227 , @xmath228 , @xmath229 for @xmath230 and @xmath231 .
let @xmath232 be the components of @xmath233 ; note that each @xmath234 is a rolle leaf of @xmath149 .
thus for @xmath220 and each @xmath74 , the set @xmath235 is an admissible integral manifold of @xmath149 with core @xmath234 and definable part @xmath236 .
[ closure_properties ] the collection of all @xmath1-sets is closed under taking finite unions , finite intersections , coordinate projections , cartesian products , permutations of coordinates and topological closure .
closure under taking finite unions , coordinate projections and permutations of coordinates is obvious from the definition and the properties of nested pfaffian sets over @xmath0 . for topological closure ,
let @xmath193 be a basic @xmath1-set with associated data as in definition [ lambda - set ] .
then @xmath237 for some subsequence @xmath238 , so @xmath239 is a @xmath1-set by proposition [ definable_sets](1 ) . for cartesian products ,
let @xmath240 and @xmath241 be basic @xmath1-sets , and let @xmath242 , @xmath243 , @xmath244 and @xmath245 be the data associated to @xmath246 as in definition [ lambda - set ] , for @xmath247 .
we assume that both @xmath248 and @xmath249 are connected ; the general case is easily reduced to this situation .
define @xmath250 where @xmath251 ranges over @xmath252 , @xmath253 over @xmath254 , @xmath255 over @xmath256 and @xmath257 over @xmath258 .
we interpret @xmath259 and @xmath260 as sets of distributions on @xmath140 correspondingly , for @xmath247 , and we define @xmath261 and @xmath262 . since @xmath248 and @xmath249 are connected , each set @xmath263 is an admissible integral manifold of @xmath149 with core distribution @xmath151 .
it is now easy to see that for each @xmath172 , the limit @xmath264 exists in @xmath265 , and that the sequence @xmath266 is increasing and has union @xmath267 . for intersections ,
let @xmath268 be basic @xmath1-sets .
then @xmath269 , where @xmath270 .
therefore , we let @xmath271 be a basic @xmath1-set and @xmath272 be closed and definable , and we show that @xmath273 is a @xmath1-set .
let the data associated to @xmath77 be as in definition [ lambda - set ] , and let @xmath140 , @xmath149 and @xmath151 be associated to that data as before proposition [ intersection_of_npl ] .
let also @xmath153 be the open subset of @xmath140 given by that proposition with @xmath274 in place of @xmath215 .
then by that proposition , there is a @xmath155 such that for every @xmath172 the set @xmath275 is the union of the projections of @xmath1-limits @xmath276 obtained from @xmath157 .
note that each @xmath277 is the limit of a @xmath1-sequence of integral manifolds of @xmath157 with core distribution @xmath278 . replacing each sequence @xmath279 by a ( possibly finite ) subsequence if necessary
, we may assume that each sequence @xmath280 is increasing .
then each @xmath281 is a basic @xmath1-set and @xmath282 .
[ boundary ] let @xmath193 be a @xmath1-set . then @xmath283 is contained in a closed @xmath1-set with empty interior .
let the data associated to @xmath77 be given as in definition [ lambda - set ] and write @xmath93 .
note that @xmath284 fix an arbitrary @xmath172 ; since @xmath285 we may assume , by corollary [ cell_fiber_cutting ] , ( * ? ? ?
* remark 4.2 ) and after replacing @xmath13 if necessary , that @xmath286 is an immersion and has constant rank @xmath287 ; in particular , @xmath288 . if @xmath289 , then each @xmath290 has empty interior by proposition [ limit_dimension ] , so @xmath291 for some subsequence @xmath164 , and we conclude by propositions [ limit_dim ] and [ definable_sets](1 ) in this case .
so assume that @xmath292 ; in particular , @xmath293 is open for every @xmath172 and @xmath30 .
in this case , since @xmath13 is bounded , we have @xmath294 for each @xmath172 .
hence @xmath295 for some subsequence @xmath164
. now use propositions [ frontier_limit ] and [ definable_sets](1 ) .
following @xcite and @xcite , and proceeding exactly as in ( * ? ? ?
* corollary 3.11 and proposition 3.12 ) using the previous propositions , we obtain : [ complement ] 1 .
let @xmath193 be a @xmath1-set , and let @xmath296 .
then the set @xmath297 has empty interior .
2 . let @xmath298^m$ ] be a @xmath1-set .
then @xmath299^m \setminus x$ ] is also a @xmath1-set . for @xmath300 ,
let @xmath301 be the collection of all @xmath1-sets @xmath302 .
[ tinfomin ] the collection @xmath303 forms an o - minimal structure on @xmath304 . for each @xmath17 ,
let @xmath305 be the ( definable ) homeomorphism given by @xmath306 and let @xmath307 be the collection of sets @xmath308 with @xmath309 . by corollary [ tinfomin ] ,
the collection @xmath310 gives rise to an o - minimal expansion @xmath9 of @xmath0 . by proposition [ definable_sets](2 ) , every definable set is definable in @xmath9 .
but if @xmath213 is a rolle leaf of a definable nested distribution @xmath12 on @xmath37 , then @xmath311 is a rolle leaf of the pullback @xmath312 .
it follows from proposition [ definable_sets](3 ) that @xmath313 , so @xmath213 is definable in @xmath9 .
therefore , @xmath8 is a reduct of @xmath9 in the sense of definability .
first , we establish ( * ? ? ? * proposition 7.1 ) with `` @xmath1-limit '' and `` @xmath9 '' in place of `` pfaffian limit '' and `` @xmath3 '' .
to do so , we proceed exactly as in @xcite , making the following additional changes . * replacing `` admissible sequence '' with `` @xmath1-sequence '' , we obtain corresponding versions of lemma 4.8 , remark 4.9 , proposition 4.11 , corollary 4.13 and proposition 5.3 in @xcite . * using ( b1 ) , we obtain the corresponding version of ( * ? ? ?
* proposition 7.1 ) .
second , assuming that @xmath0 admits analytic cell decomposition , ( b2 ) and ( * ? ? ?
* proposition 10.4 ) imply that every @xmath1-limit over @xmath0 is definable in @xmath8 ; in particular , @xmath9 and @xmath8 are interdefinable .
hence , by ( * ? ? ?
* corollary 1 ) , @xmath9 and @xmath3 are interdefinable .
replacing once more @xmath0 by @xmath3 , theorem b is now proved .
* j - m lion * , * j - p rolin * , http://www.numdam.org/item?id=afst_1998_6_7_1_93_0[_volumes , feuilles de rolle de feuilletages analytiques et thorme de wilkie _ ] , ann .
toulouse math .
( 6 ) 7 ( 1998 ) 93112 | let @xmath0 be an o - minimal expansion of the real field .
we introduce a class of hausdorff limits , the @xmath1-limits over @xmath0 , that do not in general fall under the scope of marker and steinhorn s definability - of - types theorem .
we prove that if @xmath0 admits analytic cell decomposition , then every @xmath1-limit over @xmath0 is definable in the pfaffian closure of @xmath2 . |
over the past decade , relativistic hydrodynamics has established itself as an indispensable component in modeling the collective dynamics of the quark - gluon plasma ( qgp ) produced in relativistic heavy ion collisions @xcite .
the properties of the qgp , in particular its shear viscosity , can be deduced by studying its efficiency of converting initial geometric anisotropies into final anisotropic flows . because of this , measurements of collective flow anisotropies @xmath0 ( initially the elliptic flow @xmath1 , more recently higher - order harmonic flows @xmath2 @xcite ) have generated strong interest and fertilized the study of initial geometry fluctuations and their dynamical consequences @xcite .
the tool of choice for such studies is event - by - event hydrodynamics @xcite where each fluctuating initial density profile is evolved separately , followed by taking an average over the event ensemble to compute @xmath3-spectra , anisotropic flow coefficients and two - particle correlations in the final state which can then be compared with experimental data . due to the limited number of final state particles in each event , these observables can be measured with good statistical precision only as ensemble averages . due to algorithmic progress over the last few years
, the hydrodynamic evolution part is no longer the bottleneck in such event - by - event studies ; at least for ( 2 + 1)-dimensional simulations ( which assume longitudinal boost - invariance ) the largest fraction of the computer time is spent converting the hydrodynamic output into final particle distributions , either on a switching surface " between a macroscopic hydrodynamic description of the qgp fluid and a microscopic kinetic evolution of the dilute late hadronic rescattering stage @xcite , or on a
kinetic decoupling " surface marking the transition from a strongly coupled fluid directly to a non - interacting gas of free - streaming hadrons .
the high numerical cost of this hydro - to - particle conversion " process results from the large number of unstable hadron resonances that need to be included and whose post - freeze - out decays ( mostly due to strong - interaction processes , although for some comparisons with experimental data that have not been corrected @xcite for weak - decay feed - down , weak and electromagnetic decays must also be considered ) modify the finally observed particle distributions .
the hydro - to - hadron conversion algorithm is based on the cooper - frye formula @xcite which expresses the final hadron momentum distribution as an integral of the local equilibrium ( for ideal fluid dynamics ) or slightly off - equilibrium ( in viscous fluids ) distribution function for the particle species in question over the conversion surface .
contributions to the spectra of experimentally measured _ stable _
particles from the strong decays of unstable resonances are then calculated from the single - particle spectra for the resonances @xcite .
this requires the calculation of the directly emitted ( `` thermal '' ) particle momentum distributions for all @xmath4 hadron species with mass typically up to 2 gev via cooper - frye integrals , followed by the evaluation of the phase - space integrals @xcite for all contributing decay channels . on a typical personal computer with a single - core cpu
this calculation takes presently about 2 - 3 hours , compared to 10 - 15 minutes for the preceding hydrodynamic evolution with , say , the ( 2 + 1)-dimensional viscous fluid code vish2 + 1 @xcite .
the @xmath5gev cutoff in resonance mass is dictated by requiring convergence of the relative particle yields of the measured hadronic final state after all unstable resonances have been allowed to decay .
( the pion yields are especially sensitive to resonance feeddown . )
experimental evidence points to chemical decoupling at a temperature of @xmath6mev , i.e. close to the ( pseudo)critical temperature for the quark - hadron phase transition @xcite ; at this temperature , only resonances with masses above 2gev are sufficiently strongly boltzmann - suppressed that their decay contributions to stable particle yields can be safely ignored . here
we show that for an accurate determination of the pion and proton anisotropic flow coefficients @xmath0 a much smaller number of resonances needs to be taken into account than for the hadron yields , and that even the shape of the azimuthally averaged pion and proton transverse momentum spectra can be reliably determined by accounting for only a small subset of the @xmath4 resonance species mentioned above .
these are the observables needed for an extraction of the qgp shear viscosity from heavy - ion collision experiments @xcite . by rearranging the resonance decay table in order of decreasing importance for the calculation of @xmath3-spectra and @xmath0 coefficients instead of increasing mass , good convergence for these observables
can be achieved with a significantly reduced set of only about 20 - 30 resonances .
this speeds up the computation by a factor of 10 a significant gain in efficiency for the iterative determination of the qgp shear viscosity .
the analysis presented here uses final states generated with the @xmath7-dimensional boost - invariant viscous hydrodynamic code vish2 + 1 for 200@xmath8gev au+au collisions at the relativistic heavy - ion collider ( rhic ) and for 2.76@xmath8tev pb+pb collisions at the large hadron collider ( lhc ) at various collision centralities , with previously determined @xcite hydrodynamic input parameters .
we find very similar results at both collision energies and therefore show here only plots for lhc collisions .
since the decay contributions from different resonances to the mentioned observables depend only on their decay channels and transverse momentum distributions , we expect little sensitivity to the assumption of longitudinal boost - invariance implicit in our approach and expect our reordered resonance decay tables to perform equally well for both @xmath9-d and @xmath10-d hydrodynamic simulations , and for a wide range of input parameters ( such as qgp viscosity , thermalization time , initial entropy and energy density , etc . ) .
the momentum distributions of directly emitted ( `` thermal '' ) resonances of species @xmath11 are computed from the cooper - frye formula @xcite : @xmath12 here @xmath13 is the hydro - to - hadron conversion hypersurface , @xmath14 its surface normal vector , @xmath15 $ ] is the bose or fermi thermal equilibrium distribution function , and @xmath16 accounts for viscous corrections ( driven by the viscous pressure tensor @xmath17 on the conversion surface ) of the local phase - space distribution along @xmath13 .
we assume the quadratic form @xmath18 .
resonance decays increase the total yields of the stable hadrons and change their momentum distributions . for kinematic reasons ,
most of the light decay daughters have low transverse momenta , thus modifying the shape of light stable hadrons ( pions , kaons ) particle spectra mostly in the region @xmath19gev @xcite .
we denote the total decay contribution to the momentum distribution of stable hadron species @xmath11 by @xmath20 , and the total spectrum ( obtained by adding this to the thermally emitted spectrum @xmath21 ) by @xmath22 .
( we here include only strong and electromagnetic decays . ) the @xmath23-integrated total yield @xmath24 of decay products of species @xmath11 is denoted by @xmath25 , with @xmath26 , where the sum is over resonances @xmath27 and @xmath28 is the effective branching ratio ( see eq .
( [ eq4 ] ) below ) for the decay @xmath29 .
the contribution to @xmath25 from a particular resonance @xmath27 is not only influenced by its mass ( through the boltzmann suppression factor @xmath30 ) , but also by its spin degeneracy factor @xmath31 and its branching ratio @xmath28 into the decay channel that feeds stable particle species @xmath11 .
for each stable hadron species @xmath11 it is therefore a different set of resonances that makes the most important contributions .
our goal is to order the resonances in decreasing order of importance , for each stable particle species @xmath11 .
we here assume that the conversion surface has constant temperature @xmath32 .
the different hadron resonances have @xmath32-dependent non - equilibrium fugacities @xmath33 that ensure constant stable particle ratios equal to their chemical equilibrium values at @xmath34 and @xmath35 , independent of the hydro - to - hadron conversion temperature @xmath32 . while the actual fractions contributed by each resonance to the stable particle yields
depend on @xmath32 , the ordering of these fractions is largely @xmath32-independent .
we start from the resonance table in the azhydro package , which includes @xmath36 species of hadrons ( counting different isospin states such as @xmath37 , @xmath38 , @xmath39 as separate species ) with rest masses up to @xmath40 gev . after fixing the value of @xmath32
we look up the non - equilibrium fugacity @xmath33 for each of these 319 species from the eos s95p - pce tables constructed in ref . @xcite . for each stable particle species @xmath11
we then generate an ordered list of resonances @xmath27 that can decay directly into @xmath11 .
note that in this ordering we account not only for direct decay contribution but also for multi - step decay cascades , where @xmath27 first decays into an unstable resonance @xmath41 which further decays ( directly or through more intermediate steps ) into the stable species @xmath11 .
table [ tab:1 ] shows the beginning of this contribution table for positively charged pions , for a conversion temperature @xmath42mev .
the `` total contribution '' percentages @xmath43 in the third column are computed as @xmath44 where the effective branching ratios @xmath45 in eq .
( [ eq2 ] ) account for multi - step decay cascades as follows : @xmath46 | we show that in hydrodynamic simulations for relativistic heavy - ion collisions , strong resonance decay calculations can be performed with fewer species of particle resonances while preserving good accuracy in single particle spectra and flow anisotropies .
such partial resonance calculations boost computation efficiency by a factor of 10 which is essential for large scale event - by - event simulations . |
we consider representations of numbers in a base @xmath7 , @xmath8 , using a finite alphabet of digits .
when @xmath5 is positive and real , the classical way to obtain representations is to apply the greedy algorithm and use the ` canonical ' alphabet @xmath9 . a well known result of schmidt @xcite states that if @xmath5 is a pisot number then every positive element of the field @xmath3 has eventually periodic greedy @xmath5-expansion , i.e. @xmath10 on the other hand , apart from pisot numbers , only salem numbers may have this property .
however , there are no examples of salem numbers of this type .
alternative ways of representing numbers in a real base ( balanced system with symmetric alphabet , negative base , etc . ) were considered in @xcite , @xcite , and @xcite .
corresponding analogues of the schmidt s result were shown to hold @xcite and @xcite .
this emphasises the importance of pisot numbers in the theory of number representations .
it is natural to wonder what happens if we relax the condition that the expansion in base @xmath5 is generated by a specific algorithm and allow representations in a general alphabet @xmath4 .
we shall study the so called @xmath11-representations for @xmath7 , @xmath8 and @xmath12 finite , i.e. expressions of the form @xmath13 . in this article
we focus on the set @xmath14 and study the following question : which bases @xmath5 admit an alphabet @xmath4 for which @xmath15 ? clearly , from schmidt , any pisot number @xmath5 satisfies this property choosing @xmath16 . as we will show ,
it is satisfied also by complex pisot bases and a broad class of non - pisot and non - salem numbers , in particular , all algebraic integers without a conjugate on the unit circle , see corollary [ coro : algint ] .
surprisingly , being an algebraic integer is not a necessary condition .
for example , rational numbers have this property ( corollary [ coro : rational ] ) .
in theorem [ t : x ] , we provide a sufficient condition on the base @xmath5 so that there exists an alphabet @xmath4 allowing eventually periodic representations of every element of @xmath3 . although alphabets exist which yield eventually periodic representations for every element of @xmath3 for very general bases , perhaps all algebraic @xmath5 , pisot and salem numbers still play a crucial role in the study of our question . in fact , they are the only bases for which eventually periodic representations can satisfy a special property . roughly speaking , the represented number is in its modulus proportional to the highest power of the base used , and moreover , the proportionality is uniform for every element of the field @xmath3 .
we call such representations ` weak greedy ' , these ideas are expressed formally in definition [ d : weakgreedy ] .
this property of pisot and salem numbers may be seen as a strengthening of schmidt s theorem ; we state it as theorem [ t : simon ] .
an algebraic number @xmath17 is a zero of a polynomial @xmath18.$ ] the nonzero polynomial @xmath19 $ ] of minimal degree such that @xmath20 and @xmath21 is called the minimal polynomial of @xmath22 if the leading coefficient of the minimal polynomial satisfies @xmath23 , then @xmath17 is called an algebraic integer .
galois conjugates of @xmath17 are the roots of the minimal polynomial of @xmath17 . a very important role in the theory of numeration systems is played by pisot and salem numbers . 1 .
a pisot number is an algebraic integer @xmath24 whose galois conjugates are @xmath25 in absolute value .
a complex pisot number is an algebraic integer @xmath26 whose galois conjugates except for the complex conjugate are @xmath25 in absolute value .
a salem number is an algebraic integer @xmath24 whose galois conjugates are @xmath27 in absolute value and at least on of them is equal to @xmath6 in absolute value .
let @xmath7 , @xmath8 , and let @xmath12 be a finite set containing @xmath28 .
an expression @xmath29 is called a @xmath11-representation of @xmath30 .
if @xmath31 , then @xmath32 is called the leading index of the @xmath11-representation .
suppose we are given a set @xmath33 and a mapping @xmath34 such that for every @xmath35 @xmath36 then the map @xmath37 is a transformation @xmath38 which constructs a @xmath11-representation for every @xmath39 this representation is of the form @xmath40 it is a simple observation that the coefficients @xmath41 are determined by the equation @xmath42 moreover , every @xmath43 has a @xmath11-representation of the form .
[ ex:1 ] let us give several known examples of numeration systems based on the construction described above . 1 . for a positive real base @xmath24 ,
the greedy expansion of any @xmath44 was considered by rnyi @xcite .
it is given in the above framework with @xmath45 , @xmath46 and the alphabet of digits @xmath47 .
2 . akiyama and scheicher @xcite defined a representation for any real @xmath30 in the balanced system with @xmath48 , @xmath49 and the alphabet @xmath50 .
when the base is negative , say @xmath51 , to obtain expansions for any real @xmath30 with alphabet @xmath52 , one can use the transformation @xmath53 on @xmath54 with @xmath55 and @xmath56 .
this method was introduced by ito and sadahiro @xcite . currently , there is no analogue of the greedy algorithm that can be successfully applied to any @xmath5 .
@xmath11-representation for special complex bases are studied in @xcite and @xcite .
nevertheless , darczy and ktai @xcite , and then thurston @xcite , proved that for any non - real @xmath7 of modulus greater than 1 , there exists a finite alphabet @xmath12 such that every @xmath57 has a @xmath11-representation .
@xcite[t : thurston ] given a base @xmath7 , @xmath8 , and a finite set @xmath12 .
let @xmath33 be a bounded subset of @xmath58 such that * 0 belongs to the interior of @xmath33 ; * @xmath59 .
then every @xmath60 has a @xmath11-representation of the form .
theorem [ t : thurston ] provides a formula for the mapping @xmath61 .
set @xmath62 if @xmath63 .
such @xmath64 may not be unique , therefore we have a choice in how we define @xmath65 . [ pozn:2 ] given a base @xmath7 , @xmath8 , one can easily find an alphabet @xmath4 and a set @xmath66 satisfying the assumptions of theorem [ t : thurston ] .
for example , consider @xmath67 and @xmath68 , where @xmath69 stands for an open ball of radius @xmath70 centered at @xmath71 . in
@xcite it is shown that the alphabet can always be chosen as a symmetric subset of integers , @xmath72 .
the corresponding @xmath66 is shown in figure [ f ] .
( 60,30 ) ( 0,15)(1,0)60 ( 30,0)(0,1)30 ( 40,15)(0,1)10 ( 30,25)(1,0)10 ( 20,5)(1,0)10 ( 20,5)(0,1)10 ( 20,15)(1,1)10 ( 30,5)(1,1)10 ( 5,12.5)@xmath73 ( 39,25.5)@xmath74 ( 31,25.5)@xmath75 ( 19,25.5)@xmath5 ( 40,12.5)@xmath6 ( 23,8.5)@xmath33 ( 49.5,12.5)@xmath76 ( 30.5,12.5)@xmath28 ( 10,15)(20,25)(40,25)(50,15)(40,5)(20,5)(10,15 ) [ pozn:3 ] representations considered by rnyi , akiyama and scheicher , ito and sadahiro , or thurston , all share the following feature : the leading index @xmath32 of the @xmath11-representation of a number @xmath30 is closely related to the modulus of @xmath77
. more precisely , there exists a constant @xmath78 such that for every @xmath79 , one has latexmath:[\[\label{eq : * } @xmath11-representation of a non - zero @xmath60 obtained from theorem [ t : thurston ] .
its leading index @xmath32 is found as the maximal integer such that @xmath81 .
then we have @xmath82 the last inequality follows from the assumption that 0 is an interior point of @xmath33 .
this gives the estimate .
as we already mentioned , the question of the existence of an alphabet that allows @xmath3 to have periodic expansions is already solved for real bases @xmath5 such that @xmath83 is a pisot number , as a direct consequence of schmidt s result . for negative base @xmath51 described in example
[ ex:1 ] , an analogous result to the one of schmidt was proved by frougny and lai in @xcite . in this case , if @xmath5 is pisot , then all the elements of @xmath3 admit periodic expansions with the alphabet @xmath84 let us focus on a general case .
the equality @xmath15 automatically implies that @xmath85 .
we will now show that if any element of @xmath3 can be represented by an eventually periodic sequence of digits from @xmath85 , then the same is true for a finite alphabet @xmath86 .
[ l : integer alphabet ] fix @xmath87 , @xmath8 . 1
. let @xmath85 be such that every element of @xmath3 has @xmath11-representation .
then there exists an alphabet @xmath88 such that every element of @xmath3 has a @xmath89-representation .
2 . let @xmath85 be such that every element of @xmath3 has an eventually periodic @xmath11-representation .
then there exists an alphabet @xmath88 such that every element of @xmath3 has an eventually periodic @xmath89-representation .
3 . suppose there exists @xmath78 such that every @xmath90 has a @xmath11-representation for which @xmath91 where @xmath32 is the leading index .
then there exists @xmath86 , @xmath92 such that every @xmath90 has a @xmath89-representation for which @xmath93 where @xmath94 is the leading digit .
we begin by proving item @xmath6 .
let @xmath95 be as in the statement of our lemma . since each @xmath96 is an element of @xmath3 we may express each @xmath96 in the following form @xmath97where each @xmath98 @xmath99 , and @xmath100 .
importantly @xmath101 and @xmath102 do not depend on @xmath96 .
let us take @xmath90 arbitrarily , by our assumption there exists a sequence @xmath103 with digits in @xmath4 such that @xmath104substituting ( [ expand epsilon ] ) into ( [ representation1 ] ) we obtain @xmath105multiplying out the bracket in ( [ representation2 ] ) we see that @xmath30 can be expressed as @xmath106where each @xmath107 is the sum of at most @xmath101 elements from the set @xmath108 since each @xmath109 is is the sum of at most @xmath101 elements from the set @xmath110 we may conclude that there exists @xmath111 such that @xmath112 for all @xmath113 .
importantly we may choose @xmath114 so that it does not depend on @xmath30 .
thus by every element of @xmath3 has a @xmath115-representation .
now let us take @xmath116 arbitrarily and consider @xmath117 .
since @xmath118 by the above there exists a sequence @xmath119 with digits in @xmath120 such that @xmath121 multiplying through by @xmath102 yields @xmath122 .
thus @xmath123 has a @xmath124-representation .
since @xmath123 was arbitrary and @xmath114 does not depend on @xmath123 item @xmath6 holds .
to prove item @xmath76 we remark that in the quantity @xmath109 depends only upon @xmath125 therefore , if @xmath103 is eventually periodic so is @xmath126 thus we conclude that every @xmath127 has an eventually periodic @xmath115-representation if every @xmath90 has an eventually periodic @xmath11-representation .
if @xmath128 has an eventually periodic @xmath129-representation then @xmath30 has an eventually periodic @xmath124-representation . since @xmath30 is arbitrary item @xmath76 of our lemma holds . to prove item @xmath130 we begin by fixing @xmath131 .
consider @xmath128 , suppose it has a @xmath11-representation with leading digit @xmath32 which satisfies @xmath132 . by
we see that @xmath128 has a @xmath129-representation with leading digit @xmath133 importantly @xmath134 . multiplying through by @xmath102
we see that @xmath30 has a @xmath124-representation with leading digit @xmath94 . using the inequalities @xmath132 and @xmath134 we obtain @xmath93 where @xmath135 the parameter @xmath136 has no dependence on @xmath30 so we may conclude item @xmath130 from our lemma .
lemma [ l : integer alphabet ] will often allow us to assume , without loss of generality , that an alphabet @xmath4 satisfies @xmath137 .
first we show that negative pisot and complex pisot numbers admit an alphabet such that every element of the field @xmath3 can be represented by an eventually periodic sequence .
this is a generalization of one of the implications of the theorem of schmidt @xcite .
[ l:2 ] let @xmath5 or @xmath138 be a pisot number , or let @xmath5 be a complex pisot number .
let @xmath85 be a finite alphabet of digits .
if an @xmath90 has a @xmath11-representation with leading index @xmath32 , then it has an eventually periodic @xmath11-representation with leading index @xmath32 .
let @xmath127 and let @xmath139 , @xmath31
. for @xmath140 , define @xmath141 since @xmath142 the sequence @xmath143 is bounded .
let @xmath144 be a conjugate of @xmath5 and let @xmath145 be the field isomorphism induced by @xmath146 . if @xmath5 is complex and @xmath147 , then @xmath148 is bounded due to . for all other conjugates of @xmath5 ,
consider @xmath149 as @xmath150 , the sequence @xmath148 is bounded for any isomorphism @xmath145
. recall that @xmath5 is an algebraic integer , and thus for any @xmath151 the set @xmath152:|\sigma(z)|<h \text { for every isomorphism } \sigma\}\ ] ] is finite . since @xmath127 and @xmath153
, we can find @xmath99 such that @xmath154 $ ] and @xmath155 $ ] .
obviously , also @xmath156 $ ] for every @xmath140 .
therefore the elements of the sequence @xmath143 take only finitely many values .
in particular , there exist indices @xmath157 such that @xmath158 .
thus @xmath159 thus we have obtained an eventually periodic representation with the period @xmath160@xmath161 we have seen in remark [ pozn:2 ] that for any complex base @xmath5 with @xmath8 , one can find an alphabet so that every complex number can be represented .
moreover , by remark [ pozn:3 ] , the @xmath11-representations satisfy property .
therefore , by lemma [ l:2 ] , if @xmath5 is a complex pisot number , all elements of the field @xmath3 have eventually periodic @xmath11-representations satisfying .
this leads us to a partial answer to our original question .
[ t:2 ] for a base @xmath5 , which is pisot , complex pisot , or the negative of a pisot number , one can find an alphabet @xmath137 so that every @xmath127 has eventually periodic @xmath11-representation .
however , pisot and complex pisot bases are not the only bases for which an eventually periodic @xmath11-representation of all elements of @xmath3 can be found .
[ ex : sqrt5 ] let @xmath162 , @xmath163 .
every @xmath127 can be written in the form @xmath164 , where @xmath165 .
numbers @xmath166 have eventually periodic expansions in base @xmath167 and alphabet @xmath4 , say @xmath168 , @xmath169 .
together , we can write @xmath170 the resulting representation of @xmath30 in base @xmath5 is also eventually periodic . in the same way
, one can construct eventually periodic representations of numbers in @xmath3 in systems where @xmath5 is any root of any pisot or complex pisot number .
let @xmath171 , @xmath172 , and @xmath137 be finite such that every @xmath173 has an eventually periodic @xmath174-representation .
let @xmath175 and @xmath5 be such that @xmath176 .
then any @xmath127 has an eventually periodic @xmath11-representation .
obviously @xmath177 is a subfield of @xmath3 of dimension @xmath178 $ ]
. then @xmath179 forms a basis of @xmath3 over @xmath177 .
every @xmath127 can thus be written as @xmath180 , where @xmath181 . by the assumption
, every @xmath182 has an eventually periodic @xmath174-representation .
obviously , one can write these representations in such a way that the period length is common for every @xmath182 . the eventually periodic @xmath11-representation of @xmath30 can be found realizing that every power of @xmath183 can be written as @xmath184 .
in example [ ex : sqrt5 ] the representations of elements in @xmath185 were not obtained by any transformation of the type .
it is natural to ask whether these representations also satisfy property , i.e. the modulus of @xmath30 is proportional to the modulus of @xmath186 , where @xmath32 is the leading index of the representation .
the answer to such question is negative : one can always find sequences of rational integers @xmath187 such that @xmath188 , as @xmath189 tends to @xmath190 , while @xmath191 .
moreover , since @xmath192 , @xmath193 implies that @xmath194 at the same time , representations of rational integers in base @xmath167 and alphabet @xmath195 are unique up to the identity @xmath196 .
therefore , since @xmath188 , the leading indices of the representations of @xmath197 tend to infinity .
therefore one can not find a constant @xmath78 so that property holds in example [ ex : sqrt5 ] .
this leads us to the following definition .
[ d : weakgreedy ] given @xmath198 @xmath199 @xmath200 and @xmath78 , we say that a sequence @xmath103 is a weak greedy @xmath11-representation for @xmath30 with respect to @xmath201 if @xmath202 and @xmath203 suppose @xmath202 .
clearly @xmath204 , if @xmath103 were not a weak greedy @xmath11-representation for @xmath30 with respect to @xmath205 then we would have @xmath206 .
one might expect that a reasonable algorithm for generating @xmath11-representations would be one for which @xmath207 , that is one for which after applying the first step of the algorithm , i.e. determining the first digit , one is left with a smaller / simpler number with which one has to construct a @xmath11-representation .
this idea is the motivation behind definition [ d : weakgreedy ] .
intuitively the quantity @xmath201 describes how effectively an algorithm determines the leading digit .
[ l:3 ] let @xmath208 , @xmath8 , be an algebraic number .
assume that there exist @xmath209 such that @xmath210
. then one can not find an alphabet @xmath85 and a constant @xmath78 , so that every @xmath127 has an eventually periodic weak greedy @xmath11-representation with respect to @xmath201 .
we prove the lemma by contradiction .
suppose that @xmath5 admits an alphabet @xmath211 and a constant @xmath78 so that every @xmath127 has an eventually periodic weak greedy @xmath11-representation with respect to @xmath201 .
since @xmath208 and @xmath210 , @xmath5 is irrational . therefore one can find two sequences of integers @xmath212 and @xmath213 such that @xmath214 and @xmath215 as @xmath189 tends to infinity .
denote @xmath176 and take the eventually periodic representation of @xmath216 with the leading coefficient @xmath217 satisfying .
then @xmath218 where @xmath219 as @xmath220 , the inequality implies @xmath221 and thus for every @xmath222 , @xmath223 with @xmath224 .
note that @xmath225<+\infty$ ] .
in particular , there exist @xmath226 , @xmath227 , such that @xmath179 forms a basis for @xmath3 over @xmath177 .
this means that for @xmath228 , one can express @xmath229 as a linear combination of @xmath179 with coefficients in @xmath177 .
we can therefore write for @xmath216 , @xmath230 where @xmath231 is a linear combination of at most @xmath232 numbers @xmath233 with coefficients from @xmath177 .
consequently , @xmath234 and @xmath235 with @xmath224 for @xmath236
. however , the expression of @xmath216 in the form is unique , therefore @xmath237 .
this leads to contradiction , since by assumption , we have @xmath238
. one can extend the statement of lemma [ l:3 ] to complex bases @xmath5 which do not generate imaginary quadratic fields . for @xmath239 , the set @xmath240 is not dense in @xmath58 .
however , @xmath241 is , if @xmath5 is not quadratic .
therefore one finds a sequence @xmath242 with @xmath243 such that @xmath215 .
the rest of the proof remains the same .
note that exclusion of non - real quadratic bases is not a problem : for such bases we know the answer for the question about weak greedy expansions .
if @xmath5 is an algebraic integer , then it is a complex pisot number , and by theorem [ t:2 ] eventually periodic weak greedy expansions can be found . on the other hand
, it will be shown in the following proposition , that algebraic non - integers do not possess this property .
[ p : wg ] let @xmath7 , @xmath8 , be such that there exists @xmath244 and @xmath78 such that that every element of @xmath3 has an eventually periodic weak greedy @xmath11-representation with respect to @xmath201 .
then @xmath5 is an algebraic integer such that for every conjugate @xmath144 of @xmath5 it holds that @xmath245 or @xmath246 . by lemma [ l : integer alphabet ]
we may assume that @xmath247 .
first we show that for any @xmath5 there exists a strictly increasing sequence of integers @xmath248 such that @xmath249 denote @xmath250 real numbers such that @xmath251 .
if @xmath252 is rational , @xmath253 , then set @xmath254 .
thus @xmath255 and @xmath256 .
suppose that @xmath257 .
then @xmath28 is a limit point of the sequence @xmath258
. therefore one can find a sequence @xmath248 such that @xmath259 .
write @xmath260 whence @xmath261 since @xmath262 and @xmath263 , relation is satisfied .
denote @xmath264 , @xmath265 , @xmath266 .
we have @xmath267 . denoting further @xmath268 , and @xmath269
, we can estimate @xmath270 by assumption , @xmath271 has an eventually periodic @xmath11-representation @xmath272 with leading index @xmath273 satisfying @xmath274 combining this with the above , we can write @xmath275 , which further implies that @xmath276 as a consequence , we have for the sequence of exponents @xmath277 we can derive that the integer @xmath278 has an eventually periodic @xmath11-representation of the form @xmath279 where @xmath280 $ ] , @xmath281 , @xmath282 , and @xmath283 is the period - length . multiplying by @xmath284 , we find that @xmath5 is a zero of a polynomial @xmath285 where @xmath286 $ ] is such that @xmath287 .
since @xmath288 is monic , @xmath5 is an algebraic integer .
suppose that there exists a conjugate @xmath289 of @xmath5 such that @xmath172 .
consider the field isomorphism @xmath145 induced by @xmath290 .
we apply @xmath145 to @xmath291 as given in , @xmath292 subtracting from , we obtain @xmath293 which then gives @xmath294 where @xmath295
. we thus estimate @xmath296 if @xmath297 , then without loss of generality @xmath298 , and thus the term on the left hand side of diverges @xmath299 the right hand side of is bounded , which gives a contradiction .
[ t : simon ] given @xmath208 , @xmath8 .
suppose there exists a finite alphabet @xmath85 and a constant @xmath78 such that every @xmath127 has an eventually periodic weak greedy @xmath11-representation with respect to @xmath201 .
then @xmath83 is a pisot number or a salem number . to prove theorem [ t : simon ] we make use of the following result due to ferguson @xcite .
[ t : ferguson ] suppose that the irreducible polynomial @xmath18 $ ] has @xmath300 roots , at least one real , on the circle @xmath301 .
then @xmath302 where @xmath303 has no more than one real root on any circle in @xmath58 . by proposition
[ p : wg ] , if @xmath5 satisfies the hypothesis of our theorem then it is an algebraic integer with conjugates @xmath144 satisfying either @xmath246 , or @xmath304 .
if all conjugates lie in the closed unit disc , then @xmath83 is a pisot or salem number .
suppose that there exist a conjugate of @xmath5 satisfying @xmath245 .
then by theorem [ t : ferguson ] , we have for the minimal polynomial of @xmath5 that @xmath305 , where @xmath306 is the number of conjugates on the circle of radius @xmath83 .
moreover , @xmath307 has at most one real root on every circle in the complex plane .
thus @xmath308 for @xmath183 some root of @xmath307 , and since the degree of @xmath307 is strictly smaller than the degree of @xmath309 , we have @xmath310 . by lemma [ l:3 ]
we may conclude that @xmath5 does not allow weak greedy representations for any @xmath201 . a simple adaptation of the proof of theorem [ t : simon ] yields the following theorem which can be seen as a more direct generalization of the result of schmidt . [ t : schmidt generalisation ] fix @xmath24 .
suppose there exists a finite alphabet @xmath311 such that every @xmath312 has an eventually periodic @xmath11-representation , then @xmath5 is a pisot or a salem number .
the important observation in the proof of corollary [ t : schmidt generalisation ] is that for a positive real @xmath24 and a non - negative alphabet , any representation of a positive number is weak greedy with respect to some @xmath78 . more precisely ,
if @xmath313 has a @xmath11-representation , then one automatically has that @xmath314 where @xmath32 is the leading digit in the @xmath11-representation and @xmath201 is a constant depending on @xmath4 .
in this section we show that when a base @xmath5 and an alphabet @xmath4 allow parallel addition of @xmath11-representations , then the question of finding eventually periodic representation of every element of @xmath3 is simplified . given a base @xmath7 , @xmath8 , and an alphabet @xmath12 .
denote @xmath315 .
we say that @xmath11 allows parallel addition if there exist @xmath316 and @xmath317 such that * @xmath318 ; * for every @xmath319 with @xmath320 for @xmath321 for some @xmath32 and @xmath322 , it holds that @xmath323 , where @xmath324
. let @xmath7 , @xmath8 .
then there exists an alphabet @xmath12 such that @xmath11 allows parallel addition if and only if @xmath5 is an algebraic number and @xmath325 for every conjugate @xmath144 of @xmath5 . if this is the case , than one can choose a symmetric alphabet of integer digits , @xmath326 . using the parallel addition
, one can easily derive the following statement .
[ c:1 ] let @xmath7 , @xmath8 , and let @xmath4 be a symmetric alphabet such that @xmath11 allows parallel addition .
denote @xmath327 then 1 .
[ en:0 ] @xmath328 ; 2 .
[ en:1 ] @xmath329 , @xmath330 are closed under addition and subtraction ; 3 .
[ en:2 ] @xmath329 is closed under multiplication ; 4 .
[ en:3 ] @xmath331 ; 5 .
[ en:4 ] @xmath332 $ ] .
let us illustrate how the above corollary can be used to find an eventually periodic expansion of the number @xmath333 in base @xmath334 .
such a base is chosen as the most simple example of an algebraic non - integer base .
the base @xmath334 allows parallel addition algorithm in the alphabet @xmath195 .
addition is done via conversion of digits in @xmath335 back into the alphabet @xmath4 .
note that the `` for '' loop in the following algorithm is supposed to be computed for all the indices @xmath336 at the same time .
[ paralg ] a sequence @xmath337 such that @xmath338 for all @xmath339 and @xmath340 only for finitely many indices @xmath341 a sequence @xmath342 such that @xmath343 for all @xmath339 , @xmath344 only for finitely many indices @xmath336 , and @xmath345 .
@xmath346 @xmath347 @xmath348 @xmath349 @xmath350 @xmath351 using this algorithm , we can sum the elements of @xmath330 . in the following
, we will use the notation @xmath352 for the @xmath11-representation @xmath353 .
@xmath354 let us now find an eventually periodic @xmath11-representation of @xmath355 we have @xmath356 in particular @xmath357 thus we obtain @xmath358 the geometric series itself is already an eventually periodic @xmath11-representaion @xmath359 .
furthemore , we have @xmath360 , from which @xmath361 can be computed by the usual grade - school multiplication followed by applying algorithm [ paralg ] for the digit conversion . altogether we obtain @xmath362 where again the usual multiplication was used , now ( luckily ) without the need of the digit conversion . the approach to obtain eventually periodic expansion of @xmath363 in base @xmath364 , illustrated in the above example ,
is generalized in the following theorem .
[ t : x ] let @xmath7 be an algebraic number of degree @xmath101 , @xmath8 , and let @xmath365 $ ] be its minimal polynomial .
suppose that 1 .
@xmath366 for any conjugate @xmath144 of @xmath5 ; 2 .
[ en:22 ] @xmath367 $ ] .
then there exists a finite alphabet @xmath137 such that @xmath368 .
consider a symmetric integer alphabet @xmath4 which allows for @xmath5 parallel addition .
every @xmath127 can be written in the form @xmath369 , where @xmath370 , @xmath371 .
thanks to items [ en:3 ] and [ en:4 ] from corollary [ c:1 ] , it is sufficient to show that @xmath372 for every @xmath370 . for simplicity of notation , denote @xmath373 .
suppose first that @xmath374 and @xmath375 are coprime .
let @xmath376 and @xmath377 be defined as follows : @xmath378 a_d & 0 & \cdots & 0 & 0\\ 0 & a_{d } & \ddots & 0 & 0\\ \vdots & \vdots & \ddots & \ddots & \vdots\\ 0 & 0 & \cdots & a_d & 0 \end{pmatrix } \in\z^{d\times d}\,,\quad \vec{b}:=\begin{pmatrix } \beta^{d-1}\\ \beta^{d-2}\\ \vdots\\ \beta\\ 1 \end{pmatrix}\,.\ ] ] as @xmath379 , we have the relation @xmath380 on the set @xmath381 , define the relation @xmath382 as follows : @xmath383 if @xmath374 divides every entry of the matrix @xmath384 .
obviously , @xmath382 is an equivalence relation which partitions @xmath381 into a finite number of equivalence classes . therefore there exist indices @xmath385 such that @xmath386 for a matrix @xmath387 . since for any index @xmath189 we have @xmath388 , we can write @xmath389 now recall that @xmath375 is coprime to @xmath374 , thus there exists @xmath390 such that @xmath391 . for example , take @xmath392 , where @xmath252 is the euler totient function .
this implies that for any @xmath393 , we have @xmath394 . setting @xmath395
, we obtain that @xmath396 subtracting from , we obtain for @xmath395 that @xmath397 relation then implies @xmath398 the components of the vector @xmath399 all belong to @xmath400 $ ] . denoting its last component by @xmath71 , we obtain from that @xmath401 , whence we derive @xmath402\,.\ ] ] the assumptions of the theorem together with items [ en:4 ] , [ en:1 ] , and [ en:2 ] of corollary [ c:1 ] imply that @xmath403 moreover , @xmath404 by item [ en:3 ] of corollary [ c:1 ] , relations , , and imply that @xmath405 , as desired .
it remains to consider the case when @xmath406 . in this case
@xmath407 , where @xmath408 .
according to the above , we have @xmath409 .
moreover , @xmath410 for some @xmath411 .
thus @xmath412 . by the assumption [ en:22 ] of the theorem and by items [ en:2 ] and [ en:4 ] of corollary [ c:1 ]
, we derive that @xmath413 . by [ en:3 ] of corollary [ c:1 ]
, we have @xmath414 .
[ coro : algint ] let @xmath5 be an algebraic integer with no conjugate @xmath144 in modulus equal to 1 .
then there exists an alphabet @xmath137 such that @xmath415 . for an algebraic integer , we have @xmath416 .
[ coro : rational ] let @xmath417 , @xmath418 , @xmath419 .
then there exists an alphabet @xmath137 such that @xmath415 .
the minimal polynomial of @xmath5 is @xmath420 , thus it is sufficient to verify that @xmath421 $ ] .
since @xmath422 and @xmath374 are coprime , there exist @xmath423 such that @xmath424 . whence we have @xmath425 $ ] .
[ thm : finite ] let @xmath5 , @xmath8 , be an algebraic number with minimal polynomial @xmath426 $ ] , @xmath427 , and let @xmath390 , @xmath428 .
then @xmath429 $ ] if and only if its prime factorization is @xmath430 , @xmath431 , where each @xmath432 divides @xmath433 for all @xmath434 but one ( possibly different for each @xmath113 ) .
first assume that @xmath435 , in particular , that @xmath436 for some @xmath437 $ ] and @xmath438 .
this is equivalent to @xmath439 , i.e. @xmath5 is a root of the polynomial @xmath440 with @xmath441 $ ] .
the minimal polynomial @xmath309 of @xmath5 necessarily divides @xmath307 in @xmath442 $ ] .
consider a prime divisor @xmath422 of @xmath189 .
then all the coefficients of @xmath443 but one are divisible by @xmath422 .
if @xmath444 with @xmath445 , set @xmath336 and @xmath113 to be the minimal indices such that @xmath446 and @xmath447 . then @xmath448 is not divisible by @xmath449 now if we denote @xmath70 and @xmath283 the maximal indices for which @xmath450 and @xmath451 , we obtain that @xmath452 and thus @xmath453 from the uniqueness of the coefficient of @xmath307 that is not divisible by @xmath422 .
the equality @xmath454 follows from @xmath455 to prove the opposite direction , we show that for every prime @xmath422 dividing @xmath433 for all indices but some @xmath336-th , the number @xmath456 has a finite expansion .
the statement then follows from item [ en:4 ] of corollary [ c:1 ] , i.e. from the fact that @xmath457={\rm fin}_{\a}(\beta)$ ] .
as @xmath458 is coprime to @xmath422 , the equation @xmath459 has a solution @xmath460 .
we have @xmath461 . since @xmath462
, we obtain that @xmath463.\ ] ]
representations of numbers in base @xmath5 and alphabet @xmath4 generated by various algorithms are extensively studied , mainly for their arithmetic properties .
most common example are the greedy expansions in base @xmath24 , which use the canonical alphabet @xmath464 .
one considers the set @xmath465 of numbers with eventually periodic greedy @xmath5-expansions @xcite .
the set @xmath466 of numbers whose greedy @xmath5-expansions have only finitely many non - zero digits @xmath467 has been studied for example in @xcite . in @xcite
, the set @xmath468 of numbers whose greedy @xmath5-expansions use only non - negative powers of the base has been introduced .
the subject of this paper is representations in base @xmath7 with digits in an alphabet @xmath4 not necessarily issued out of a given algorithm . instead of @xmath465 , @xmath466
, we can consider @xmath330 , @xmath329 as defined in corollary [ c:1 ] .
numbers with only non - negative powers of @xmath5 in their @xmath11-representation can be seen as polynomials in @xmath5 with coefficients in @xmath4 , i.e. form the set @xmath469 $ ] .
comparison of the behaviour of these sets is surprising . while @xmath468 is never closed under addition except for integer @xmath24 , akiyama et al .
@xcite show that there exists a finite set @xmath137 such that @xmath469=\z[\beta]$ ] if and only if @xmath5 is an algebraic number such that @xmath5 and its conjugates lie either all on the unit circle , or all outside the unit circle .
the so - called ( f ) property has been studied in @xcite .
they showed for the greedy @xmath5-expansions that the equality @xmath470\cap[0,+\infty)$ ] forces @xmath5 to be a pisot number . on the other hand , allowing any representation , we can derive from @xcite that there exists an alphabet @xmath137 such that @xmath471 $ ] for any algebraic number @xmath5 without a conjugate on the unit circle .
it is not known whether this statement can be reversed .
our main tool in showing that certain classes of numbers satisfy @xmath15 is parallel addition in the numeration system @xmath11 , only known to be possible for algebraic bases with no conjugate on the unit circle .
our computational experiments suggest that every such number @xmath5 admits an alphabet @xmath4 such that @xmath15 .
it remains an open question whether this equality can hold for some algebraic number @xmath5 having a conjugate of modulus one .
it can be commented that salem numbers belong to this class .
already when considering greedy @xmath5-expansions generated by the map @xmath472 , only very restricted results are known .
schmidt @xcite conjectured , that for salem numbers @xmath5 , the orbit of any rational @xmath473 is eventually periodic , which amounts to stating that equality @xmath474 holds .
when trying to prove this conjecture , boyd in @xcite has shown that for any salem number @xmath5 of degree @xmath475 the orbit of @xmath6 under the map @xmath472 is eventually periodic .
on the other hand , in @xcite the same author provides data which put the schmidt s conjecture in doubt .
however , even if @xmath474 were not true for salem numbers @xmath5 when considering the greedy @xmath5-expansions , one may still ask about possibility of finding an alphabet of digits @xmath4 so that @xmath15 is valid .
this work was supported by the czech science foundation , grant no .
13 - 03538s .
we also acknowledge financial support of the grant agency of the czech technical university in prague , grant no . sgs14/205/ohk4/3t/14 . | we study periodic expansions in positional number systems with a base @xmath0 , and with coefficients in a finite set of digits @xmath1 we are interested in determining those algebraic bases for which there exists @xmath2 such that all elements of @xmath3 admit at least one eventually periodic representation with digits in @xmath4 . in this paper
we prove a general result that guarantees the existence of such an @xmath4 .
this result implies the existence of such an @xmath4 when @xmath5 is a rational number or an algebraic integer with no conjugates of modulus @xmath6 .
we also consider eventually periodic representations of elements of @xmath3 for which the maximal power of the representation is proportional to the absolute value of the represented number , up to some universal constant .
we prove that if every element of @xmath3 admits such a representation then @xmath5 must be a pisot number or a salem number .
this result generalises a well known result of schmidt @xcite . |
the composite fermion approach@xcite has had a great number of extremely impressive successes in describing the physics of electrons in high magnetic fields . in this picture ,
fractional quantum hall systems in total magnetic field @xmath8 are described in terms of noninteracting composite fermions " in an effective magnetic field @xmath9 , where @xmath10 is the flux quantum , @xmath11 is the electron density , and @xmath1 is a positive integer .
this maps , for example , fractional quantum hall states at filling fractions of the form @xmath12 to integer quantum hall states for the composite fermions at filling fraction @xmath13
. we will denote such composite fermions with @xmath14 flux quanta attached to them as @xmath15cf .
jain s original approach to composite fermions@xcite constructed highly accurate trial wavefunctions by taking simple wavefunctions for the noninteracting ( composite ) fermions in the effective magnetic field , multiplying by jastrow factors , and then projecting the result into the lowest landau level
. the early successes of this method were impressive@xcite , despite the fact that the method was limited by the extreme numerical difficulty of performing projections for systems with more than roughly 10 electrons .
a major theoretical breakthrough came when jain and kamilla@xcite discovered a new way of writing composite fermion trial wavefunctions ( described below ) , which involves a very minor modification of the projection .
these new trial states seemed to be just as good as the originally proposed wavefunctions and could be numerically generated even for systems with many electrons ( 40 electrons or more ) . since that time , many important studies have been achieved using this method@xcite .
however , so far this method has been restricted to cases where the effective magnetic field has the same sign as the external magnetic field .
results using this method have been published for filling fractions of the form @xmath0 with @xmath16 but not for @xmath5 . in the current paper
, we extend the work of jain and kamilla@xcite so that we are able to handle states with @xmath17 .
the @xmath18 states take more computational resources than the case of @xmath4 , and the difference in the computational resources between the two cases increases with the absolute value of the effective flux .
however , the computational problems turn out to be more severe for the smallest @xmath19 , where the number of particles in the system increases most slowly with each flux added .
fortunately , we probably need not go to particularly large systems to understand the physics of small @xmath19 . for large @xmath19 , describing the approach to the fermi liquid - like composite fermion state , the system size is already large for relatively small effective flux , i.e. the differences of the computational requirements in the case of @xmath4 and @xmath17 become relatively less important .
although the case of negative @xmath2 has not previously been studied for large systems , we point out that for @xmath3 the series of states with negative @xmath2 and the series of states with positive @xmath2 are essentially equivalent due to an exact particle - hole symmetry in the lowest landau level .
in fact , below , we exploit this symmetry to check the validity of our method .
once we have verified the method , we can study the properties of the @xmath6 series for negative @xmath2 and compare the results to those of the positive @xmath2 members of this same series as well as to those of the @xmath3 series .
the outline of this paper is as follows . in section [ sec : jk ]
we briefly review the jain - kamilla method .
as mentioned above , the method has only been used previously for the case of positive @xmath2 . in appendix
[ app : jk ] we show in detail how the crucial projection scheme of jain - kamilla can easily be generalized to handle negative @xmath2 also .
it is easy to see from the result how much additional numerical complexity is involved for negative @xmath2 . in section
[ sec : phsym ] we test our approach by examining the @xmath20cf series @xmath21 . in particular , we examine ground state energies , excitation spectra , and energy gaps .
we pay particular attention to the mass of the composite fermion and the scaling of the gap with @xmath2 . for @xmath4 , results are already available in the literature@xcite . for @xmath17
we use our generalization of the jain - kamilla method to calculate energies directly , and we compare these energies to energies obtained by particle - hole conjugating the @xmath4 series . appendix [ app : phconj ] describes the particle - hole conjugation transformation in depth .
this comparison establishes the accuracy of our method . in section [ sec : cf4results ] we move on to examine the @xmath22cf series of states @xmath23 .
again , for @xmath4 some results are already available in the literature@xcite . however , for @xmath17 our results are new .
again , we examine ground state energies , excitation spectra , energy gaps , and composite fermion effective masses .
( by using particle - hole conjugation , we could give results for filling fractions @xmath24 ) similarly ) .
we are able to make some comparison of our energy gaps to the experimental work of pan _
_ et al.__@xcite . throughout this paper
we assume complete spin - polarization of the electrons .
this should be a reasonable assumption for real experiments at sufficiently high magnetic fields .
jain s original proposal@xcite was to construct trial wavefunctions for fractional quantum hall states by writing @xmath25 \phi_0^{2 m } \right\}\ ] ] where the determinant is a slater determinant of noninteracting single fermion wavefunctions @xmath26 in effective magnetic field @xmath27 , and @xmath28 indicates projection to the lowest landau level . here
, @xmath29 is the wavefunction of a completely filled landau level , @xmath30 where @xmath31 is the dimensionless complex coordinate on the plane , @xmath32 and the usual gaussian factors @xmath33 are understood to be included in the measure of the hilbert space and will not be written explicitly for simplicity of notation . choosing a set of single particle wavefunctions @xmath26 to fill the @xmath2 lowest effective landau levels ( i.e. , such that @xmath34 $ ] represents the ground state of an _ integer _ quantum hall state @xmath35 ) , one obtains through eq .
[ eq : jain0 ] extremely good trial wavefunctions for fractional quantum hall states @xmath0 . as discussed above , the projection in eq .
[ eq : jain0 ] is exceedingly hard to implement for systems with more than roughly 10 electrons . for this reason ,
jain and kamilla@xcite looked for an essentially equivalent formulation that would be computationally simpler . in their approach
they begin by rewriting the wavefunction as @xmath36 \right\}\ ] ] where @xmath37 and then make the approximation that one can interchange the order of projection and taking the determinant to obtain a new trial wavefunction @xmath38 \phi_0^{2m}\ ] ] with @xmath39 although @xmath40 appears to be a single slater determinant , it is somewhat more complicated because each @xmath41 is actually a function of all of the particle positions through @xmath42 .
nonetheless , this new trial wavefunction is far simpler to evaluate numerically .
furthermore , extensive numerical work@xcite has shown that for small systems @xmath40 is just as good a trial state as @xmath43 and that both are extremely accurate we might consider @xmath44 with @xmath45 ( and @xmath46 ) . in the cases we have checked we have found that wavefunctions built with this version of @xmath47 are also extremely similar to those built from eq .
[ eq : tildepsi1 ] . ] . in the original work by jain and kamilla ,
it was shown how to calculate @xmath47 on a sphere for the case when @xmath48 has the same sign as the magnetic field @xmath8
( i.e. , @xmath16 ) . in appendix
[ app : jk ] we repeat the derivation for the case where @xmath48 has the opposite sign from @xmath8 ( i.e. , @xmath5 ) .
a discussion is also given there of the relative computational effort required to perform the relevant computations numerically .
this technique allows one to also obtain low energy spectra of these fractional quantum hall states , by similarly composite - fermionizing low energy excited states of noninteracting fermions as discussed in ref . .
in this paper we will perform all calculations using a spherical geometry@xcite with a monopole of charge @xmath49 flux quanta at the center .
the composite fermions then see an effective flux @xmath50 . in the presence of this effective flux ,
single particle states are described by two quantum numbers , @xmath51 and @xmath1 . here
@xmath52 is the angular momentum with @xmath53 corresponding to the landau level number " or
shell " index , and @xmath1 is the @xmath54-component of angular momentum . a state with @xmath2 filled composite fermion landau levels corresponds to @xmath55 . a low energy exciton is now formed by taking a composite fermion out of the highest occupied shell ( or landau level ) @xmath56 shell with some @xmath57 . choosing to work with states of zero total @xmath54-angular momentum , we take the state with @xmath58 and write this state as @xmath59 . using vector coupling ( clebsh - gordon ) coefficients@xcite we can construct exciton eigenstates of angular momentum @xmath60 as@xcite @xmath61 which serve as extremely accurate trial wavefunctions for the low energy excited states of the above discussed composite fermion ground states .
@xmath49 & @xmath2&@xmath62&@xmath63 & & & & @xmath64 & @xmath62&@xmath63 & & & + 9&-2&-1&6&-0.5391(1)&0.0906(9)&0.118(1)&1&3&4&-0.53949(2)&0.0928(2)&0.118(2 ) + 12&&-2&8&-0.5338(1)&0.091(1)&0.115(1)&&4&5&-0.53412(2)&0.0931(2)&0.1124(4 ) + 15&&-3&10&-0.5303(1)&0.083(2)&0.109(2)&&5&6&-0.53090(2)&0.0838(3)&0.1079(4 ) + 18&&-4&12&-0.5282(1)&0.080(2)&0.104(3)&&6&7&-0.52873(2)&0.0801(4)&0.1040(5 ) + 21&&-5&14&-0.5266(1)&0.078(2)&0.101(2)&&7&8&-0.52721(2)&0.0824(4)&0.1021(4 ) + 24&&-6&16&-0.5257(1)&0.079(5)&0.100(3)&&8&9&-0.52607(2)&0.0768(5)&0.1013(6 ) + @xmath65&&-@xmath65&@xmath65&-0.5173(1)&0.070(3)&0.089(3)&&@xmath65&@xmath65&-0.51803(3)&0.069(4)&0.0907(9 ) + 21&-3&-1&12&-0.4985(1)&0.061(2)&0.067(2)&2&3&10&-0.49870(5)&0.061(1)&0.069(1 ) + 26&&-2&15&-0.4980(1)&0.056(3)&0.063(3)&&4&12&-0.49826(5)&0.054(1)&0.066(1 ) + 31&&-3&18&-0.4979(1)&0.061(3)&0.067(3)&&5&14&-0.49803(5)&0.053(1)&0.063(1 ) + 36&&-4&21&-0.4979(1)&0.056(3)&0.069(3)&&6&16&-0.49804(5)&0.053(1)&0.065(1 ) + 41&&-5&24&-0.4976(1)&0.048(4)&0.058(4)&&7&18&-0.49797(6)&0.050(2)&0.064(2 ) + @xmath65&&-@xmath65&@xmath65&-0.4968(2)&0.042(8)&0.060(8)&&@xmath65&@xmath65&-0.4972(2)&0.041(3)&0.058(2 ) + 37&-4&-1&20&-0.4847(1)&0.052(3)&0.052(3)&3&3&18&-0.48467(5)&0.051(1)&0.053(2 ) + 44&&-2&24&-0.4851(1)&0.050(3)&0.050(3)&&4&21&-0.48525(7)&0.041(2)&0.052(2 ) + 51&&-3&28&-0.4855(1)&0.042(4)&0.048(4)&&5&24&-0.48564(8)&0.041(3)&0.048(3 ) + 58&&-4&32&-0.4857(1)&0.036(4)&0.043(4)&&6&27&-0.48593(6)&0.041(3)&0.046(3 ) + @xmath65&&-@xmath65&@xmath65&-0.4875(1)&0.014(7)&0.030(6)&&@xmath65&@xmath65&-0.48802(4)&0.022(8)&0.036(3 ) +
using particle - hole symmetry of the lowest landau level , one can exactly map states at filling fraction @xmath67 into states at filling fraction @xmath68 ( so long as we maintain complete spin polarization ) . as mentioned above , in this paper
we perform all calculations using a spherical geometry@xcite . on the sphere
, the lowest landau level has @xmath69 single particle eigenstates with @xmath49 the total number of flux quanta penetrating the sphere . thus states with @xmath70 electrons can be precisely mapped to their particle - hole conjugate states with @xmath71 electrons ( i.e. with @xmath70 holes ) . in appendix
[ app : phconj ] we show that , on a sphere , given an eigenstate @xmath72 with @xmath70 electrons and energy @xmath73 one can write the energy @xmath74 of its particle - hole conjugate wavefunction @xmath75 as @xmath76 where @xmath77 is the energy of the completely filled landau level .
this , of course , implies that the excitation spectrum of any given state is precisely the same as the excitation spectrum of its particle - hole conjugate state .
we will now focus on @xmath20cf states of the form @xmath21 .
the state with @xmath2 is particle - hole conjugate of the state with @xmath78 .
( for example , @xmath79 which is @xmath80 is conjugate of @xmath81 which is @xmath82 )
. extensive numerical work has already been performed for positive @xmath2 , calculating accurate ground state energies and energy gaps@xcite .
[ eq : pheq ] this means that we already know the ground state energies and energy gaps for negative @xmath2 . here
, however , we calculate these quantities directly using our negative @xmath2 trial wavefunctions and compare to the particle hole conjugated results to establish the validity of our approach .
table [ tab : results_m1 ] summarizes the numerical results for the ground state energies of states in this series calculated using monte - carlo@xcite . by separating it into a slater - determinant of the projected pseudo single - particle - wavefunctions and a jastrow - factor , tends to produce numerical instabilities in the determinant algorithm
this is easily understood , since two particles approaching each other at a distance @xmath83 will cause the respective @xmath84 to grow as @xmath85 in the same manner , producing two linearly dependant columns in the matrix . now , due to numerical errors , instead of obtaining zero when two columns become the same , the evaluation yields numerical errors following approximately the same power law behavior , that will dominate the result . since the short distance behavior of the wavefunction is known , extrapolation allows us to correct for such events that occur with a probability of less than @xmath86 in all of the presented calculations . ] quantum hall states that are particle - hole conjugate of each other , are expected to have the same excitation spectrum .
the low energy spectra shown here for for @xmath87 have been calculated with the method of jain and kamilla@xcite .
the spectra for the particle - hole conjugate states with @xmath88 are calculated using the new negative effective flux trial wavefunctions described in this paper .
if the calculations were exact ( rather than just approximate ) the corresponding spectra of the particle - hole conjugate pairs would match exactly .
although our calculations , being based on approximate trial wavefunctions for @xmath20cfs , are not exact , we still see remarkably good agreement , suggesting that our new trial wavefunctions are approximately as accurate as the previously described trial wavefunctions for @xmath87 .
the spectra shown here correspond ( as marked ) to particle hole conjugate pairs at filling fractions , from top to bottom @xmath89 , @xmath90 and @xmath91 corresponding to flux @xmath92 respectively .
note that our new trial wavefunctions show an excellent reproduction of the increasingly nontrivial features as @xmath93 is approached . in the top panel
the data points have been shifted slightly horizontally from integer angular momentum values for better distinguishability . ]
the ground - state energies we obtain using negative @xmath2 trial wavefunctions show an outstanding agreement with the values obtained by particle hole conjugating positive @xmath2 trial wavefunctions .
we observed slight differences on the fourth significant digit , that show that the trial state with composite fermions in positive effective flux is very slightly better than the one with negative effective flux introduced here .
the excellent agreement of our negative @xmath2 wavefunctions with particle - hole conjugation of positve @xmath2 wavefunctions extends to the excited states , generated as outlined in section [ sec : jk ] . as examples , figure [ fig : spectra_m1 ] shows excitation spectra for @xmath79 , @xmath94 and @xmath95 and their respective particle - hole conjugate states .
we also give , in table [ tab : results_m1 ] values for magnetoroton gaps ( which are the lowest energy neutral excitations ) as well as large @xmath96 gaps ( which are presumably the transport gap ) .
these results , along with the excellent results for the ground - state energies confirm the validity of our approach to calculating cf wavefunctions at negative effective flux , which enables us to consider in the following section filling fractions above 1/4 in the series @xmath97 , that were previously inaccessible .
@xmath98&@xmath2&@xmath99&@xmath100 & & & & @xmath64 & @xmath99&@xmath100 & & & + 4&-1&-3&9&-0.47481(3)&0.0929(3)&0.1241(3)&1&3&15&-0.37706(1)&0.0169(1)&0.0305(1 ) + 5&&-4&12&-0.45940(3)&0.0939(4)&0.1224(4)&&4&20&-0.36501(1)&0.0203(1)&0.0276(1 ) + 6&&-5&15&-0.44996(3)&0.0837(4)&0.1183(4)&&5&25&-0.35800(1)&0.0132(1)&0.0251(1 ) + 7&&-6&18&-0.44355(3)&0.0812(3)&0.1135(4)&&6&30&-0.35303(1)&0.0147(2)&0.0241(2 ) + 8&&-7&21&-0.43895(3)&0.0830(4)&0.1110(5)&&7&35&-0.34955(1)&0.0151(2)&0.0243(2 ) + 9&&-8&24&-0.43540(4)&0.0781(5)&0.1083(5)&&8&40&-0.34686(2)&0.0124(3)&0.0245(3 ) + 20&&@xmath101&@xmath101&-&-&-&&19&95&-0.33577(2)&0.0109(5)&0.0242(5 ) + -@xmath65&&@xmath65&@xmath65&-0.40999(4)&0.0668(5)&0.097(3)&&@xmath65&@xmath65&-0.32748(8)&0.009(2)&0.021(1 ) + 6&-2&-1&19&-0.40432(3)&0.0243(2)&0.0243(2)&2&1&21&-0.38699(1)&0.0182(2)&0.0182(2 ) + 8&&-2&26&-0.39826(3)&0.0231(3)&0.0257(3)&&2&30&-0.37411(1)&0.0186(2)&0.0193(2 ) + 10&&-3&33&-0.39484(3)&0.0190(3)&0.0283(3)&&3&39&-0.36726(1)&0.0142(4)&0.0220(3 ) + 12&&-4&40&-0.39257(3)&0.0207(4)&0.0288(4)&&4&48&-0.36276(2)&0.0128(3)&0.0219(4 ) + 14&&-5&47&-0.39092(3)&0.0179(5)&0.0275(5)&&5&57&-0.35969(2)&0.0125(4)&0.0208(5 ) + 16&&-6&54&-0.38976(3)&0.0203(4)&0.0268(5)&&6&66&-0.35742(2)&0.0129(5)&0.0183(5 ) + 18&&-7&61&-0.38885(3)&0.0176(8)&0.0235(9)&&7&75&-0.35569(2)&0.0108(6)&0.0179(4 ) + @xmath65&&-@xmath65&@xmath65&-0.38163(3)&0.0151(1)&0.0276(2)&&@xmath65&@xmath65&-0.34277(5)&0.0077(13)&0.020(2 ) + 12&-3&-1&43&-0.38486(3)&0.0152(5)&0.0152(5)&3&1&45&-0.37306(2)&0.0132(4)&0.0132(4 ) + 15&&-2&54&-0.38266(3)&0.0154(7)&0.0169(7)&&2&58&-0.36765(3)&0.0137(6)&0.0137(6 ) + 18&&-3&65&-0.38126(3)&0.0140(8)&0.0190(8)&&3&71&-0.36427(3)&0.0119(7)&0.0171(6 ) + 21&&-4&76&-0.38017(3)&0.0126(8)&0.0192(10)&&4&84&-0.36182(2)&0.0120(8)&0.0175(8 ) + 24&&-5&87&-0.37940(3)&0.0126(10)&0.0196(9)&&5&97&-0.36006(4)&0.012(1)&0.017(1 ) + 27&&-6&98&-0.37883(3)&0.0134(14)&0.0195(14)&&6&110&-0.35869(2)&0.010(1)&0.016(1 ) + @xmath65&&-@xmath65&@xmath65&-0.37404(3)&0.010(1)&0.0238(8)&&@xmath65&@xmath65&-0.34842(4)&0.0088(12)&0.021(2 ) + 20&-4&-1&75&-0.37206(3)&0.0123(6)&0.0123(6)&4&1&77&-0.36788(3)&0.0109(7)&0.0109(7 ) + 24&&-2&90&-0.37172(4)&0.010(1)&0.0117(10)&&2&94&-0.36491(3)&0.0118(9)&0.013(1 ) + 28&&-3&105&-0.37165(4)&0.010(1)&0.0148(7)&&3&111&-0.36285(3)&0.011(1)&0.012(1 ) + 32&&-4&120&-0.37142(3)&0.010(1)&0.013(1)&&4&128&-0.36133(3)&0.0093(9)&0.014(1 ) + 36&&-5&135&-0.37138(2)&0.012(1)&0.016(1)&&5&145&-0.36021(3)&0.011(1)&0.014(2 ) + @xmath65&&-@xmath65&@xmath65&-0.3705(1)&0.010(3)&0.019(3)&&@xmath65&@xmath65&-0.35129(5)&0.0087(19)&0.017(2 ) +
the @xmath22cf series of composite fermion states , corresponding to filling fractions around @xmath102 , has been the subject of some recent experimental work@xcite , yet the branch of filling fractions above 1/4 has been mostly inaccessible to numerical investigations .
the numerical approach we took for examining these states permits us to calculate the excitation spectra for systems with a moderate number of effective magnetic flux quanta @xmath103 for the cf - system . for negative @xmath2 ,
the calculational complexity of the wavefunction increases with @xmath104 , consequently the achievable system size is reduced compared to the systems with positive @xmath2 . for this very reason , extrapolation of the results to infinite system size in order to obtain the gap in the thermodynamic limit , and thus the mass of the cf , is difficult for these states .
nonetheless , the data obtained so far reveals that the spectra of states with @xmath2 and @xmath105 at the same system size appear to be more and more similar as @xmath19 increases and @xmath102 is approached .
table [ tab : table_m2 ] summarizes the numerical results for the groundstate energies and gaps of states around @xmath102 , and figure [ fig : spectra_m2 ] shows several examples of excitation spectra revealing the above mentioned similarity .
the method developed here permits us to calculate the dispersion curves for the low energy excitations at filling fractions @xmath106 for negative @xmath2 , i.e. above a quarter filling , that were previously inaccessible . a comparison to the spectra of states below and above @xmath102 shows , that the excitation spectra for @xmath2 and @xmath105 show very similar features and seem to become more similar as @xmath2
is increased .
as above , the @xmath16 spectra are calculated using the method of jain and kamilla@xcite whereas the @xmath18 spectra are calculated using the method discussed in the current paper . ]
this figure illustrates the extrapolation of the gaps to the thermodynamic limit for different filling fractions in the series @xmath106 by means of a simple linear regression of the available data points over the reciprocal particle number .
states with @xmath2 and @xmath105 are displayed together .
the state at @xmath79 can be obtained in two different manners , i.e. as @xmath22cf in negative flux or as @xmath20cf in positive flux , which accounts for the two additional sets of data in the upper left .
the magnetoroton gaps of these different @xmath79 states are close to indistinguishable , whereas the large @xmath96 gap is slightly bigger for negative effective flux .
as discussed in the text , extrapolation is least certain for @xmath107 due to finite size effects . ]
masses of the composite fermions near @xmath102 and extrapolated large @xmath96 gaps from which they were deduced at filling fractions of the series @xmath97 .
data from ref . is from larger system size calculations and should be considered to be more accurate . comparing these more accurate results to our smaller system calculations gives us an estimate of the finite size errors of our results which presumably will hold even in the @xmath18 case . except for the marked asymmetry between @xmath79 and @xmath108
we do not see much sign of an asymmetry around @xmath102 which is observed experimentally@xcite .
we also do not see signs of the experimentally observed diverging effective mass as @xmath102 is approached .
our values of the effective mass are somewhat less than that obseved experimentally in general .
however , it is known that finite well width corrections tend to increase the effective masses@xcite .
two additional data points are shown in this figure ( not mentioned in the legend ) which give results obtained by treating @xmath79 as a @xmath20cf state . ]
it is interesting to note that the @xmath109 similarity in spectra found here differs from the particle - hole symmetry correspondence of @xmath2 with the @xmath110 states in the @xmath21 series shown above in section [ sec : phsym ] .
one might naively expect that the particle - hole symmetry @xmath111 for states in the @xmath20cf series @xmath21 above survives the composite fermionization attachment of two more jastrow factors and results in an approximate symmetry for @xmath97 states .
for example , we could take the symmetry related @xmath3 pair ( @xmath112 ) and ( @xmath113 ) at @xmath114 on table [ tab : results_m1 ] and add two jastrow factors so @xmath115 resulting in an approximate symmetry relating ( @xmath116 ) to ( @xmath117 ) in table [ tab : table_m2 ] .
however , when we examine the overall shape of the resulting excitation spectra , we find very little relation between the two spectra related by @xmath118 for @xmath22cfs .
in contrast we see in figure [ fig : spectra_m2 ] that the @xmath109 related states have quite similar dispersions , and the similarity appears to increase with increasing @xmath19 .
we comment that another interesting test of our approach can be obtained by constructing a trial wavefunction for @xmath79 as the @xmath119 member of the @xmath97 series .
we find that the ground state energy of this wavefunction is , within numerical precision , precisely the same as that of the laughlin @xmath79 trial wavefunction , which leads us to believe that we have precisely constructed that state .
similarly , we can examine the excitation spectrum of @xmath79 by using a @xmath20cf wavefunction of the series @xmath120 with @xmath80 or by using a @xmath22cf wavefunction of the series @xmath106 with @xmath119 .
we find that the spectra obtained in these two approaches are quite similar to each other ( albeit not quite identical ) , which gives us still further confidence in our approach .
the values of the magneto - roton gap appear to be almost exactly the same in both cases . yet , examining the large @xmath96 gap , the @xmath22cf trial wavefunction using negative flux ( @xmath119 ) yields a slightly larger value , which decreases slightly more rapidly with the system size though , and seems to extrapolate to almost the same value at infinite @xmath70 ( see figure [ fig : gap_thermo ] ) .
figure [ fig : gap_thermo ] shows the extrapolation to the thermodynamic limit of the data given in table [ tab : table_m2 ] , comparing gaps at @xmath2 with those of states at @xmath105 .
a comparison of the excitations at @xmath79 considered in the two different manners discussed above is also displayed there .
further , this figure gives us an idea of the quality of our extrapolation .
where the extrapolation is not particularly smooth , we can not claim to deliver more than a rough result .
the extrapolation of @xmath107 appears to be the most difficult , since it is not easy to distinguish a clear linear dependence of the large @xmath96 gap as a function of @xmath121 for the initial data set .
the reason for this problem appears to be that the magnetoroton gap is located at a large value of @xmath60 , so that in the smaller systems , the magnetoroton gap is located very close to the largest @xmath96 available , or even coincides with this point . for the @xmath122 case
, we can see the error in this extrapolation clearly by comparing our extrapolated result to a result of a similar calculation using larger system sizes from ref . .
this comparison is show in figure [ fig:4cfmasses ] ( see below ) .
we might guess that the error in extrapolation for @xmath82 is of similar magnitude . in order to obtain the composite fermion effective mass
, we equate the activation gap @xmath123 ( determined from the excitation energy at the maximum angular momentum , i.e. biggest particle - hole separation ) to the cyclotron energy of cfs in their effective magnetic field : @xmath124 since the gap is measured in units of the coulomb interaction , we write @xmath125 .
further taking into account @xmath126 for gaas and the free electron mass as our point of reference , we find the dimensionless normalized effective mass @xmath127})$ ] to be given by @xmath128 .
the @xmath22cf - masses we obtain are displayed in figure [ fig:4cfmasses ] together with the large @xmath96 gaps from which they are deduced .
in addition , for @xmath129 ( positive @xmath2 ) we have shown data from ref .
where larger systems were used than we have used here ( when we also use larger system sizes , our results agree very will with those of ref . ) . for @xmath79 , as discussed above , we have shown results that treat this either as a @xmath119 point of the @xmath106 series or a @xmath80 point of the @xmath120 series . for the latter case , we have likewise included the result from the above mentioned larger system calculation . in figure [ fig:4cfmasses ]
, we have intentionally displayed data extrapolated from a restricted set of small systems sizes with positive @xmath2 so as to match the same set of system sizes that we study for negative @xmath2 where we can not go to very large systems .
one can estimate the finite size error for our negative @xmath2 calculations by examining the deviations between these restricted extrapolations at positive @xmath2 compared to the larger system calculations of ref .
( also displayed in our figure ) .
the behavior of the negative @xmath2 fractions seems to roughly mirror the behavior of the positive @xmath2 fractions . as was seen previously in ref .
for @xmath4 it is seen that the effective mass increases with @xmath130 .
as @xmath102 is approached from either side , it is not clear if the effective mass will converge to a constant as would be predicted by theory ( up to logarithmic corrections@xcite ) . in the experiments of ref . , a striking asymmetry of the effective mass beween the high field and low field sides of @xmath102 has been observed .
while we can not rule out some asymmetry from our data , we certainly can not claim to see the extremely strong differences that are observed experimentally .
this , however , is not surprising .
experimentally , the asymmetry is attributed to the proximity of a wigner crystal state@xcite .
since we are using a trial wavefunction approach , we should not see the effects of any imminent phase transition .
perhaps the most interesting data point in figure [ fig:4cfmasses ] is the one at @xmath79 . whether we treat this point as the @xmath119 member of the series @xmath106 or the @xmath80 member of the series @xmath120 , we find almost identical results of a very large gap , which establishes a continuity between the cf masses around @xmath93 and @xmath102 .
furthermore , we note that this point is quite asymmetric with its reflection at @xmath108 .
certainly the hypothesis of constant effective composite fermion mass does not extend all the way from @xmath102 out to both @xmath108 and @xmath79 .
below quarter filling , one observes a continuous increase of this mass .
a similar trend appears at small values of the effective magnetic field above this point , but then the mass drops down again at @xmath79 , the final point of this series . generally , our values for the effective mass seem to be lower than those measured in the experiments of ref . by a factor of roughly @xmath131 .
this error is rather expected , since similar discrepancies have been observed in previous studies based on the composite fermion picture@xcite .
it is known , however , that taking into account the finite width of the 2d electron gas changes the interaction so as to increase the effective mass@xcite .
we note that we are not able to find any evidence of the divergence of @xmath132 as we approach @xmath102 from either side , which is observed experimentally in ref . .
this is not surprising for several reasons .
first of all , the experiment only sees strong divergences extremely close to @xmath102 which we can not access numerically .
furthermore , one might suspect that the disorder might be the source of the divergences in the measurements used by pan .
more importantly , however , even if there were genuine infra - red divergences@xcite of the effective mass as @xmath102 is approached , one would not necessarily expect such divergences to be properly represented in a trial wavefunction approach .
we emphasize that the most important achievement of this paper is not any particular numerical result .
if genuine numbers were desired for comparison to experiment , we would want to use a more realistic interaction , accounting for finite well width@xcite , as well as perhaps landau - level mixing , and we would want to use a more powerful computer to analyze ever larger systems . instead we would like to emphasize in this paper that we have clearly demonstrated that we can extend the approach of jain and kamilla@xcite to treat negative effective magnetic field , and we can study these negative @xmath2 composite fermion wavefunctions for reasonably large systems , which has not been done before . in this paper
we have tested this method by using particle - hole symmetry for the @xmath20cf series and we have found our method to be quite accurate . we have then applied this method to @xmath22cfs to study , for the first time , large systems for filling fractions @xmath106 with negative @xmath2 .
our main physical result is that the effective mass appears to be roughly symmetric around and close to @xmath102 although larger system calculations would be desirable .
the authors acknowledge helpful conversations with e. h. rezayi .
g.m . acknowledges support from the french ministry of science , and thanks both the ecole doctorale de paris and lucent technologies for their support that made participation in this research project possible .
the starting point for the composite fermion trial wavefunctions@xcite are the single particle eigenfunctions of the quantum mechanical problem of a particle in a magnetic monopole field on a sphere .
( we use spherical coordinates with the azimuth @xmath133 ranging from @xmath134 to @xmath135 , and @xmath136 the longitude ranging from @xmath134 to @xmath137 ) .
the monopole harmonics are given by@xcite @xmath138 with @xmath139 , @xmath140 , @xmath141 , @xmath142 , @xmath143 and @xmath144 are the jacobi polynomials .
this monopole harmonic represents an eigenstate of a particle on a sphere in a radial magnetic field with @xmath145 flux quanta penetrating the sphere , where a positive sign refers to outwards pointing flux . here
, the angular momentum of the eigenstate is @xmath146 and the @xmath54 component of the angular momentum @xmath147 .
further , @xmath11 is the ll index @xmath148 .
the above expression assumes the haldane gauge@xcite , where the singularities of the vector potential are chosen to be located symmetrically on both north- and southpole of the sphere . here
, we focus on the case where @xmath149 , since the @xmath150 case has already been discussed in detail in ref . .
for the rest of this paragraph , we thus assume @xmath149 .
expanding the jacobi polynomials in terms of the spinor coordinates @xmath151 and @xmath152 , one obtains @xmath153 this expression can equally well be obtained from the relation for complex conjugation of the monopole harmonics@xcite , if one corrects kamilla s formula by replacing @xmath104 by @xmath154 in the appropriate places .
we now use the @xmath155 as single particle wavefunctions ( written as @xmath26 in the main text ) and composite fermionize by attaching jastrow factors . as discussed in the main text
, we can construct the many particle composite fermion trial wavefunctions by bringing the jastrow factors inside the slater determinant of single particle states , which may then be projected individually .
note , that in the spherical geometry , the jastrow factor becomes @xmath156 the details of the required projection @xmath157 are discussed next .
first , we remark that the jastrow factor , according to kamilla s notation , whereas @xmath1 , that had been used for this purpose in section [ sec : jk ] to comply with the common notation for the filling fractions of jain s series , is already used for the eigenvalue of @xmath158 here . ]
@xmath159 is a lll function , with @xmath160 zeros in @xmath161 , i.e. it is a lll wavefunction for flux @xmath162 .
since @xmath70 is in general a big number , we have @xmath163 .
the resulting wavefunction , in turn , has to be a valid wavefunction for a total number of flux @xmath164 . secondly ,
since projection is a linear operation , we may consider the action of projection on each of the basis states @xmath165 separately , by expanding @xmath159 in this basis .
in general multiplication by a basis state @xmath166 followed by projection can be described as a linear operator called hereafter @xmath167 .
@xmath168 since we know the entire basis of the subspace that we project upon , namely the lll for flux @xmath169 with states @xmath170 , and @xmath171 , the projection operator is @xmath172 .
we now show , how this leads to an expression for @xmath167 as a differential operator in the coordinate representation , in which ( [ projection_of_basis_states ] ) becomes @xmath173 integration over the longitudinal angle @xmath136 singles out one nonzero scalar product for @xmath174 , and the one remaining integral over the azimuthal angle @xmath133 yields a well known binomial coefficient . simplifying the normalization factors of @xmath165 on both sides , we have : @xmath175 using the explicit form of the normalization @xmath176 and remarking that the fractions of factorials that are left in this expression equal those that appear by the multiple derivation of a monomial @xmath177 @xmath178 we may deduce @xmath167 by comparison of both sides : @xmath179 let us remark that this result reproduces the known result , that the projection on the lll is achieved by performing the habitual procedure of moving all @xmath180 s and @xmath181 s to the far left , and replacing them with derivatives according to @xmath182 nevertheless , performing this explicit projection gives us a supplementary information in the form of a weight factor @xmath183 for the different landau levels before projection , which of course does not matter for the problems discussed here , but may play a role in other cases@xcite . practically , we would like to obtain a form of ( [ eq : projop ] ) with the derivatives moved to the extreme right , which may be calculated using a straightforward application of the leibniz rule for multiple derivatives in both @xmath184 and @xmath185 : @xmath186 this yields a triple sum with the inner summation ranges being dependant on the outer summation index @xmath187 .
one finds that the summation ranges can be made independent of @xmath187 since the summand becomes zero outside of the given intervals .
as such the sum over @xmath187 may be evaluated using @xmath188 since the result yields a kronecker delta , one of the remaining sums becomes trivial , and after shifting the remaining summation index , the final result is revealed to be exactly like ( [ eq : projop ] ) , but with all derivatives placed at the very right . the projected composite fermion wavefunction is nothing but this operator applied to the single particle jastrow factor : @xmath189 in order to perform numerical calculations with this wavefunction , we need to evaluate the derivatives explicitly .
one may use jain and kamilla s approach@xcite , to commute the derivatives through the jastrow factors as @xmath190 \intertext{with } & \hat u_i = j_i^{-p } \frac{\partial}{\partial u_i}j_i^p,\text { and } \,\hat v_i = j_i^{-p } \frac{\partial}{\partial v_i}j_i^p\end{aligned}\ ] ] constructing a many - particle wavefunction out of these projected cf wavefunctions in the form of a slater determinant , one may factor out the jastrow factors again , and thus obtains a form which resembles single particle wavefunctions , on a basis of projected states @xmath191 with : @xmath192\end{aligned}\ ] ] of course , this only appears to be a one particle wavefunction , since there is an implicit dependence of the positions of all other electrons in the system hidden in the operators @xmath193 and @xmath194 .
the complexity of this expression increases with the total number of derivatives per term , given by @xmath195 for negative @xmath104 compared to @xmath196 for positive effective flux .
in the lowest landau level , on the sphere , there are @xmath197 single particle eigenstates where @xmath100 is the flux through the sphere .
we label these eigenstates by the @xmath54-component of their angular momentum @xmath1 .
for a two - body interaction we can write the hamiltonian as @xmath198 as usual , the normal ordering of the operators accounts for the uniform positive background . as usual , the fermion operators have anticommutation relations @xmath199
. the interaction matrix @xmath200 must have the following symmetries @xmath201 furthermore , for any rotationally ( translationally ) invariant interaction , we must have angular momentum conservation , which implies that the matrix element is zero unless @xmath202 we define the vacuum state @xmath203 , to be the state which contains no electrons at all . the filled landau level , which we can also think of as the vacuum for holes is written as @xmath204 for holes , given by @xmath205 which also obey the usual anticommutations @xmath206 .
we now rewrite the hamiltonian in terms of these hole operators . using the commutation relations as well as above
described symmetries of @xmath200 we obtain @xmath207 where @xmath208 and @xmath209 we show below that for any rotationally invariant interaction , @xmath210 is actually independent of @xmath1 .
furthermore , it is very easy to show that the energy of the entirely filled landau level is given by @xmath211 thus we have @xmath212 and @xmath213 is the number of holes ( i.e. , the eigenvalue of @xmath214 ) . a general state containing @xmath70 electrons in the lll can be written as : @xmath215 electrons .
note that both of these states live " in the same lowest landau level which has @xmath216 single particle eigenstates ( indexed by @xmath1 ) .
now , if @xmath72 is an eigenstate of @xmath217 with eigenvalue @xmath218 ( assumed to be real ) , then , since @xmath219 has precisely the same structure as @xmath217 we see that @xmath75 is an eigenstate of @xmath219 with the same eigenvalue .
thus , we obtain the particle hole conjugation relation @xmath220 where the @xmath223 s are rotation matrices as in ref . and @xmath224 . setting @xmath225 , and @xmath226 and summing over @xmath227 as prescribed in eq .
[ eq : udef ] we obtain @xmath228^ * \label{eq : ures}\ ] ] where we have used the orthogonality@xcite @xmath229^ * = \delta_{m_2',m_3'}\ ] ] as well eq . [ eq : angcons ] .
[ eq : ures ] must be true for any rotation , we can integrate over all rotations and use@xcite @xmath230^ * = \mbox{constant}\ ] ] independent of @xmath1 , which shows that @xmath210 is independent of @xmath1 .
r. k. kamilla , j. k. jain , and s. m. girvin phys .
b * 56 * , 12411 - 12416 ( 1997 ) ; k. park and j. k. jain , phys . rev
* 80 * , 4237 - 4240 ( 1998 ) ; k. park , v. melik - alaverdian , n. e. bonesteel , and j. k. jain phys .
b * 58 * , r10167-r10170 ( 1998 ) ; k. park and j. k. jain phys . rev . lett . * 83 * , 5543 - 5546 ( 1999 ) ; v. w. scarola , k. park , and j. k. jain phys .
b * 61 * , 13064 - 13072 ( 2000 ) ; k. park and j. k. jain phys . rev
. lett . * 84 * , 5576 - 5579 ( 2000 ) ; t. sbeouelji , k. park , j. k. jain , and n. meskini phys . rev .
b * 62 * , r4802-r4804 ( 2000 ) ; k. park and j. k. jain phys . rev .
b*62 * , r13274-r13277 ( 2000 ) ; v. w. scarola , k. park , and j. k. jain phys .
b * 62 * , r16259-r16262 ( 2000 ) ; s. s. mandel and j. k. jain , phys . rev .
b * 63 * , 201310 ( 2001 ) ; phys . rev .
b * 64 * , 081302 ( 2001 ) ; phys . rev .
b * 64 * , 125310 ( 2001 ) ; phys . rev . lett . * 89 * , 096801 ( 2002 ) ; t. sbeouelji and n. meskini , phys . rev .
b * 64 * 193305 ( 2001 ) .
k. park , n. meskini , and j. k. jain , j. phys .
cond matt * 11 * , 7283 ( 1999 ) ; v. w. scarola , s. y. lee , and j. k. jain , phys .
rev . b*66 * , 155320 ( 2002 ) ; k. park and j. k. jain phys .
lett . * 81 * , 4200 - 4203 ( 1998 ) . | the method of jain and kamilla [ prb * 55 * , r4895 ( 1997 ) ] allows numerical generation of composite fermion trial wavefunctions for large numbers of electrons in high magnetic fields at filling fractions of the form @xmath0 with @xmath1 and @xmath2 positive integers . in the current paper
we generalize this method to the case where the composite fermions are in an effective ( mean ) field with opposite sign from the actual physical field , i.e. when @xmath2 is negative .
we examine both the ground state energies and the low energy neutral excitation spectra of these states . using particle - hole symmetry
we can confirm the correctness of our method by comparing results for the series @xmath3 with @xmath4 ( previously calculated by others ) to our results for the conjugate series @xmath3 with @xmath5 .
finally , we present similar results for ground state energies and low energy neutral excitations for the states with @xmath6 and @xmath5 which were not previously addressable , comparing our results to the @xmath3 case and the @xmath7 , @xmath6 cases . |
the cern programme in experimental kaon physics of the last decade has been carried out by the na48 series of experiments .
na48 has accomplished several physics subprogrammes based on data taken with @xmath2 , @xmath3 , @xmath4 and neutral hyperon beams in 19972004 .
the principal components of the experimental setup ( modified and upgraded in the course of operation ) are a beam line followed by a vacuum decay volume , a magnetic spectrometer consisting of four drift chambers , a trigger scintillator hodoscope , a liquid krypton electromagnetic calorimeter , a hadron calorimeter , and and a muon detector @xcite .
the present paper reports a number recent precise measurements of cp violation ( cpv ) parameters : 1 ) the indirect cpv parameter @xmath0 with @xmath5 decays ; 2 ) the direct cp violating charge asymmetries of dalitz plot slopes @xmath6 in @xmath7 and @xmath8 decays .
[ eta ] the interest in a precise measurement of the indirect cpv parameter @xmath9 stems , in particular , from the fact that its recent measurements by ktev and kloe experiments published in 2004 and 2006 , respectively , differ by @xmath10 , or more than four standard deviations , from the previous world average . the na48 measurement of @xmath0 @xcite is based on a data set taken during two days of dedicated running in 1999 . the directly measured quantity is the ratio of the decay rates @xmath11 ; these decays are characterized by similar signatures involving two reconstructed tracks of charged particles . then @xmath0 is computed as @xmath12 in this approach the @xmath2 and @xmath3 lifetimes @xmath13 and @xmath14 , and the branching fractions @xmath15 and @xmath16 are external inputs taken from the best single measurements .
the data sample contains about @xmath17 2-track triggers .
event selection is similar for the @xmath5 and @xmath18 modes .
a crucial difference is electron vs pion identification based on the ratio of particle energy deposition in the em calorimeter to its momentum measured by the spectrometer ( expected to be close to 1 for electrons ) .
particle identification efficiencies were directly measured and corrected for .
samples of @xmath19 @xmath5 and @xmath20 @xmath21 candidates were selected , with about @xmath22 background contamination in each .
acceptance corrections and background subtraction were performed by monte carlo simulation .
trigger efficiencies were measured directly with the data and corrected for .
the most relevant systematic uncertainties come from precision of simulation of kaon momentum spectrum , precision of radiative corrections , and precision of trigger efficiency measurement .
the final result is @xmath23 this leads , subtracting the @xmath24 direct emission contribution , but retaining the inner bremsstrahlung contribution , to @xmath25 finally , the cp violation parameter is computed according to ( [ eta - main ] ) to be @xmath26 the result in in agreement with the recent kloe and ktev measurements , while it contradicts the 2004 pdg average .
the latter disagreement is understood to be due to the improved treatment of the radiative corrections in the recent analyses .
@xmath27 and @xmath8 decays are among the most promising processes in kaon physics to search for cpv phenomena .
the decay density is parameterized ( up to radiative and @xmath28 rescattering effects studied separately @xcite ) by a polynomial expansion @xmath29 where @xmath30 , @xmath31 , @xmath32 are the so called linear and quadratic dalitz plot slope parameters ( @xmath33 ) , and the two lorentz invariant kinematic variables @xmath34 and @xmath35 are defined as @xmath36 here @xmath37 is the charged pion mass , @xmath38 and @xmath39 are the kaon and pion four - momenta , the indices @xmath40 correspond to the two pions of the same electrical charge , and the index @xmath41 to the pion of different charge .
a non - zero difference @xmath42 between the slope parameters @xmath43 and @xmath44 describing the decays of @xmath45 and @xmath46 , respectively , is a manifestation of direct cp violation expressed by the corresponding slope asymmetry @xmath47 the above slope asymmetry is expected to be strongly enhanced with respect to the asymmetry of integrated decay rates .
a recent full next - to - leading order chpt computation @xcite predicts @xmath6 to be of the order of @xmath48 within the sm .
calculations involving processes beyond the sm @xcite allow a wider range of @xmath6 , including substantial enhancements up to a few @xmath49 .
a measurement of the quantity @xmath6 was performed with a record data sample collected in 200304 with simultaneous @xmath45 and @xmath46 beams @xcite .
the measurement method is based on the study of ratios of @xmath34 spectra of @xmath45 and @xmath46 decays , and exploits cancellations of major systematic effects due to the simultaneous collection of @xmath45 and @xmath46 decays , and regular inversions of magnetic fields in the beam line and the spectrometric magnet , which allows achieving @xmath50 precision .
the event samples are practically background - free , and contain @xmath51 @xmath27 candidates , and @xmath52 @xmath8 candidates ( the @xmath45/@xmath46 flux ratio , on which however the results do not depend , is 1.8 ) .
the cp violating charge asymmetries of the linear slope parameter of the dalitz plot of the @xmath27 and @xmath8 decays were measured to be @xmath53 the archived precision is more than an order of magnitude better that those of the previous measurements . the results do not show evidences for large enhancements due to non - sm physics , and can be used to constrain certain sm extensions predicting enhanced cpv effects .
a number of recent measurements of cpv parameters in kaon decays by the na48 collaboration at cern are presented .
the achieved precisions are similar to or better than the best previous ones .
00 v. fanti _
( na48 ) , .
et al_. ( na48 ) , phys . lett . *
b645 * ( 2007 ) 26 .
e. gmiz , j. prades , i. scimemi , jhep * 10 * ( 2003 ) 042 .
shabalin , itep preprint * 8 - 98 * ( 1998 ) .
g. dambrosio , g. isidori , g. martinelli , phys . lett .
* b480 * ( 2000 ) 164 .
( na48/2 ) , phys .
b649 * ( 2007 ) 349 .
( na48/2 ) , phys . lett .
* b633 * ( 2006 ) 173 .
et al_. ( na48/2 ) , eur .
j. * c52 * ( 2007 ) 875 . | recent precise measurements of cp violation parameters in kaon decays at the na48 experiment : indirect cpv parameter @xmath0 , and charge asymmetries in @xmath1 decays , are presented .
cp violation , kaon decays |
a large part of modern physics is devoted to nonlinear classical and quantum phenomena in various systems .
such effects as the generation of the second harmonic or optical rectification are known from classical physics , while quantum electron pumping through a small sample due to interference of wave functions is a quantum nonlinear effect .
experiments on nonlinear electrical transport often combine classical and quantum contributions .
a macroscopic sample without inversion center @xcite exhibits a current - voltage characteristic which with increasing voltage departs from linearity due to terms proportional to the square of the applied voltage .
if now an oscillating ( ac ) voltage is applied , a zero - frequency current ( dc ) is generated .
if the sample is sufficiently small , quantum effects can appear due to the wave nature of electrons .
the uncontrollable distribution of impurities or small variations in the shape of the sample result in quantum contributions to the dc which are random . for a mesoscopic conductor with terminals
@xmath0 we can describe the quadratic current response in terms of second order conductances @xmath1 .
they relate voltages @xmath2 applied at contacts or neighboring gates @xmath3 at frequency @xmath4 to the current at zero frequency at contact @xmath5 , @xmath6 the second order conductances include in detail the role of the shape and the nearby conductors ( gates ) .
they depend on external parameters like the frequency of the perturbation , temperature , magnetic field or the connection of the sample to the environment .
we concentrate here on the quantum properties of nonlinear conductance through coherent chaotic samples .
chaos could result from the presence of impurities ( disorder ) or random scattering at the boundaries ( ballistic billiard ) .
due to electronic interference the sign of this effect is generically random even for samples of macroscopically similar shape .
@xcite when averaged over an ensemble , the second order conductances vanish . as a consequence , for a fully chaotic sample
there is no classical contribution to the dc and the nonlinear response is the result of the sample - specific quantum fluctuations .
interestingly enough , from a fundamental point of view these fluctuations of nonlinear conductance are sensitive to the presence of coulomb interactions and magnetic field .
while interactions strongly affect the fluctuations amplitude , their sign is easily changed by a small variation of magnetic flux @xmath7 , similarly to universal conductance fluctuations ( ucf ) in linear transport .
more importantly , without interactions the current ( [ eq : iv ] ) through a two - terminal sample is a symmetric function of magnetic field , just like linear conductance .
however , the idea that coulomb interactions are responsible for magnetic - field asymmetry in nonlinear current was recently proposed theoretically @xcite and demonstrated experimentally in different mesoscopic systems .
@xcite ( various aspects of nonlinear quantum @xcite and classical @xcite charge and spin transport @xcite have been discussed later on . )
it is useful to consider ( anti ) symmetric second order conductance @xmath8 defined as @xmath9 where @xmath10 is a combination of voltages at the gates and contacts varied in the experiment and @xmath11 accounts for the spin degeneracy .
we emphasize that , depending on the way voltages are varied , experiments probe different linear combinations of second order conductance elements @xmath1 of eq .
( [ eq : iv ] ) . from now on we will simply call @xmath12 conductances and if no confusion is possible leave out the expression `` second order '' . in the presence of a dc perturbation
the mesoscopic averages of antisymmetric @xcite and symmetric @xcite conductances vanish , and it is their sample - to - sample fluctuations that are measured .
experiments are usually performed for strongly interacting samples and the magnetic - field components @xmath13 allow one to evaluate the strength of interactions . @xcite in previous theoretical works on nonlinear transport through chaotic dots several important issues have been discussed using random matrix theory ( rmt ) .
@xcite snchez and bttiker @xcite found the fluctuations of @xmath14 in a dot with arbitrary interaction strength at zero temperature and broken time - reversal symmetry due to magnetic field .
polianski and bttiker considered the statistics of both @xmath14 and @xmath15 for arbitrary flux @xmath7 , the temperature @xmath16 , and the dephasing rate .
@xcite the fluctuations of relative asymmetry @xmath17 and the role of the contact asymmetry on this quantity were discussed in ref .. the results of rmt approach were compared with experimental data of zumbhl @xcite and angers @xcite previously we considered statistics of @xmath8 for the dots where only one dc voltage was varied .
however , to avoid parasitic circuit effects some experiments are performed varying several voltages simultaneously .
surprisingly , the importance of the chosen combination of varied voltages ( bias mode ) was not addressed before in the literature .
it turns out that an experiment where only one of the voltages is varied @xcite or two voltages are asymmetrically shifted @xcite measure different combinations of nonlinear conductances @xmath1 .
for example , in a weakly interacting dot in the first mode we found that @xmath18 , @xcite but in the second bias mode the fluctuations of nonlinear current are strongly reduced , so that @xmath19 .
it is also important to generalize the previous treatment of the nonlinear current to mesoscopic systems biased by an ac - voltage at _ finite _ frequency .
the resulting dc is sometimes called `` photovoltaic current '' .
we expect that in such mesoscopic ac / dc converters the interactions lead to significant magnetic field - asymmetry in the dc - signal .
the rectification effect of mesoscopic diffusive metallic microjunctions was theoretically considered by falko and khmelnitskii @xcite assuming that electrons do not interact .
therefore , a magnetic - field asymmetry was not predicted and was also not observed in subsequent experiments . @xcite the fact that the interactions induce a magnetic field - asymmetry of the photovoltaic current when the size of the sample is strongly reduced was recently demonstrated in aharonov - bohm rings by angers @xcite however , it turns out that for an ac perturbation another quantum interference phenomenon , also quadratic in voltage , random in sign and magnetic field - asymmetric , contributes to the dc .
due to _ internal _
ac- perturbations of the sample , the energy levels are randomly shifted and a phenomenon commonly referred to as `` quantum pumping '' @xcite appears .
brouwer demonstrated that two voltages applied out of phase generate pumped current linear in frequency , while a single voltage pumps current quadratic in frequency @xmath4 .
@xcite although theory usually considers small ( adiabatic ) frequencies , a photovoltaic current could be induced by voltages applied at arbitrary frequency . at small @xmath4
the pumping contribution vanishes and only the rectification effect survives .
in contrast , it is not clear what the ratio of pumping current to rectification current is at large @xmath4 . to distinguish between different mechanisms
it is therefore important to consider rectification in a wide range of frequencies in detail . at the locally applied gate , which slightly changes the shape of the dot (
shown dashed ) , or microwave antenna emitting photons with energy @xmath20 into the dot . bottom : rectification sources include external bias @xmath21 , top gate voltage @xmath22 with capacitance @xmath23 , and parasitic coupling of @xmath24 due to stray capacitances @xmath25 .
microwave antenna can emit photons to the contacts and lead not only to photon - assisted ac transport but also to a rectified dc.,title="fig : " ] we point here to a crucial difference between rectification and pumping contributions to the photovoltaic effect .
rectification results from external perturbations or the perturbations that can be reduced to the exterior by a gauge transformation .
typical examples are external ac - bias , or gate voltage which shifts all levels uniformly , @xcite or a bias induced by parasitic ( stray ) capacitance which connects sources of possible internal perturbations to macroscopic reservoirs , @xcite see the bottom panel in fig .
[ fig : dot ] .
pumping , on the other hand , is due to internal perturbations like those of a microwave antenna @xcite or a locally applied gate voltage , @xcite see the top panel in fig .
[ fig : dot ] .
internal and external sources affect the schrdinger equation and its boundary conditions , respectively . in experiment
pumping and rectification , often considered together under the name of photovoltaic effect , @xcite are hard to distinguish .
can one clearly separate quantum pumping from rectification effects ? to distinguish them
it was proposed to use magnetic field asymmetry of dc as a signature of a true quantum pump effect . in refs . and
rectification by ( non - interacting ) quantum dot was due to stray capacitances of reservoirs with pumping sources .
the rectified current was found to be symmetric with respect to @xmath26.@xcite while such field - symmetric rectification dominated in the experiments of switkes @xcite and dicarlo @xcite at mhz frequencies , an asymmetry @xmath27 observed at larger ghz frequencies seemed to signify a quantum pump effect .
@xcite it was noted that the coulomb interactions treated self - consistently do not lead to any drastic changes in the mesoscopic distribution of a pumped current.@xcite probably , that is why the effect of interactions on the rectification have not been considered yet , even though the coulomb interaction in such dots is known to be strong.@xcite however , as it turned out later , coulomb interactions are responsible for magnetic - field asymmetry in nonlinear transport through quantum dots .
@xcite similarly this could be expected for rectification as well .
then the magnetic field asymmetry alone can not safely distinguish pumping from rectification .
therefore we thoroughly examine the frequency dependence of the magnetic - field ( anti)symmetric conductances @xmath8 . here
we neglect any quantum pumping effects and their interference with rectification . @xcite
while the role of coulomb interactions and the full frequency dependence in quantum pumping are yet to be explored , here we answer two important questions concerning a competing mechanism , rectification : ( 1 ) in the dc limit @xmath28 for a strongly interacting quantum dot @xmath14 and @xmath15 are of the same order .
is this also the case at finite frequencies ? ( 2 ) how are the experimental data affected by the bias mode for alternating voltages ?
a number of very recent experiments on nonlinear dc transport @xcite and ac rectification @xcite have used submicron ring - shaped samples with a relatively large aspect ratio . in this work
we develop a model of a ring which includes chaotic dynamics due to possible roughness of its boundary and/or the presence of impurities .
similarly to quantum dots , the two - terminal nonlinear conductance of such a ring is field - asymmetric because field - asymmetry exists in each arm . in particular
, this leads to deviations of the phase in ab oscillations from @xmath29 which characterizes linear conductance obeying onsager symmetry relations .
experiments find that the amplitude and phase of ab oscillations exhibit rather curious properties .
for example , the dc experiment of leturcq @xcite finds that during many ab oscillations with period @xmath30 the phase is well - defined .
the experiment demonstrates that a nearby gate can vary the phase of the ab oscillations over the full circle .
the amplitude of the second harmonic @xmath31 is strongly suppressed . on the other hand ,
the dc experiment @xcite and ac experiment @xcite of angers find that the phase can be defined only for few oscillations at low magnetic fields . for high frequencies ,
the phase fluctuates strongly as function of frequency .
both in the nonlinear and the rectified current the amplitude of the second harmonic @xmath31 in ab oscillations is always comparable with the first harmonic @xmath30 .
this is in contrast with the experiments in ref .. although we do not fully address all these questions here , our model of a chaotic ring allows us to consider them at least on a qualitative level .
to introduce the reader to the problem of nonlinear transport in sec .
[ sec : bias ] we first qualitatively discuss the coulomb interaction effect in the simplest dc problem . in reality the statistical properties of conductances @xmath1 in eq .
( [ eq : iv ] ) are sensitive to electronic interference but to assess the role of coulomb interactions we can consider a specific sample .
in contrast to linear transport , it turns out that the nonlinear current strongly depends on the way voltages at the contacts and/or nearby conductors are varied from their equilibrium values ( bias mode ) .
for example , we find that the experiments when only one voltage at the contact is varied @xcite or when two contact voltages are shifted oppositely @xcite measure different nonlinear currents . indeed ,
for a current @xmath32 , bilinear in voltages , its second derivative should depend on the chosen direction in the space of voltages @xmath33 .
interestingly , a sample with weak interactions is very sensitive to the choice of the bias mode , which we attribute to the strong effect of capacitive coupling of the sample with nearby conductors . to make our arguments quantitative and
consider the role of magnetic flux @xmath7 for a quantum dot which is ( generally ) ac - biased at arbitrary frequency @xmath4 , in sec .
[ sec : dc ] we take electronic interference into account . having done that , we illustrate the interplay between interactions and interference on several important examples .
first , we consider nonlinear transport due to a constant applied voltage and then consider rectification of ac voltages .
for a two - terminal dot , in a generally asymmetric circuit ( capacitive couplings included ) , in sec .
[ sec:2terminal ] we find the statistics of ( anti ) symmetric conductances @xmath8 defined in eq .
( [ eq : ivfield ] ) .
both @xmath14 and @xmath15 vanish on average .
quantum fluctuations of @xmath15 strongly depend on the interaction strength , circuit asymmetry and bias mode .
this is in accordance with our qualitative picture . on the other hand ,
the antisymmetric component @xmath14 depends only on interactions .
our arguments agree with recent experiments in quantum dots:@xcite depending on the bias mode different features of the nonlinear conductance tensor are probed .
the fluctuations of nonlinear current can be minimized or maximized ( on average ) , which becomes important for weakly interacting electrons .
curiously , for symmetric coupling ( transmission and capacitance ) of contacts and dot the bias mode in which the voltages at the contacts are changed in opposite directions generally _ minimizes _ fluctuations of @xmath15 .
consequently , such a mode is more advantageous for the observation of @xmath14 or a cleaner linear signal . near the end of sec .
[ sec:2terminal ] we also demonstrate how to take into account possible classical circuit - induced asymmetry @xcite due to the finite classical resistance of the wires . in sec .
[ sec : rectify ] we present results elucidating the role of interaction in rectification through two - terminal dots .
usually there are two important time - scales : the dwell time @xmath34 an electron spends inside the dot and the charge relaxation time @xmath35 of the dot . for a given geometry ,
the dwell time depends on the coupling of the dot with reservoirs , but the charge relaxation time is also sensitive to the interaction strength .
we have @xmath36 for strong interactions and @xmath37 in the weak interaction limit .
our results for fluctuations of @xmath38 are obtained for arbitrary frequency @xmath4 .
although the fluctuations of both @xmath39 and @xmath40 monotonically decrease when @xmath41 , as functions of frequency @xmath4 they behave differently . at nonadiabatic frequencies
@xmath42 the nearby gate short - circuits currents .
this effect is even in magnetic field and thus affects only @xmath15 . as a result
, for a high - frequency voltage the asymptotes of @xmath14 and @xmath15 are generally different and strongly depend on the bias mode . since the regime of parameters is quite realistic , we expect that the predicted difference of @xmath40 and @xmath39 should be experimentally observable . in the noninteracting limit
our results qualitatively agree with those in diffusive metallic junctions .
our model of a ring consisting of a chaotic dot with a ballistic arm which encloses an ab flux is presented in section [ sec : phase ] .
although it is impossible to find the full mesoscopic distribution of the ab phase @xmath43 , its shape can be discussed qualitatively .
since @xmath44 is similar to the asymmetry parameter @xmath17 in quantum dots , its distribution can become very wide for a particular choice of the bias mode . on average @xmath45 in our model , and we find the dependence of the fluctuations of @xmath46 on temperature , interactions , and number of channels of the contacts and the arm .
our treatment allows a straightforward generalization to treat ac voltages applied to the ring .
the technical calculations are presented in the appendix .
[ ] [ ] [ 0.8]@xmath23[][][0.8]@xmath47 at reservoirs connected by @xmath48 ballistic channels and capacitances @xmath49 and by voltages @xmath50 applied at additional gates with capacitances @xmath51 .
transport through the dot is sensitive to the total magnetic flux @xmath7 through the area of the dot .
( center and right ) forward and reverse connection of ref .
exchange voltages at the contacts and classical resistors @xmath52.,title="fig : " ]
the 2d quantum dot , see the left panel in fig .
[ fig:3dot ] , is biased with several voltages @xmath53 at @xmath54 ballistic quantum point contacts ( qpcs ) with @xmath55 orbital channels .
the reservoirs can be capacitively coupled to the dot via capacitances @xmath56 .
an additional set of voltages @xmath57 is applied to ( several ) gates with capacitances @xmath51 .
all perturbations are assumed to be at the same frequency @xmath4 , which is not necessarily small ( adiabatic ) .
the dot is in the universal regime , @xcite when the thouless energy @xmath58 is large .
the dots with area @xmath59 ( taken circular ) are either diffusive with mean free path @xmath60 , or ballistic , with @xmath61 and chaotic classical dynamics ( in the latter case the substitution @xmath62 should be used ) .
the mean level spacing ( per spin direction ) @xmath63 and the total number of ballistic channels @xmath64 together define the dwell time @xmath65 .
we also require that @xmath66 when we can treat the nonlinearity only to @xmath67 .
scattering is spin - independent and this spin degeneracy is accounted for by the coefficient @xmath11 .
the noninteracting electrons are treated using the scattering matrix approach and random matrix theory ( rmt ) for the energy - dependent scattering matrix @xmath68 . for details
we refer the reader to reviews .
@xcite in this approach the fundamental property of a dot is its scattering matrix @xmath69 distributed over circular ensembles of proper symmetry , see ref .
( an alternative method is the hamiltonian approach based on the properties of the dot s hamiltonian @xmath70 taken from a gaussian ensemble .
@xcite ) transport properties of chaotic dots in rmt for matrices @xmath69 or @xmath70 are usually expressed in terms of an effective , magnetic field - dependent number of channels .
predictions based on this approach are in good agreement with experiment . for multichannel samples with @xmath71 we use the diagrammatic technique described in refs . and .
however , when interactions are present , this treatment should be modified . the approach which assumes that in a pointlike scatterer the interactions appear in the form of a self - consistent potential
was introduced by bttiker and co - authors @xcite on the basis of gauge - invariance and charge conservation .
this ( hartree ) approach neglects contributions leading to coulomb blockade ( fock terms ) , but is a good approximation for open systems .
if the screening in the dot inside the medium with dielectric constant @xmath72 is strong , @xmath73 , an rpa treatment of coulomb interactions is sufficient . for large dots , @xmath74 ,
the details of screening potential on the scale @xmath75 are not important and we can assign an electric potential @xmath76 defined by excess electrons at @xmath77 at any point @xmath78 of the sample . if additionally the number of ballistic channels @xmath64 is much smaller than the dimensionless conductance of a closed sample , @xmath79 , the potential drops over the contacts and therefore in the interior of the dot it can be taken uniform ( `` zero - mode approximation '' ) .
@xcite this potential shifts the bottom of the energy band in the dot and thus modifies the @xmath69-matrix . as a consequence ,
electrons with kinetic energy @xmath80 have an electro - chemical potential @xmath81 in the contact @xmath5 and @xmath82 in the dot .
( we point out that we neglect the quantum pumping in the dot and consequently the @xmath69-matrix depends only on one energy . ) recently , brouwer , lamacraft , and flensberg demonstrated that this self - consistent approach gives the leading order in an expansion in the inverse number of channels @xmath83 .
@xcite therefore , our analytical results present the leading order effect , valid for @xmath83 . in the self - consistent approach
the influx of charge changes the internal electrical potential of the dot @xmath84 , which in turn affects the currents incoming through each conducting lead and/or redistributes charges among the nearby conductors ( gates ) .
such capacitive coupling can often be estimated simply from the geometrical configuration .
for example , the capacitance of a dot covered by a top gate at short distance @xmath85 is @xmath86 and a single quantum dot has @xmath87 .
the ratio of charging energy @xmath88 to mean level spacing @xmath89 characterizes the interaction strength .
it is proportional to the ratio of the smallest geometrical scale to the effective bohr s radius , @xmath90 .
we refer to interactions as strong if @xmath91 and weak if @xmath92 .
and @xmath93 , the dashed curves correspond to equal currents .
thin line shows fixed @xmath94 and @xmath95 is a function of source voltage @xmath96 .
thick line corresponds to fixed @xmath97 , such that @xmath95 depends only on @xmath98 .
full and empty dots on the right figure correspond to the forward or reverse configurations shown in fig .
[ fig : dot].,title="fig : " ] we suppose for simplicity that at equilibrium the voltages @xmath99 are set . in the following
we consider the situation when the ( single ) gate voltage @xmath100 is held fixed at its equilibrium value .
experiments can be performed in different _ bias modes _ , usually either ( i ) with fixed drain voltage @xmath94 or ( ii ) at fixed @xmath97 ( the variations of the voltages at the contacts are equal in magnitude but opposite in sign ) .
these different modes correspond to straight lines in the @xmath101 plane shown in fig .
[ fig : vaxes ] .
let us consider the nonlinear current as a function @xmath102 , where @xmath103 and @xmath93 are deviations of contacts voltages from equilibrium . for generality
we consider below a situation when the linear combination @xmath104 is held fixed and the only variable is @xmath105 this corresponds to a rotation of the original @xmath106 axes such that the new coordinate axis @xmath10 makes an angle @xmath107 with the @xmath108 line , as illustrated in fig .
[ fig : vaxes ] .
the value of @xmath107 fully characterizes the bias mode .
now the two modes introduced above are simply ( i ) @xmath109 which implies @xmath110 ; and ( ii ) @xmath111 , which implies @xmath112 and corresponds to an asymmetric variation of the voltages .
the linear current depends only on @xmath113 ( dashed lines on the left panel in fig .
[ fig : vaxes ] correspond to the lines of equal currents ) and in any bias mode the measured linear current @xmath114 is the same for a given @xmath113 . if we consider the nonlinear current @xmath115 as a function of @xmath116 , it is by construction a bilinear function of @xmath116 . as in the linear case
the current must vanish if the voltages are the same and thus @xmath117 for @xmath118 .
therefore , the bilinear function must be of the form @xmath119\ , ( x - y)\end{aligned}\ ] ] with unknown ( generally fluctuating ) parameters @xmath120 and @xmath121 $ ] .
it is important that the qualitative behavior of @xmath102 depends on the interaction strength : one could expect that transport depends not only on voltages in the leads , but also on the internal nonequilibrium potential @xmath122 of the sample .
this potential can be found if potentials in all reservoirs and the nearby gate are known . in the limit of weak interactions
the equilibrium point @xmath100 is important , and if we reverse the bias voltage , @xmath123 the current is fully reversed , that is @xmath124 . for the current defined in eq .
( [ eq : simplei ] ) it is possible only when @xmath125 .
another way to see this is to use the usual expression for the total current in terms of scattering matrices . in this formula
the current depends on the difference between fermi distributions in the leads @xmath126 , and its expansion up to the second order yields @xmath127 .
the lines of equal current are curved and directions @xmath128 correspond to zero current directions .
thus the dependence of current on the angle @xmath107 is strong .
in addition this approach predicts that the current through a two - terminal sample is symmetric with respect to the magnetic flux inversion .
in contrast , for strong interactions , the value of @xmath100 is irrelevant and the nonequilibrium electrical potential @xmath122 is independent of @xmath100 . in this case
current depends only on the voltage difference @xmath113 and thus @xmath129 for linear transport .
therefore we do not expect any nontrivial dependence of the nonlinear current on the choice of the bias mode .
it is noteworthy that qualitative considerations can predict neither the sign , nor the magnitude of @xmath120 .
the only general conclusion which we can make for a weakly or strongly interacting dot is @xmath130 and @xmath131 , respectively .
experiments extract derivatives of @xmath115 with respect to the applied voltages .
importantly , this derivative depends on the chosen direction @xmath107 .
the nonlinear current measured in this bias mode is @xmath132 the current is zero when @xmath133 and @xmath134 , and the bisectrix of the angle between the two zero - current directions at @xmath135 maximizes @xmath136 .
sometimes experiments extract information on nonlinearity from measurements in different connections schematically shown in the central and right panels in fig .
[ fig:3dot ] : `` forward '' connection corresponds to @xmath137 , while `` reverse '' connection for the same voltage configuration corresponds to @xmath138 .
the gate voltages @xmath139 are kept fixed . in fig .
[ fig : vaxes ] these forward and reverse points are indicated by black and white dots , respectively . to find the nonlinear conductance marlow @xcite and lfgren @xcite
determine the difference of conductance at these measurement points .
lfgren @xcite use the term `` rigidity '' for samples for which @xmath140 in the points @xmath141 and @xmath142 .
@xcite equation ( [ eq : simplei ] ) gives the nonlinear contribution @xmath143 to the full conductance @xmath144 : @xmath145\,.\end{aligned}\ ] ] thus for a sample which is called rigid this implies @xmath146 .
since @xmath147 would mean that there is no second - order response , we must have @xmath148 which is the case for samples with strong interaction .
in other words , `` rigidity '' in samples which exhibit @xmath149 current is equivalent to strong coulomb interactions .
on the other hand , comparison of data at another pair of points @xmath150 and @xmath151 gives @xmath152 and provides _ additional _ information about the two fluctuating quantities @xmath153 .
reference expects that a left - right ( lr)-symmetric system has @xmath154 .
therefore rigid and lr - symmetric sample should necessarily have @xmath155 and thus could not exhibit a second - order current @xmath149 .
this point is discussed more quantitatively in sec .
[ sec:2terminal ] .
it is important to note that to find the linear dc current one needs to know only @xmath156 , while for the nonlinear current in general one needs two variables @xmath157 or any independent pair of their linear combinations .
the projection of the vector @xmath158 on the @xmath159const plane uniquely defines the nonlinear current .
this projection can be parametrized by the pair of cartesian @xmath160 or axial coordinates @xmath161 .
however , if in the experiment the voltages @xmath162 were fixed , this would not be enough to define @xmath160 uniquely . in this case ref .
points to the importance of the reference point @xmath100 .
indeed , one could arrive at the point with a given @xmath163 from any equilibrium point and the measured current would depend on @xmath100 .
we prefer to characterize the measurement by the pair @xmath164 instead of three variables @xmath158 because of the simplicity of the final results .
the weaker the interaction ( or the stronger the capacitive coupling of the sample to the nearby gate ) the more important the role of @xmath107 chosen in experiment .
we illustrate this important conclusion by quantitative results for nonlinear conductance @xmath165 in the following sections .
we point out that conductance with respect to the voltage difference @xmath166 is often used , even when a linear combination @xmath10 is actually varied in experiment .
voltages @xmath10 and @xmath167 are related , @xmath168 , and one can straightforwardly find @xmath169 .
now we quantify the qualitative arguments of sec . [
sec : bias ] and consider the more general situation of a dc current generated by an ac bias . if at first we neglect coulomb interactions , the nonlinear dc current @xmath170 in response to the fourier components @xmath171 of the ac voltages applied at the contacts @xmath172 , can be expressed with the help of the dc - conductance matrix @xmath173 of the dot at the energy @xmath72
@xcite @xmath174.\ ] ] if we now include interactions using a self - consistent potential @xmath175 this formula is modified : @xcite in eq .
( [ eq : pedersen ] ) the fourier components of the voltages at _ all _ contacts are shifted down by the fourier component of the internal potential @xmath176 @xmath177\nonumber \\ & \times & \frac{f(\e)-f(\e+\hbar\oo)}{\hbar\oo}.\label{eq : sumg}\end{aligned}\ ] ] in eq .
( [ eq : uomega ] ) the index @xmath178 runs not only over real leads @xmath179 , but also over all gates @xmath180 . however , when @xmath181 the ac conductance @xmath182 is absent and only capacitive coupling @xmath183 remains in the numerator .
we point out that the matrix @xmath184 of dynamical ac conductance at frequency @xmath4 given in eq .
( [ eq : sumg ] ) should not be confused with the degenerate matrix @xmath185 of energy - dependent dc conductances of electrons with kinetic energy @xmath72 given in eq .
( [ eq : gdc ] ) .
the results of ref . can be expressed in terms of the dc conductances @xmath186 and frequency - dependent characteristic potentials
@xmath187 , @xmath188 stands for the real part of @xmath187 , which is in general a complex quantity .
in contrast to eq .
( [ eq : pedersen ] ) , eq .
( [ eq : current4omega ] ) is expressed via differences of voltages applied to all present conductors .
therefore , the current is gauge - invariant .
the charge conservation , @xmath189 , is obvious from eq .
( [ eq : gdc ] ) . from this point on we consider eq .
( [ eq : current4omega ] ) , a specific expression of eq .
( [ eq : iv ] ) , in detail for several regimes . in sec .
[ sec:2terminal ] we discuss the nonlinear current due to dc applied voltages ( previously considered in ref . ) and the importance of different bias modes in experiments in two - terminal quantum dots . in sec .
[ sec : rectify ] we consider the frequency dependence of @xmath40 and @xmath39 . in the static limit @xcite @xmath190 the integrand in the first line of eq .
( [ eq : current4omega ] ) simplifies to @xmath191 and for @xmath192 the derivatives @xmath187 are real and expressed via subtraces of the hermitian wigner - smith matrix @xmath193 @xcite @xmath194 for a two - terminal sample the nonlinear current through the first lead is @xmath195+(u_2-u_1)(v_1-v_2)^2\right].\end{aligned}\ ] ] the characteristic potentials in the last term of eq .
( [ eq : current ] ) are sensitive to the asymmetry of the contacts . indeed , in a strongly interacting dot @xmath196 and @xmath197 .
the current magnitude grows with asymmetry due to the last term in eq .
( [ eq : current ] ) . on the other hand ,
the sign of @xmath198 is random because of quantum fluctuations of @xmath199 around zero .
@xcite as a consequence , if in an experiment the fermi level is shifted by @xmath200 ( or the shape of the dot is changed ) the sign of nonlinearity can be inverted .
different modes of bias having been discussed in sec .
[ sec : bias ] , we concentrate here on the ( anti)symmetric conductances through the quantum dot at fixed gate voltages . when the reservoir voltages are varied in the @xmath107 direction , the nonlinear current is given by the expression @xmath201\cos\eta\tilde v^2,\end{aligned}\ ] ] and one can define exactly the unknown parameters @xmath153 which we introduced in the qualitative argument leading to eq .
( [ eq : iexample ] ) . depending on @xmath107
one measures different linear combinations of conductances . if we consider conductances @xmath202 in units of @xmath203 , eqs .
( [ eq : ivfield ] ) and ( [ eq : deriviv ] ) yield @xmath204}\end{aligned}\ ] ] expressed in terms of fluctuating functions @xmath205 and a traceless matrix @xmath206 : @xmath207,\\ \label{eq : chi2s } \chi_{2,s}(\e ) & = & \delta\left(\frac{c_0\tan\eta+c_2-c_1}{e^2\nu_s } + \frac{n_2-n_1}{n}\frac{\tr \s^\dagger\dd_\e\s}{2\pi i } \right.\nonumber \\ & & \left .
+ \frac{1}{2\pi i } \tr\lambda\{\s^\dagger,\dd_\e\s\}\right).\end{aligned}\ ] ] standard calculations using the wigner - smith and/or @xmath69-matrix averaging @xcite yield @xmath208 .
this result signifies that the nonlinear current through a quantum dot is indeed a quantum effect . as a consequence
the size of the measured nonlinearity must be evaluated from correlations of @xmath8 .
the functions @xmath209 and @xmath210 are uncorrelated , and their autocorrelations @xcite readily allow one to find statistical properties of @xmath211 .
our results can be expressed in terms of diffuson @xmath212 or cooperon @xmath213 in a time representation , @xmath214 and @xmath215 . both can be introduced using the @xmath69-matrix correlators @xcite ( correlations of retarded and advanced green functions lead to the same expression up to a normalization constant @xcite ) .
we have @xmath216 we also introduce the electrochemical capacitance @xmath217 @xcite which relates the non - quantized mesoscopically averaged excess charge @xmath218 in the dot in response to small shift of the voltages @xmath219 at all gates .
in addition the charge relaxation time @xmath220 of the dot is conveniently introduced by this electrochemical capacitance and the total contact resistance , @xmath221 the denominator of eq .
( [ eq : defg ] ) is a self - averaging quantity , @xmath222 . using the diffusons and cooperons defined in eq .
( [ eq : cum1 t ] ) we find the following correlations of @xmath14 and @xmath15 : @xmath223 there are two very different contributions to eq .
( [ eq : main ] ) , @xmath224 due to quantum interference and @xmath225 defined by the classical response of the internal potential to external voltage .
the terms denoted by @xmath226 are sensitive to temperature , magnetic field , and decoherence .
asymptotical values of @xmath227 in the low temperature , @xmath228 , or high temperature limits , @xmath229 , are @xmath230 and @xmath231 , respectively . the term denoted by @xmath225 and given by eq .
( [ eq : x ] ) contains only quantities specifying the geometry of the sample and gates and the bias mode . in a real experiment the coupling due to capacitances
@xmath232 is usually stronger then that of the external gates , @xmath233 .
symmetrization of the circuit @xmath234 can diminish the value of @xmath225 .
if in addition @xmath235 and @xmath111 ( used in the experiments @xcite ) we have @xmath236 .
thus such a symmetric setup and bias mode minimize the fluctuations of the nonlinear current and actually would be best for an accurate measurement of _ linear _ transport .
indeed , this regime is not affected by the fluctuations of capacitive coupling @xmath237 of the dot with the nearby gate and thus minimizes fluctuations of @xmath15 around 0 .
fluctuations of @xmath8 are given by different expressions , see the first line of eq .
( [ eq : main ] ) , where the first term is due to @xmath238 or @xmath239 .
importantly , @xmath239 contains both quantum @xmath240 and classical @xmath225 contributions .
if the classical term dominates , @xmath241 , the current is mostly symmetric , @xmath242 .
this could be expected either for a weakly interacting dot or a very asymmetric setup , @xmath243 .@xcite however , if the classical term is reduced due to , e.g. , the bias mode , the fluctuations of @xmath14 and @xmath15 become comparable . this experimentally important conclusion remains valid for _ any interaction strength_. ( particularly , it leads to a very wide distribution of the aharonov - bohm phase considered in sec .
[ sec : phase ] . )
experiments of zumbhl @xcite and leturcq @xcite are performed in this regime when @xmath111 and @xmath236 .
data in ref . demonstrate that the part of the total current symmetrized with respect to magnetic field is by far dominated by linear conductance . from eq .
( [ eq : main ] ) we expect mesoscopic fluctuations in linear conductance to be @xmath244 times larger then those of @xmath245 .
thus only when the number of channels is decreased will the nonlinear @xmath15 become noticeable . a clear observation of @xmath15 without linear transport contribution was performed in a dc aharonov - bohm experiment by angers @xcite in the mode @xmath246 ( only one contact voltage was varied ) .
this allowed to evaluate the interaction strength from the ratio of @xmath247 .
experiments of marlow @xcite and lfgren @xcite measure the full two - terminal conductance and extract nonlinear conductance properties related to various spatial symmetries of the dot .
although the current through a weakly interacting sample is field - symmetric , this is not true in general .
samples of ref .
differ in `` rigidity '' and degree of symmetry .
rigid samples , @xmath248 , with left - right(lr ) and up - down ( ud)-symmetry should have @xmath249 and @xmath250 respectively , according to the expectations of lfgren @xcite ( indices @xmath251 and @xmath252 mean the symmetric and antisymmetric part in magnetic field ) .
due to quantum fluctuations , in experiment none of these symmetry - relations can be exactly fulfilled , see eq .
( [ eq : main ] ) . according to eq .
( [ eq : deriviv ] ) , the difference in the full conductances @xmath253 measured between different points probes different characteristic potentials .
reference defines three differences @xmath254 for three pairs of points in the forward and reverse connection discussed after eq .
( [ eq : ivangle ] ) . using eq .
( [ eq : main ] ) we find ( i ) @xmath255 , ( ii ) @xmath256 , and ( iii ) @xmath257 .
the ensemble average of these differences vanishes and their fluctuations for @xmath258 are given by @xmath259 where @xmath260 is found from eq .
( [ eq : x ] ) at @xmath261 . in weakly
interacting dots @xmath262 and only magnetic - field symmetric signals @xmath263 and @xmath264 survive . in strongly interacting ( `` rigid '' ) dots @xmath265 and @xmath266 becomes similar to @xmath264 .
we point out that even if the rigid samples are made symmetric with respect to left - right inversion , the quantum fluctuations of the sample properties are unavoidable and @xmath267 at @xmath268 .
for high magnetic fields and arbitrary interactions @xmath269 and experiment should observe @xmath270 .
clearly fluctuations exist also for large magnetic fields beyond the range of applicabilty of rmt . experimental data
( see inset of fig .
6 in ref . )
show that @xmath271 .
it is hard to make a quantitative comparison with refs . and , since the quantum fluctuations in the nonlinear conductance exist possibly on the background of classical effects due to macroscopic symmetries .
we expect that quantum effects become more pronounced as contacts are narrowed . to conclude this subsection we briefly discuss here the case of a macroscopically asymmetric setup .
if the experiment were aimed to measure large @xmath14 compared to @xmath15 , one would try to minimize @xmath15 by adjusting the setup .
such a procedure minimizes the value of @xmath225 in eq .
( [ eq : x ] ) .
for @xmath272 the role of asymmetric contacts @xmath243 was discussed in ref ..
analogously , one could consider a more general case of @xmath232 and an arbitrary bias mode @xmath107 .
this is especially important if the difference @xmath273 can not be neglected due to occasional loss of contact symmetry .
the results of an experiment could also be affected by the presence of classical resistance loads @xmath52 between macroscopic reservoirs and the dot ( shown in fig .
[ fig:3dot ] ) .
swapping of such resistances in the experiment , when connection is switched between `` forward '' and `` reverse '' @xcite affects the voltage division between loads . if we assume the capacitive connection of the dot and reservoirs is still the same , the modification of the expression for @xmath187 in eq .
( [ eq : uomega ] ) is straightforward , @xmath274 .
naturally , at large @xmath275 , @xmath276 , the main drop of the voltage occurs over the resistor @xmath275 and not over the qpcs . as a consequence ,
if @xmath277 , values of @xmath278 can become unequal due to @xmath279 and this leads to the classical circuit asymmetry which we do not consider here . here
we consider the dc generated by a quantum dot subject to an ac bias at the frequency @xmath4 . in experiment at high bias frequency
@xmath280 current is usually measured at zero frequency .
in contrast , at small bias frequency @xmath281 higher harmonics ( for instance the second harmonic @xmath282 ) can be measured . however , up to corrections small due to @xmath281 , the second harmonic is just equal to the rectified current , @xmath283 . therefore , to leading order , our results for the rectified current describe both experiments .
generally , there are several important time - scales characteristic for time - dependent problems in chaotic quantum dots . to see how they appear let us first consider frequency - dependent linear transport of noninteracting electrons .
its statistics usually depend only on the flux - dependent time scales @xmath284 , see eq .
( [ eq : channels ] ) . if we consider an analog of ucf @xmath285 for the frequency - dependent conductances introduced in eq .
( [ eq : sumg ] ) , we find @xmath286\sinh^2 \pi t\tau/\hbar}.&&\end{aligned}\ ] ] the presence of @xmath287 in the diffuson and cooperon contributions in the second line of eq .
( [ eq : linear ] ) is due to the energy dependence of the scattering matrix @xmath68 , which usually brings up imaginary corrections to the matrix - element correlators . in a dc - problem
@xmath28 it is usually useful to introduce a dimensionless number of channels @xmath288 modified by the magnetic field , see eq .
( [ eq : channels ] ) . in this limit at @xmath289
the integration in eq .
( [ eq : linear ] ) becomes straightforward and summation is then performed over @xmath290 . for equal magnetic fields , @xmath291
, we have @xmath292 , but @xmath293 is strongly modified by large fields , @xmath294 , which suppresses the weak localization correction and diminishes ucf .
however , for an ac problem ( especially for @xmath280 ) it is more convenient to express results in terms of dimensionless quantities @xmath295 .
for example , from eq .
( [ eq : linear ] ) the statistics of conductance can be easily evaluated : @xmath296 and the real and imaginary parts of conductance are similar and uncorrelated at high frequency @xmath42 .
inclusion of interactions introduces an ( additional ) dependence on @xmath220 , the charge - relaxation time defined in eq .
( [ eq : excess ] ) . to leading order in @xmath83 the effect of interactions
is often to substitute @xmath297 in the noninteracting results , e.g. , for the linear conductance @xcite or shot noise .
@xcite interestingly , the subleading corrections depend on both @xmath220 and @xmath34 , e.g. , in the weak localization correction in the absence of magnetic field.@xcite when the magnetic field is increased to values which finally break time - reversal symmetry , the appearance of different time scales @xmath284 is expected , see e.g. , eq .
( [ eq : linear ] ) .
therefore at intermediate magnetic fields , when @xmath298 , and the interactions taken into account , @xmath299 , the solution of an ac problem is expected to show a complicated dependence on all these time scales . indeed
, if we consider the rectified current such an interplay between @xmath284 and @xmath220 does appear .
we find @xmath208 and present below results for correlations of @xmath14 and @xmath15 : @xmath300\left ( \frac{4\pi\cos^2\eta } { \delta}\frac{c_\m}{c_\sigma}\right)^2\frac{n_1 ^ 3 n_2 ^ 3}{n^6 } \label{eq : vargs}.\end{aligned}\ ] ] here the functions @xmath301 are finite - frequency generalizations of eq .
( [ eq : fraw ] ) @xmath302 the subscripts @xmath303 of @xmath304 illustrate the origin of these functions : they result from averaging of different scattering properties over the energy band defined by @xmath305 .
the function @xmath306 is a characteristic of the internal potential @xmath307 , see eq .
( [ eq : uomega ] ) .
the function @xmath308 results from the energy averaging of the dc conductance @xmath185 .
such averaging appears because both @xmath309 defined in eq .
( [ eq : sumg ] ) and @xmath185 in eq .
( [ eq : current4omega ] ) are coupled to the fermi distribution .
the function @xmath310 is @xmath311 and in the static limit @xmath28 it is given by eq .
( [ eq : x ] ) .
we point out that when the interactions are negligible , @xmath312 , the role of the bias mode is significant .
a quantum dot with ( fully broken ) time - reversal symmetry can be labeled by dyson symmetry parameter ( @xmath313 ) @xmath314 .
when the setup is ideal , @xmath315 , and @xmath316 , the fluctuations of @xmath8 at large frequencies @xmath42 are @xmath317 in chaotic quantum dots the role of the thouless energy @xmath318 of the open systems is often played by the escape rate @xmath319 . if we take this into account , our result ( [ eq : fk ] ) qualitatively agrees with that of falko and khmelnitskii @xcite obtained for open diffusive metallic junctions .
however , when @xmath320 , the fluctuations of @xmath15 are much smaller and for @xmath321 they become comparable with those of the antisymmetric conductance ( [ eq : fkasym ] ) .
however , very often experiments are performed in samples , where the interaction is not weak . since @xmath14 and @xmath15 are comparable for strong coulomb interactions in the static limit @xmath322,@xcite we concentrate here on this experimentally relevant regime of @xmath323 and take an ideal symmetric setup , @xmath235 and @xmath315
. then we have @xmath324 below we consider in detail the case @xmath316 and how this bias mode affects the behavior of @xmath325 .
several frequency regimes can be separated : adiabatic @xmath326 , intermediate , where @xmath327 , and high frequencies @xmath328 .
the asymptotes of the functions defined in eqs .
( [ eq : fu ] ) , ( [ eq : fg ] ) , and ( [ eq : xomega ] ) in these regimes are presented in table [ tab : table ] for reference . for adiabatic frequencies @xmath326 the integrands in eqs .
( [ eq : fu ] ) and ( [ eq : fg ] ) do not oscillate on the short time scale @xmath329 . at such small frequencies @xmath330
are equal to @xmath227 of eq .
( [ eq : fraw ] ) and @xmath331 can be neglected .
this is essentially the zero frequency regime considered before for nonlinear dc transport .
as the frequency grows , an intermediate regime is reached when max @xmath332 and @xmath333 start to differ .
the scattering properties at large energy difference @xmath334 are uncorrelated and the response of the dot is randomized .
therefore both the conductance averaged over a large energy window @xmath335 and the response of the internal potential @xmath307 to the ac voltage at @xmath336 are strongly suppressed , see table [ tab : table ] . as a result ,
if @xmath310 is still negligible , both @xmath337 and @xmath338 decrease with growing frequency as @xmath339 .
.asymptotes of @xmath340 at @xmath341 [ tab : table ] [ cols="^,^,^,^ " , ] it is noteworthy that a recent experiment in ab rings @xcite finds that @xmath342 grows with frequency until @xmath343 and then decreases @xmath344 . while we predict a monotonic decrease of @xmath345 , this growth could be the result of quantum pumping or an interference of the pumping and rectification ( both effects were neglected here ) .
in this section we consider nonlinear transport through a chaotic aharonov - bohm ( ab ) ring .
the nonlinear conductance @xmath143 exhibits periodic ab oscillations and non - periodic fluctuations , similarly to the linear conductance @xmath346 .
however , since coulomb interactions produce asymmetry of @xmath143 with respect to magnetic field inversion , the phase of these oscillations is not pinned to @xmath347 . as a quantum effect
this ab phase is characterized by a mesoscopic distribution .
the width of this distribution represents a typical fluctuation .
we first discuss what kind of distribution could be expected in a chaotic ab ring and then calculate the fluctuation of the ab phase .
let us assume that @xmath143 as a function of magnetic flux @xmath7 can be expanded into the series of well - defined fourier harmonics similarly to the linear conductance @xmath346 : @xmath348 the phase @xmath46 of the main ( first ) harmonic @xmath349 is obtained from the ratio of the ( anti ) symmetrized conductances defined in eq .
( [ eq : ivfield ] ) @xmath350 we can not find the full mesoscopic distribution of the phase @xmath351 .
we can gain some insight in the behavior of this phase by investigating a similar quantity , namely , the asymmetry parameter @xmath17 considered previously for chaotic dots .
@xcite based on eq .
( [ eq : tan ] ) we argue that the statistical properties of @xmath352 and the ab phase @xmath46 should be similar .
in quantum dots the parameter @xmath353 is given by the ratio @xmath354 , see eqs .
( [ eq : defg ] ) , ( [ eq : chi2a ] ) , and ( [ eq : chi2s ] ) . the functions @xmath355 at @xmath356 are convolved separately with @xmath357 , and at @xmath358 ( which we consider below ) they are evaluated at the fermi energy .
the properties of @xmath355 and the dependence of @xmath359 on the bias mode were described after eq .
( [ eq : x ] ) .
the function @xmath359 can have a nonzero ( classical ) average @xmath360 defined by the interaction strength , geometry of the setup , and the bias mode @xmath107 .
since @xmath361 and the fluctuations of @xmath355 are small as @xmath362 , the mesoscopic distribution of @xmath363 is narrow and concentrated close to 0 .
however , @xmath364 is possible if @xmath236 , e.g. , for symmetric contacts and the bias mode @xmath111 . in this case
, the distribution of @xmath352 becomes wide regardless of the interaction strength .
[ t]@xmath365[l]@xmath366 [ l][][0.8]@xmath367 [ l][][0.8]@xmath368 [ l][][.7]@xmath369 [ l][][.7]@xmath370 of @xmath371 .
( main plot ) if the contacts are asymmetric ( bold curve , @xmath368 ) the distribution is narrow , while for symmetric contacts ( dashed , @xmath367 ) it is almost uniform .
as shown in the inset , for symmetric contacts at large @xmath64 the distribution becomes uniform , compare bold curve for @xmath370 and dashed for @xmath369.,title="fig:",width=302 ] + the role of the classical contribution on the shape of @xmath372 is demonstrated in the main plot in fig .
[ fig : delta ] for @xmath111 , where the distributions for asymmetric , @xmath373 , and symmetric contacts , @xmath374 , are presented .
while the distribution is almost uniform , when the classical contribution @xmath225 is absent , it is highly peaked near zero when @xmath225 dominates .
if @xmath225 is absent , the correlations between @xmath14 and @xmath15 are significant at small @xmath64 .
this leads to a nonuniform distribution of @xmath372 , which is peaked at @xmath375 and @xmath376 when @xmath370 , see the inset in fig . [
fig : delta ] .
when @xmath64 grows , the correlations between @xmath14 and @xmath15 vanish and therefore the distribution becomes uniform . such a distribution could be easily obtained if we make the natural assumption that @xmath8 are independent and distributed by the gaussian law with the same width .
these numerics were performed for @xmath111 , when the mesoscopic distribution of @xmath17 becomes insensitive to the interaction strength .
the role of interactions appears only if @xmath316 , when the classical contribution @xmath225 becomes dominant .
similarly , we expect that the distribution of the phase of ab oscillations is also strongly affected by the bias mode .
if the bias mode is chosen such that the classical contribution @xmath225 vanishes , the phase @xmath46 strongly fluctuates _ even for weak interactions_. it would be very interesting to check this surprising conclusion experimentally .
let us now consider the fluctuations of the ab phase . since the scattering theory turned out to be very useful for the discussion of the nonlinear/ rectified current through a chaotic quantum dot , we extend this theory to rings .
we make two key assumptions ( discussed in the appendix in more detail ) that the magnetic flux through the annulus of the ring is smaller then the flux quantum @xmath377 and that the mean free path @xmath378 , the radius @xmath379 , the width of the ring @xmath380 and the contacts @xmath381 satisfy the condition @xmath382 . in this case
the rmt can be applied to such chaotic rings as well . unlike the experiments on large open rings with high aspect ratio @xmath383,@xcite the recent experiments @xcite are performed in rings of submicron size , which are effectively zero - dimensional .
the treatment of such rings is similar to chaotic quantum dots , and the fluctuations of @xmath13 can be expressed in terms of the diffuson @xmath212 and the cooperon @xmath213 , see eq .
( [ eq : main ] ) .
the only problem is to find the expression for the effective number of channels as a function of magnetic field , similar to eq .
( [ eq : channels ] ) . channels .
the model consists of a quantum dot with @xmath54 channels combined with a ballistic arm with @xmath384 channels.,title="fig:",width=226 ] + the model we propose for a chaotic ab ring combines chaos and a ring geometry : a chaotic dot is attached to a long ballistic arm which serves to include an ab flux large compared to the fraction of the flux through the sample .
this model is shown in fig .
[ fig : model ] , where the ring with @xmath385 ballistic channels in the contacts 1,2 is modeled by a dot with @xmath386 channels and a ballistic arm with @xmath384 channels in contacts 3,4 .
the parameter @xmath387 , the ratio of @xmath388 to the total number of channels @xmath54 , can vary between 0 when the arm is much narrower then the contacts and 1 in the opposite limit .
the electronic phase is randomized in the quantum dot , but when electrons propagate along the arm their phase is determined by the geometry and applied magnetic field .
this model is a reasonable approximation for the real experiment , it takes into account the long time spent by electron inside the ring and the randomness of its motion .
the discussion of the model and the details of calculation of @xmath389 are presented in the appendix . in experiment
the fourier transform is often taken over the total flux ( or applied magnetic field ) and the flux through the hole @xmath390 can not be separated from the flux through the dot @xmath391
. then the dependence of the diffuson and cooperon on magnetic field is non - periodic , which is indeed observed in the form of nonperiodic fluctuations in the ( non-)linear conductance and phase slips of ab oscillations .
a possible weakness of this model is in its spatial separation of chaotic scattering and the main part of magnetic field , but in the limit when the arm is much wider then the contacts @xmath392 and @xmath393 such a separation is not important and the averaged properties of ab oscillation phase become independent of the arm s width .
if the flux @xmath394 through the dot is much smaller then the flux @xmath395 through the hole , the nonperiodic fluctuations and the periodic ab oscillations are well - separated , which is usually the case in experiment.@xcite in view of this separation we can neglect the flux through the chaotic dot , @xmath396 , to find the statistics of the ab phase .
we assume that the averaging is taken over a magnetic field range containing many ab oscillations but still small compared to the characteristic field of the nonperiodic fluctuations .
in such a simplified model of a chaotic ab ring @xmath288 are given by eq .
( [ eq : dcfinal ] ) with @xmath397 and the parameter @xmath398 . the effective number of channels is @xmath399 using this expression for the @xmath288 we can evaluate the quantum fluctuations of linear conductance in ab rings . at low temperature a typical fluctuation of @xmath346 at @xmath400 is @xmath401 and the amplitude of ab oscillations is @xmath402 , which reaches maximum when the widths of the arm and the contacts are equal , @xmath403 .
the first two moments of @xmath44 can be found analytically .
it is zero on average , @xmath404 , since the numerator and denominator in eq .
( [ eq : tan ] ) are independent random quantities . equation ( [ eq : dcsimple ] ) show that @xmath405 are the same functions of @xmath406 , so that all necessary ingredients can be expressed in terms of the functions @xmath407 and @xmath408 of @xmath409 defined in eqs .
( [ eq : fu ] ) and ( [ eq : fg ] ) .
we omit now the index @xmath212 for brevity and denote the average over magnetic field by @xmath410 , to find @xmath411/ \left[\overline{({\cal f}_{u } ( \phi)+x(\oo)){\cal f}_{g } ( \phi)\cos\phi } \right.\nonumber \\ & & \mbox { } \left.+ \overline{\frac{{\cal f}_{u } ( \phi)}{2}}\overline{{\cal f}_{g } ( \phi)\cos\phi}+ \overline{\frac{{\cal f}_{g } ( \phi)}{2}}\overline{{\cal f}_{u } ( \phi)\cos\phi}\right],\end{aligned}\ ] ] where the function @xmath310 is defined by the setup geometry in eq .
( [ eq : xomega ] ) . again , in the static limit @xmath28 we have @xmath412 and @xmath225 defined by eqs .
( [ eq : fraw ] ) and ( [ eq : x ] ) . in this case
( [ eq : tandelta ] ) can be rewritten as @xmath413\overline{\cos\phi{\cal f}(\phi)}}{\overline{\cos\phi{\cal f}^2(\phi)}}.\end{aligned}\ ] ] in the limits of high , @xmath414 , and low temperature , @xmath415 the asymptotical values of @xmath416 are @xmath417 very important is the case of symmetric contacts , @xmath235 , and antisymmetric bias mode , @xmath111 , which is used in ref .. then @xmath225 vanishes and the average @xmath418 becomes independent of interaction strength and as a function of @xmath16 it is very weak .
that is not the case if @xmath419 , for example , when only one of the voltages changes , @xmath261.@xcite then the statistics of the ab phase becomes temperature and interaction dependent due to the presence of @xmath225 .
the limit @xmath420 corresponds to a uniformly chaotic ring , which we suppose to be closer to the experimental situation .
then the dependence on @xmath54 drops out and the high / low temperature asymptotic read @xmath421 this result clearly demonstrates that the phase of the oscillations is expected to deviate strongly from 0 , especially if the temperature is low and the number of channels in the contacts is diminished .
the temperature is taken into account only in the form of temperature - averaging and the dephasing ( previously considered for nonlinear transport of noninteracting electrons in refs . )
is not included .
we expect our model for chaotic ab rings to work both for experiments at small frequencies @xcite and for large frequencies.@xcite similarly to quantum dots , the generalization on the finite - frequency case is obvious , if we use eq .
( [ eq : xomega ] ) . even in cases
where rmt can not be assured to be valid for open diffusive rings , the dependence of the ab phase on interaction strength , temperature , and number of external channels given by eq .
( [ eq : anglewidearm ] ) should be correct qualitatively .
the experiment of leturcq @xcite is performed in a bias mode @xmath111 when @xmath422 .
( [ eq : anglewidearm ] ) gives @xmath423 .
the phase of the oscillations is evaluated from data according to eq .
( [ eq : tan ] ) over a large range of fields . in experiment
the ab phase is varied continuously as a function of the gate voltage at one of the arms of the ring .
the data demonstrate that the phase @xmath46 indeed changes in a wide range and is usually far from 0 .
this substantiates our conclusion that in the mode when the classical contribution is minimized , @xmath236 , the mesoscopic distribution of @xmath46 is very wide . experiment of angers @xcite varies voltage in a different way , @xmath109 , and therefore has @xmath424 . we would expect the phase @xmath425 to take values closer to @xmath375 and the antisymmetric component of the oscillations be relatively smaller even for large fields .
although phases close to 0 are indeed observed , the field averaging is taken only over first few oscillations . in this range @xmath14 , the numerator in eq .
( [ eq : tan ] ) is still small and grows linearly with magnetic field .
averaging over a larger field - range similar to ref . could not be performed because of the phase slips .
another interesting question is a difference in data @xcite for the relative magnitudes @xmath426 of the second @xmath31 and main harmonic @xmath30 , see eq .
( [ eq : expansion ] ) . in the nonlinear transport regime
this harmonic is small compared to its contribution in the linear transport , @xmath427,@xcite while in ref . they were comparable , @xmath428 .
our model also predicts the mesoscopically averaged contribution of @xmath31 into linear and nonlinear conductance to be comparable with that of @xmath30 .
our approach assumes full quantum coherence of the ring , and probably the difference in data is due to decoherence .
in this paper , we consider mesoscopic chaotic samples ( quantum dots or rings ) and find the statistics of their nonlinear conductance @xmath143 .
this transport coefficient characterizes nonlinear dc current due to dc - bias or a rectified current due to ac bias or photon - assisted transport . for chaotic samples ,
the nonlinear effect is of quantum origin , which is clear from the fact that its ensemble average over similar samples vanishes .
the linear response of the sample in two - terminal measurements is always symmetric with respect to magnetic field inversion .
however , the coulomb interactions lead to magnetic field asymmetry of the nonlinear dc response , which fluctuates due to the electronic interference . for the quantum dots we consider the fluctuations of ( anti ) symmetrized components @xmath8 of the nonlinear conductance . in chaotic rings
the statistics of the phase of ab oscillations in the nonlinear transport regime , closely related to the ratio @xmath429 , is of interest . unlike the linear conductance measurements , in mesoscopic nonlinear transport experiments the way voltages are varied ( `` bias mode '' ) turns out to be important , especially for a weakly interacting sample .
we demonstrate this fact qualitatively and discuss the role of coulomb interactions .
quantitative self - consistent treatment of interactions allows us to consider magnetic - field asymmetry in chaotic quantum dots with many channels . using eqs .
( [ eq : main])([eq : x ] ) we show that the fluctuations of @xmath15 are strongly affected by the geometry of the setup and discuss how the bias mode influences data of recent dc experiments .
another important issue is rectification of ac bias , which is quadratic in applied voltage , random , and asymmetric with respect to the magnetic flux inversion .
the photovoltaic dc current can be due to rectification of external perturbations or quantum pumping by internal perturbations .
both rectification and quantum pumping share the aforementioned properties , and it is important to clearly separate them especially when the frequency of perturbations is high ( nonadiabatic ) .
we consider here only the effects of the external perturbations and discuss the dependence of the fluctuations of @xmath8 on frequency @xmath4 .
we show that the fluctuations of both @xmath14 and @xmath15 , presented in eqs .
( [ eq : vargs])([eq : xgeneral ] ) , decrease monotonically as @xmath41 .
however , contrary to naive expectations , their asymptotical behavior can be very different .
since at high frequencies the response of the dot to the external bias becomes rather capacitive then resistive , the coupling to the nearby gates can be strongly enhanced .
if the experiment is performed in a bias mode where such coupling contributes , the symmetrized @xmath430 can become much larger then @xmath431 valid for a strongly interacting quantum dot .
the same conclusion holds in the weakly interacting limit , when @xmath432 and @xmath433 .
in addition , we show that recent experiments in chaotic aharonov - bohm rings might be considered similarly to quantum dots .
the multiply connected geometry alone leads to ab oscillations , yet the mesoscopic distribution of their phase is expected to be qualitatively similar to that of @xmath434 in quantum dots .
therefore , the bias mode should strongly affect the shape of mesoscopic distribution of the ab phase .
the model of an ab ring , which we develop , consists of a dot and a long ballistic arm and takes into account both chaos and a ring geometry . as an application of our model
we consider fluctuations of the ab phase .
unlike the ab phase in the linear conductance , pinned to @xmath435 by the onsager symmetry relations , the fluctuations of the ab phase in nonlinear transport are shown to depend on the bias mode , interaction strength , and temperature .
we thank hlne bouchiat , piet brouwer , renaud leturcq , david snchez , maxim vavilov , and dominik zumbhl for valuable discussions .
we also thank the authors of ref . for sharing their results with us before publication .
this work was supported by the swiss national science foundation , the swiss center for excellence manep , and the strep project subtle .
in this appendix we determine the diffuson and cooperon contributions to the @xmath69-matrix correlators of the random scattering matrix of a chaotic aharonov - bohm ( ab ) ring .
this calculation is performed using random matrix theory ( rmt ) .
first we explain what approximations should be made to ensure validity of rmt .
our starting point is the assumption that the @xmath69-matrix of the ring is uniformly distributed over the unitary group .
this means that the ring is essentially zero - dimensional , similarly to quantum dots .
rmt is applicable if all energy - scales are much smaller then the thouless energy @xmath318 and the total flux through the annulus of the ring is much smaller then @xmath377 .
assume the ring of radius @xmath379 and width @xmath436 to be diffusive with diffusion coefficient @xmath437 . to evaluate @xmath318 we neglect with transversal motion of an electron and
find @xmath438 as a solution to laplace equation along the circumference of the ring .
rmt can be applied to a closed ring if the dimensionless conductance is large , @xmath439 , which is usually satisfied for a weak disorder even if @xmath436 . an open ring with ballistic contacts of the width @xmath381 gains a new energy parameter , the escape rate @xmath440 , where @xmath64 is the total number of ballistic channels .
the scattering matrix @xmath69 is uniformly distributed and independent of the exact positions of the contacts ( and therefore the length of the arms ) if @xmath441 . in this case
the main drop of the potential occurs in the contacts . if a magnetic field is applied , the rmt is valid if the total flux through the annulus of the ring is much less then the flux quantum , @xmath442 .
due to narrow contacts the time - reversal symmetry ( trs ) of the @xmath69-matrix can be broken at a much smaller scale , @xmath443 . since in our rings
@xmath444 , a full crossover to the broken trs can be considered .
how well are these conditions fulfilled in the experiment ? in ref .
chaos was mainly due to diffusive scattering on the boundary and @xmath445 .
the width of the arm is 2 - 4 channels , while the number of channels in the contacts is @xmath446 , estimated from the linear conductance measurements , so @xmath447 . in semiballistic samples of ref .
( obtained by etching , and therefore having diffusive boundary scattering ) @xmath448 and the mean free path is estimated @xmath449m@xmath450 m , the side length .
therefore , we have a similar estimate for the ratio @xmath451 .
although this ratio is not parametrically large due to , e.g. , weak disorder @xmath452 , we believe that such ab rings still can be assumed zero - dimensional due to their good conducting properties together with relatively narrow contacts . in our calculations
we make a further simplification by spatially separating chaotic scattering which randomizes the electronic phase and the long ballistic arm attached to it . to find the correlators of the @xmath69-matrix elements we use a simplified model , see fig .
[ fig : abdot ] , which combines chaos and a ring geometry .
a chaotic @xmath54-channel dot is attached to a long multi - channel ballistic arm with @xmath453 orbital channels .
we assume that the size of the dot @xmath454 and the length of the arm @xmath455 are such that @xmath456 to ensure that in the hierarchy of different fluxes the main flux @xmath395 is concentrated in the region embraced by the arm , the flux through the dot @xmath394 is much smaller , but still much larger then the flux through the cross section of the arm .
the amplitude of ab oscillations depends on the width of the arm @xmath457 .
the wider the arm ( relatively to the contacts ) the closer the results should be to a uniformly chaotic ring . for the case
when @xmath420 we expect it to be valid for the chaos uniformly distributed over the ring .
indeed , in this case an electron makes @xmath458 windings around the arm before exiting . in this appendix
it is more convenient for us to work with an energy - dependent matrix @xmath68 , and the final transformation to time - representation is rather obvious .
the total scattering matrix @xmath69 is of size @xmath459 due to scattering channels in the contacts 1 and 2 .
chaotic scattering in the @xmath54-channel quantum dot is characterized by the @xmath460 matrix @xmath461 .
the scattered electron can either exit the sample through the @xmath385 channels ( projection operator @xmath462 ) or propagate into the arm with @xmath384 channels .
electrons propagate through this arm ballistically and gain phases which depend on the flux through the hole . in the absence of backscattering the electronic amplitudes at energy @xmath72
are related to the path length @xmath455 and magnetic field phase @xmath365 : @xmath463 the scattering matrix of the arm is @xmath464 .
each time an electrons enters the arm either through the third or fourth lead , the matrix @xmath465 contributes to the scattering amplitude of the process .
the total scattering matrix @xmath69 is determined from the following equation : @xmath466 where multiple @xmath467 windings are taken into account .
both @xmath468 and @xmath469 are field and energy dependent .
once we are interested only in pair correlators of @xmath470 for @xmath471 , the diffuson @xmath212 and cooperon @xmath213 of our scattering matrix are expressed via correlators of the dot , @xmath472,@xmath473 , and tr @xmath474 .
the correlators of the @xmath461-matrix are known , see eq . ( [ eq : cum1 t ] ) for their time representation , and for @xmath212 and @xmath213 we derive @xmath475 the flux penetrating the dot is denoted as @xmath394 and the phase @xmath476 gained in the arm depends on the flux @xmath395 through the hole .
the traces read @xmath477}.\end{aligned}\ ] ] since we assumed that the area of the arm is small compared to that of the dot , the energy - dependence of eq .
( [ eq : tr ] ) can be neglected compared to that of @xmath478 in eq .
( [ eq : dcu ] ) .
we also assumed that since the arm is much longer than the size of the dot , @xmath479 , the phases @xmath480 of open trajectories in the arm correspond to the flux @xmath481 through the hole .
therefore , the effective number of channels @xmath288 , similar to eq .
( [ eq : channels ] ) for quantum dots is @xmath482 the energy - dependent cooperon and diffuson in energy representation are given by @xmath483 , x=\coop,\diff$ ] .
notice that when @xmath291 the cooperon @xmath213 is nonperiodic in the total flux @xmath484 due to finite flux through the material of the sample , @xmath485 .
c. a. marlow , r. p. taylor , m. fairbanks , i. shorubalko , and h. linke , phys .
* 96 * , 116801 ( 2006 ) .
r. leturcq , d. snchez , g. gtz , t. ihn , k. ensslin , d. c. driscoll , and a. c. gossard , phys .
lett . * 96 * , 126801 ( 2006 ) .
d. m. zumbhl , c. m. marcus , m. p. hanson , a. c. gossard , phys .
lett . * 96 * , 206802 ( 2006 ) .
a. v. andreev and l. i. glazman , phys .
* 97 * , 266806 ( 2006 ) .
b. braunecker , d. e. feldman , and f. li , phys .
b * 76 * , 085119 ( 2007 ) .
a. lfgren , c. a. marlow , t. e. humphrey , i. shorubalko , r. p. taylor , p. omling , r. newbury , p. e. lindelof , and h. linke , phys .
b * 73 * , 235321 ( 2006 ) .
a. a. bykov , z. d. kvon , l. v. litvin , yu .
v. nastaushev , v. g. mansurov , v. p. migal , and s. p. moschenko , pisma zh .
. fiz . * 58 * , 538 ( 1993 ) [ jetp lett . * 58 * , 543 ( 1993 ) ] ; a. a. bykov , l. v. litvin , n. t. moshegov , and a. i. toropov , superlattices microstruct . * 23 * , 1285 ( 1998 ) .
m. g. vavilov , l. dicarlo , and c. m. marcus , phys . rev .
b * 71 * , 241309(r ) ( 2005 ) showed that the interference between parasitic rectification and quantum pumping produces current asymmetric to @xmath486 even in noninteracting dot . | we investigate the nonlinear current - voltage characteristic of mesoscopic conductors and the current generated through rectification of an alternating external bias . to leading order in applied voltages both the nonlinear and the rectified current are quadratic .
this current response can be described in terms of second order conductance coefficients and for a generic mesoscopic conductor they fluctuate randomly from sample to sample .
due to coulomb interactions the symmetry of transport under magnetic field inversion is broken in a two - terminal setup .
therefore , we consider both the symmetric and antisymmetric nonlinear conductances separately .
we treat interactions self - consistently taking into account nearby gates .
the nonlinear current is determined by different combinations of second order conductances depending on the way external voltages are varied away from an equilibrium reference point ( bias mode ) .
we discuss the role of the bias mode and circuit asymmetry in recent experiments . in a photovoltaic experiment
the alternating perturbations are rectified , and the fluctuations of the nonlinear conductance are shown to decrease with frequency .
their asymptotical behavior strongly depends on the bias mode and in general the antisymmetric conductance is suppressed stronger then the symmetric conductance .
we next investigate nonlinear transport and rectification in chaotic rings . to this extent
we develop a model which combines a chaotic quantum dot and a ballistic arm to enclose an aharonov - bohm flux . in the linear two - probe conductance the phase of the aharonov - bohm oscillation is pinned while in nonlinear transport phase rigidity is lost .
we discuss the shape of the mesoscopic distribution of the phase and determine the phase fluctuations . |
charge symmetry breaking ( csb ) in the @xmath8 interaction , which amounts to the difference between the @xmath9 and the @xmath10 interactions , can not be studied in free space for lack of direct or indirect @xmath9 scattering data and also because none of the two possible @xmath11 @xmath9 and @xmath10 systems is bound .
furthermore , it can not be inferred from the only three - body @xmath0 hypernucleus known to date , the @xmath12 , in which csb effects are highly suppressed . however , the two four - body @xmath11 @xmath0 hypernuclei , with @xmath13 and with @xmath14 , each one with two particle - stable levels @xmath4 and @xmath5 , suggest substantial csb splitting of the @xmath1 hypernuclear ground state ( see fig . [
fig : a=4 ] ) : @xmath15 ) level diagram ( in mev ) .
the @xmath4 @xmath0 separation energies @xmath16 , loosely termed @xmath0 binding energies , taken from a recent measurement at mami @xcite for and from emulsion work @xcite for , are marked under the @xmath4 energy levels .
the @xmath17 separation energies follow from @xmath18-ray measurements of the excitation energies @xmath19 @xcite denoted by arrows , and are marked above the @xmath17 energy levels .
csb splittings are shown to the right of the levels .
results from recent measurements are highlighted in red in the online version .
figure adapted from @xcite.,scaledwidth=70.0% ] until recently , this relatively large observed csb splitting could not be reproduced in _ ab - initio _ four - body calculations with the widely used hyperon - nucleon ( @xmath20 ) nijmegen soft - core meson exchange models nsc97@xmath21 @xcite ; see refs .
. a maximal value of @xmath22 kev was reached in model nsc97@xmath23 @xcite .
the csb model used in these past calculations is the @xmath24 mixing model of dalitz and von hippel @xcite . in this model , the pure - isospin @xmath12 @xmath25 and @xmath26 @xmath27 octet hyperons which share the @xmath28 central point of the su(3)@xmath23 octet , as shown in fig .
[ fig : octet ] , are admixed by csb in forming the physical @xmath0 and @xmath29 hyperons .
the model relates then the mass - mixing matrix element @xmath30 to electromagnetic mass differences of su(3)@xmath23 octet baryons : @xmath31=1.143\pm 0.040~{\rm mev}. \label{eq : deltam}\ ] ] lattice qcd calculations yield so far only half of this value for the mass - mixing matrix element @xcite .
the reason apparently is the omission of qed from these calculations @xcite .
octet baryons with their underlying leading quark structure .
note the @xmath12 @xmath25 and @xmath26 @xmath27 hyperons , sharing the @xmath28 central point , which are admixed by csb in the physical @xmath0 and @xmath29 hyperons.,scaledwidth=70.0% ] interaction diagram describing a si @xmath32 interaction followed by a csb @xmath24 mass - mixing vertex.,scaledwidth=70.0% ] the mass - mixing matrix element ( [ eq : deltam ] ) serves as insertion in @xmath8 csb diagrams generated by the @xmath33 strong - interaction ( si ) coupling potential @xmath34 , as shown in fig .
[ fig : diag_csbls_h ] , leading to a concrete expression of @xmath35 @xmath8 matrix elements in terms of @xmath36 @xmath33 matrix elements @xcite : @xmath37 where the z component of the isospin pauli matrix @xmath38 assumes the values @xmath39 for protons and neutrons , respectively , the isospin clebsch - gordan coefficient @xmath40 accounts for the @xmath41 amplitude in the @xmath42 @xmath43 state , and the space - spin structure of this @xmath43 state is taken identical with that of the @xmath44 state embracing @xmath35 . the csb scale coefficient 0.0297 in ( [ eq : ome ] )
follows from the @xmath24 mass - mixing matrix element @xmath45 given above , @xmath46 where the factor 2 accounts for the two possibilities of inserting @xmath47 in fig .
[ fig : diag_csbls_h ] , to the left of the si circle or to its right ( as drawn ) . and excitation energies @xmath48 in ncsm calculations of @xmath49 states @xcite as a function of @xmath50 , using lo @xmath51eft @xmath20 interactions with cutoff 600 mev @xcite , including ( left ) or excluding ( right ) @xmath52 coupling . ]
since the charge symmetric si @xmath33 coupling , according to eq .
( [ eq : ome ] ) , is the chief provider of the csb @xmath8 matrix element , it is natural to ask how strong the @xmath53 coupling is in realistic microscopic @xmath20 interaction models . in fig .
[ fig : lo ] we show results of no - core shell - model ( ncsm ) calculations of levels @xcite , using the bonn - jlich leading - order ( lo ) chiral effective field theory ( @xmath51eft ) @xmath20 si potential model @xcite , in which @xmath33 coupling is seen to contribute between 3 to 4 mev to the total binding of and almost 40% of the @xmath54 excitation energy @xmath48 .
a similar effect on @xmath48 also occurs in the nijmegen nsc97 models @xcite .
recall that in a meson exchange model , one - pion exchange ( ope ) , forbidden by isospin in the si @xmath8 diagonal potential , contributes as strongly as possible to the @xmath33 coupling potential .
with si @xmath33 potential energy contributions of order 10 mev @xcite , and with a csb scale of order 3% , eq .
( [ eq : ome ] ) could yield csb contributions of order 300 kev . as shown below
, the bonn - jlich lo @xmath51eft @xmath20 interaction potentials @xcite are able to produce this order of magnitude by applying eq .
( [ eq : ome ] ) to each one of the @xmath33 @xmath36 components in this lo version .
disregarded in this procedure are csb contributions arising from meson mixings , such as @xmath55 and @xmath56 .
these were found negligible , @xmath57 in four - body @xmath58 calculations by coon et al .
@xcite and are disregarded here .
mixing contribution to @xmath7 is opposite in sign to the positive @xmath59 exchange contribution from @xmath6 mixing @xcite . ]
the present work extends our letter report @xcite on csb level - splitting calculations in the @xmath1 mirror hypernuclei , adding calculational details , and furthermore comparing the csb splittings derived from these _ ab initio _ calculations with those derived by a straightforward evaluation of ope csb contributions .
the paper is organized as follows : in sect .
[ sec : meth ] we review briefly the bonn - jlich lo @xmath51eft approach followed in our ncsm four - body calculations , as well as providing details of the application of this ncsm technique .
results of these calculations , updating and extending those of ref .
@xcite , are presented in sect .
[ sec : res ] , with further discussion centered on the role of ope in sect .
[ sec : disc ] .
the paper ends with a brief summary and outlook in sect .
[ sec : sum ] .
the version of the ncsm approach which is particularly suitable for dealing with few - body systems employs translationally invariant harmonic - oscillator ( ho ) bases formulated in relative jacobi coordinates @xcite in which two - body and three - body interaction matrix elements are evaluated .
antisymmetrization is imposed with respect to nucleons , and the resulting hamiltonian is diagonalized in a finite four - body ho basis , admitting all ho excitation energies @xmath60 , @xmath61 , up to @xmath50 ho quanta .
this ncsm nuclear technique was extended recently to light hypernuclei @xcite and is applied here in a particle basis , with full account of the different masses within baryon iso - multiplets , to the and mirror hypernuclei , using momentum - space chiral model interactions specified in sect .
[ subsec : lo ] .
some technical details of the present application of the ncsm methodology to the @xmath62=4 mirror hypernuclei are relegated to the unpublished appendix a. while it was possible to obtain fully converged binding energies , with kev precision , for the @xmath62=3 core nuclei @xmath63h and @xmath63he , it was not computationally feasible to perform calculations with sufficiently large @xmath64 to demonstrate convergence for and . in these cases extrapolation to an infinite model space , @xmath65 , had to be employed . extrapolated energy values @xmath66 are obtained in the present work by fitting an exponential function , @xmath67 with parameters @xmath62 and @xmath68 , to @xmath69 sequences in the vicinity of the variational minima with respect to the ho basis frequency @xmath70 .
the reliability of such extrapolations is then reflected in the independence of @xmath66 of the frequency @xmath70 . in our fitting procedure ,
only the last three @xmath64 values which are the most reliable ones , were used . between @xmath71 and @xmath72 , for @xmath4 ( upper curves ) and for @xmath5 ( lower curves ) on the model - space size parameter @xmath64 for ho values of @xmath73 around the variational minima , together with their extrapolated values , in _ ab initio _ ncsm calculations using lo chiral eft coupled - channel @xmath20 potentials @xmath36 , with cutoff momentum @xmath0=600 mev @xcite , plus @xmath35 derived from @xmath36 using eq .
( [ eq : ome]).,scaledwidth=70.0% ] it is worth noting that the present work focuses on _ differences _ @xmath74 of @xmath0-hyperon separation energies in and , obtained as @xmath75- [ e({}^3\text{h})-e({}^4_\lambda\text{h})],\ ] ] where converged energy values of @xmath63h and @xmath63he are used together with extrapolated energy values for and . in general , the differences @xmath76 are much more stable as function of @xmath64 than the absolute energies are .
this is demonstrated in fig . [
fig : csbnmaxom ] where the dependence of the separation - energy differences @xmath74 , for the @xmath4 ( upper curves ) and the @xmath5 ( lower curves ) states , on the size of the model space is shown for ho frequencies @xmath70 around the variational minima of absolute energies at @xmath77 mev for @xmath78=0(1 ) , together with their extrapolated values .
the values of @xmath74 exhibit fairly weak @xmath64 and @xmath70 dependence compared to the behavior of the absolute energies , and to a lesser extent the behavior of the @xmath0 separation energies , and the employed extrapolation scheme is found sufficiently robust for our purposes . with regard to the use of @xmath79 extrapolated values based on the last three @xmath64 values
, it was found that including the last four @xmath64 values in the fit resulted in @xmath76 values that differed by @xmath80 kev .
@xmath51eft interactions are used throughout this work , with n3lo @xmath81 and n2lo @xmath82 interactions , @xcite respectively , both with momentum cutoff @xmath83 mev . for the si
@xmath20 coupled - channel potentials @xmath36 we use the bonn - jlich su(3)-based lo @xmath51eft approach @xcite plus @xmath35 evaluated from @xmath36 by using eq .
( [ eq : ome ] ) . in principle
, the power counting underlying the eft scheme allows to include two @xmath8 csb contact terms , as done in n3lo @xmath81 versions to account quantitatively for the charge dependence of the low - energy @xmath81 scattering parameters @xcite .
given , however , that low - energy @xmath10 cross sections are poorly known and @xmath9 scattering data are unavailable , the corresponding low - energy constants can not be determined , unless they are fitted to the two csb splittings @xmath84 of the @xmath1 hypernuclear mirror levels , in which case the @xmath1 csb calculation reduces to tautology .
this unfortunate occurrence can not be remedied by going from lo to nlo @xmath51eft @xmath20 potentials . for this reason , and anticipating that hypernuclear csb is driven by the relatively long - range ope , we disregard csb contact terms .
the @xmath51eft potentials @xmath36 are regularized in momentum space , using the standard choice @xcite @xmath85 in order to remove high - energy components of the hadronic fields involved . in lo , @xmath36 consists of regularized pseudoscalar ( ps ) @xmath86 , @xmath87 and @xmath88 meson exchanges with coupling constants constrained by su(3)@xmath23 , plus five central interaction low - energy constants ( also called contact terms ) simulating the short range behavior of the @xmath20 coupled channel interactions , all of which are regularized according to ( [ eq : cutoff ] ) with a cutoff momentum @xmath89 , varied from 550 to 700 mev .
two of the five contact terms connect @xmath8 to @xmath90 in spin - singlet and triplet @xmath91-wave channels , and are of special importance for the calculation of csb splittings .
the dominant meson exchange interaction is ope which couples the @xmath8 channel exclusively to the @xmath11 @xmath90 channel .
@xmath87-meson exchange also couples these two @xmath20 channels .
this lo @xmath36 ( @xmath92 ) reproduces reasonably well , with @xmath93 , the scarce @xmath20 low - energy scattering data .
it also reproduces the binding energy of , with a calculated value @xmath16()=110@xmath9410 kev for cutoff 600 mev @xcite , consistent with experiment ( 130@xmath9450 kev @xcite ) and with faddeev calculations reported by haidenbauer et al .
isospin conserving matrix elements of @xmath92 , given in momentum space , are evaluated here in a particle basis with full account of mass differences within baryon iso - multiplets , while isospin breaking @xmath95 @xmath96 and @xmath97 @xmath98 transitions are suppressed .
the coulomb interaction between charged baryons ( @xmath99 , @xmath100 ) is included .
calculations consisting of fully converged @xmath63h and @xmath63he binding energies , and ( , ) @xmath4 and @xmath17 binding energies extrapolated to infinite model spaces from @xmath101 for @xmath102 are reported in the next section .
the calculated binding energies of the core nuclei , 8.482 mev for @xmath63h and 7.720 mev for @xmath63he , reproduce very well the known binding energies .
the @xmath82 interaction , was excluded from most of the hypernuclear calculations after verifying that , in spite of adding almost 80 kev to the @xmath0 separation energies @xmath103 and somewhat less to @xmath104 , its inclusion makes a difference of only a few kev for the csb splittings @xmath105 in both the @xmath4 and @xmath5 states .
this section is divided to two parts , one in which the explicit csb potential @xmath35 of eq .
( [ eq : ome ] ) is excluded , in order to allow comparison with past calculations , and one in which @xmath35 is generated from the lo @xmath51eft @xmath106 strong interactions used here .
we start by comparing in table [ tab : b_l(l4h0+nocsb ) ] ( @xmath4 ) and table [ tab : b_l(l4h1+nocsb ) ] ( @xmath5 ) our ncsm calculations to nogga s yakubovsky - equations calculations for , both using the same lo @xmath51eft @xmath20 interactions with no explicit csb potential @xmath35 , and also to nogga s recent calculations using nlo @xcite . with uncertainties in calculated @xmath16 values arising from different @xmath81 input in different lo calculations , and also from the suppressed @xmath82 interaction , all of which
are conservatively estimated to be of the order of @xmath1070.1 mev , we cite lo results up to the first decimal point . .@xmath103
( ) values calculated in @xmath51eft approaches , without explicit @xmath35 , for various cutoff momenta @xmath0 ( in mev ) .
@xmath108 ( ) stands for the mean@xmath94spread of these values . [
cols="<,^,^,^,^,^",options="header " , ] [ tab : opeexc600 ] for the finite values of the cutoff @xmath109 listed in tables [ tab : opegs550 ] , [ tab : opegs600 ] , [ tab : opeexc550 ] and [ tab : opeexc600 ] , the dependence of the csb ope contributions on @xmath109 for a given @xmath110 is weak to moderate , and the limiting case of @xmath111 poses no convergence problem .
however , once @xmath109 is increased beyond roughly 700 mev , the ore contribution may no longer be ignored , with a @xmath112-subtracted central contribution that augments the ope @xmath112-subtracted central contribution and a tensor contribution that reduces considerably the ope tensor contribution .
we conjecture that the failure of the nsc97 @xmath20 models to reproduce the large size of the observed g.s .
csb splitting @xmath7 arises from a strong cancellation between the ope and ore tensor csb contributions which in these models overshadow the central csb contributions . finally , the dependence of the total , ` updated dvh ' ope csb contribution on the strong - interaction cutoff @xmath110 , for a given @xmath113 , is considerably weaker for the @xmath4 than that given in table [ tab : csb ] using eq .
( [ eq : ome ] ) to derive @xmath35 .
in this work we discussed the extension of the ncsm from few - body nuclear to few - body hypernuclear applications and provided details of our recent letter publication on _ ab initio _ calculations of csb in the @xmath62=4 mirror hypernuclei @xcite .
these calculations are the first microscopic calculations to generate a large positive value of @xmath7 commensurate with experiment , although with a considerable momentum - cutoff dependence within the bonn - jlich lo @xmath51eft coupled - channel @xmath20 potential model @xcite .
the calculational extrapolation uncertainties involved in the evaluation of @xmath7 were estimated to be in the range of 10 to 20 kev at most . in the bonn - jlich @xmath51eft approach
, the relatively large value derived for @xmath114 arises from the @xmath115 ct of the si @xmath116 coupling potential , appearing to have no relationship with the large ope csb contribution anticipated by dvh @xcite .
this is a direct consequence of using the relationship given by eq .
( [ eq : ome ] ) between si and csb . by removing the short - range @xmath117 term from the ope @xmath8 csb potential , and using a dwba - like evaluation of this csb potential
, we were able to recover the dvh original estimate of the central ope csb contribution , updated to present - day coupling constants .
furthermore , choosing a cutoff @xmath110=600 mev , which is closer to reproducing @xmath118 than the lower cutoff considered here , as large values as @xmath107200 kev for @xmath7 , and small and negative values @xmath11940 kev for @xmath120 , emerge for the combined central plus tensor ope csb contribution , in striking agreement with experiment .
similar estimates were obtained by one of the authors @xcite using a @xmath106 effective @xmath36 to which eq .
( [ eq : ome ] ) was applied to generate the corresponding @xmath35 .
future applications of the ncsm to @xmath121-shell hypernuclei are desirable , in view of the few csb mirror - level splittings known in this mass range @xcite .
the lesson of this latter work is that genuine csb splittings become smaller as one goes to heavier hypernuclei . in this respect ,
given the particularly large observed value of @xmath7 in the @xmath62=4 mirror hypernuclei considered in the present work , these hypernuclei provide a unique test ground for csb models beyond nuclear physics .
the starting point of the _ ab initio _ ncsm calculations is the hamiltonian for a system of nonrelativistic nucleons and hyperons interacting by realistic two - body @xmath81 and @xmath20 , and also three - nucleon interactions : @xmath122 in the present work , considering the @xmath62=4 mirror hypernuclei , the momenta @xmath123 , masses @xmath124 and coordinates @xmath125 for @xmath126 correspond to nucleons and those for @xmath127 to hyperons . the hamiltonian form ( [ eq : h ] ) is then rewritten in terms of _ relative _ jacobi coordinates , momenta and their associated masses .
there are several different sets of jacobi coordinates , the first of which is defined by @xmath128 where @xmath129 .
this particular set is a natural one for implementing antisymmetrization with respect to nucleons , and is subsequently used for diagonalization of the hamiltonian . here , @xmath130 is proportional to the center of mass coordinate of the @xmath62-baryon system and @xmath131 ( @xmath132 ) is proportional to the relative coordinate of the @xmath133 baryon with respect to the center of mass of @xmath134 baryons . the kinetic energy term in eq .
( [ eq : h ] ) is then rewritten in terms of jacobi cooredinates ( [ eq : cset1 ] ) : @xmath135 since the various interactions @xmath136 in ( [ eq : h ] ) do not depend on @xmath130 , the center of mass kinetic energy can be omitted from ( [ eq : ht ] ) , and one can use an ho basis depending on coordinates @xmath137 , @xmath138 and @xmath139 , e.g. @xmath140 here @xmath141 , @xmath142 are ho quantum numbers corresponding to coordinate @xmath137 describing the relative motion of the first two nucleons ; @xmath143 , @xmath144 corresponding to @xmath138 describe the relative motion of the third nucleon with respect to the nucleon pair ; and @xmath145 , @xmath146 associated with @xmath139 describe the relative motion of the hyperon with respect to the three - nucleon cluster .
the spin quantum numbers referring to single - particle states are omitted , @xmath147 is the spin of the two - nucleon pair , and the @xmath148 quantum numbers denote respective angular momenta .
we work in the isospin basis , @xmath149 is the isospin of the nucleon pair , and the nucleon single - particle isospin is also suppressed in ( [ eq : b1 ] ) .
the hyperon isospin quantum number @xmath150 holds for @xmath0 and @xmath151 hyperons , respectively , thereby allowing for explicit admixtures of @xmath151 hyperons into @xmath0 hypernuclear states , induced by the @xmath152 @xmath33 coupling potential .
the basis ( [ eq : b1 ] ) is truncated in ncsm calculations by requiring that the total number of ho quanta does not exceed a chosen value @xmath64 , @xmath153 thereby defining the size of the model space .
moreover , all ho wave functions in ( [ eq : b1 ] ) depend on a single ho frequency @xmath70 which is a free parameter in ncsm calculations . the basis ( [ eq : b1 ] ) is antisymmetric with respect to exchanging nucleons @xmath154 and @xmath155 upon requiring @xmath156 for the two - nucleon system .
it is , however , not antisymmetric with respect to nucleon exchanges @xmath157 and @xmath158 .
the procedure of fully antisymmetrizing the three - nucleon cluster in the basis ( [ eq : b1 ] ) , recalling that it is disconnected from the hyperon quantum numbers , is described in detail e.g. in ref .
the resulting fully antisymmetric three - nucleon cluster basis elements can be expanded as linear combinations of the original basis ( [ eq : b1 ] ) .
incidentally , the set of coordinates ( [ eq : cset1 ] ) is also suitable for evaluating three - nucleon interaction matrix elements which are naturally expressed as functions of the jacobi coordinates @xmath137 and @xmath138 @xcite . the basis ( [ eq : b1 ] ) is , however , inappropriate for evaluating two - body interaction terms .
another set of jacobi coordinates suitable for basis expansion when @xmath81 and @xmath20 interaction matrix elements are calculated is obtained by keeping to @xmath130 , @xmath137 and introducing two new variables , @xmath159 a basis depending on coordinates @xmath137 , @xmath160 , @xmath161 , with two - body subclusters , may be defined e.g. as @xmath162 where , similarly to ( [ eq : b1 ] ) , the ho state @xmath163 associated with the coordinate @xmath137 describes the nucleon pair and the ho state @xmath164 associated with @xmath165 corresponds to the relative - coordinate hyperon nucleon channel , with @xmath166 , @xmath167 and @xmath168 standing for the spin , total angular momentum and isospin of the @xmath20 pair , respectively .
the ho state @xmath169 associated with the coordinate @xmath161 describes the relative motion of the @xmath81 and @xmath20 clusters .
properties of ho wave functions and jacobi coordinates allow basis elements defined in ( [ eq : b1 ] ) to be expanded in basis ( [ eq : b2 ] ) as follows : @xmath170 where the orthogonal transformation between the jacobi coordinates @xmath171 and @xmath172 was employed , and @xmath173 is the general ho bracket for two particles , defined e.g in ref .
@xcite . here ,
@xmath174 and @xmath175 are the nucleon and hyperon ( @xmath176 , @xmath151 ) masses defined as @xmath177 with @xmath178 , @xmath179 , @xmath180 , @xmath181 , and @xmath182 denoting the masses of the neutron , proton , @xmath183 , @xmath29 , and @xmath184 hyperons , respectively , and @xmath185 is the projection of the total isospin t , @xmath186 for ( , ) respectively .
the transformation ( [ eq : rcpl ] ) conserves the total @xmath78 and @xmath187 and also , quite importantly , the total number of ho quanta , @xmath188 using the expansion ( [ eq : rcpl ] ) , it is straightforward to evaluate matrix elements of two - body interactions in the basis ( [ eq : b1 ] ) , @xmath189 where the matrix elements on the right hand sides are diagonal in all quantum numbers of the states ( [ eq : b2 ] ) except for @xmath141 , @xmath142 and @xmath190 , @xmath191 , respectively , for isospin conserving interactions .
equally straightforward is the evaluation of two - body interactions defined in momentum space , since transformations analogous to those in ( [ eq : cset1 ] ) and ( [ eq : cset2 ] ) can be introduced for momenta @xmath123 by substituting @xmath192 .
both local and non - local interactions can be accommodated within the ncsm methodology .
realistic @xmath81 and @xmath20 interactions are , however , usually defined in the particle basis , not in the isospin basis . to evaluate the corresponding matrix elements of @xmath193 between good - isospin basis states ( [ eq : b2 ] ) we use the following prescription @xmath194 here , only the isospin quantum numbers of states ( [ eq : b2 ] ) are displayed .
the basis elements are decomposed via clebsch
gordan coefficients and the potential matrix elemets are evaluated between two - nucleon states @xmath195 with single - nucleon isospin projections @xmath196 and @xmath197 . in this procedure
the isospin breaking transitions @xmath198 are suppressed , but the resulting isospin - basis defined @xmath81 interaction depends parametrically on the isospin of the @xmath20 cluster , as well as on the total isospin and its projection .
similarly , a particle - basis defined @xmath20 interaction @xmath199 is evaluated as @xmath200 where the potential matrix elements are evaluated between hyperon
nucleon states @xmath201 with @xmath202 and @xmath197 the isospin projections of hyperon @xmath203 and nucleon @xmath204 , respectively .
again , the isospin - breaking transitions @xmath205 are suppressed .
this procedure gives excellent agreement with particle - basis calculations as demonstrated in ref .
@xcite . for the @xmath62=3,4 hypernuclear systems ,
the difference between calculated total energies in particle basis and isospin basis using relations ( [ eq : vnnavg ] ) and ( [ eq : vnyavg ] ) was found to be only few kev .
we are grateful to petr navrtil for helpful advice on extensions of nuclear - physics ncsm codes , to johann haidenbauer , and andreas nogga for providing us with the input lo eft @xmath20 potentials used in the present work , and to nir barnea , ji mare and ulf meiner for useful discussions on issues related to this work .
the research of d.g . was supported by the granting agency of the czech republic ( gacr ) , grant no . p203/15/04301s .
j. haidenbauer , u .- g .
meiner , a. nogga , h. polinder , in _ topics in strangeness nuclear physics _ , lecture notes in physics * 724 * , eds .
p. bydovsk , j. mare , a. gal ( springer , new york , 2007 ) , pp . 113 - 140 . | charge symmetry breaking ( csb ) in the @xmath0-nucleon strong interaction generates a charge dependence of @xmath0 separation energies in mirror hypernuclei , which in the case of the @xmath1 mirror hypernuclei @xmath2 ground states is sizable , @xmath3 kev , and of opposite sign to that induced by the coulomb repulsion in light hypernuclei .
recent _ ab initio _ calculations of the ( , ) mirror hypernuclei @xmath4 and @xmath5 levels have demonstrated that a @xmath6 mixing csb model due to dalitz and von hippel ( 1964 ) is capable of reproducing this large value of @xmath7 . these calculations are discussed here with emphasis placed on the leading - order chiral eft hyperon - nucleon bonn - jlich strong - interaction potential model used and the no - core shell - model calculational scheme applied .
the role of one - pion exchange in producing sizable csb level splittings in the @xmath1 mirror hypernuclei is discussed .
hypernuclei , hyperon - nucleon interactions , charge symmetry breaking |
during recent years , quantum discord [ 1 - 14 ] has become the main focus of fundamental research in the discipline of quantum information theory .
it quantifies the total non - classical correlations in a quantum state .
recently , quantum discord for multipartite system has been investigated [ 15 - 17 ] .
its extension to @xmath3-partite @xmath0 state is given by jianwei xu [ 18 ] .
the dynamics of quantum discord has been extensively studied in different contexts by various authors [ 19 - 45 ] .
it has been shown that quantum discord is more robust than entanglement for bipartite systems .
furthermore , the quantum discord for a qubit - qutrit [ 46 ] and qubit - qudit [ 47 ] systems has been proposed .
quantum discord has been used in studies of quantum phase transition [ 48 , 49 ] and to measure the quantum correlation between relatively accelerated observers [ 50 ] .
the geometric interpretation of the geometric discord has been discussed by yao et al . [ 51 ] .
the lower and upper bounds of quantum discord have also been investigated [ 52 , 53 ] .
its experimental evidences can be seen in references [ 54 - 57 ] .
recently , the geometric measure of quantum discord ( gmqd ) has been proposed [ 58 , 59 ] , which quantifies the amount of non - classical correlations of a state in terms of its minimal distance from the set of genuinely classical states .
however , most of the studies in this connection deal with bipartite quantum states .
it has also been studied over two - sided projective measurements by [ 60 ] .
recently , the geometric quantum discord ( gqd ) for multipartite quantum states have been introduced by jianwei xu [ 61 ] and its lower bound is given .
motivated from the previous studies regarding the influence of environment on the bipartite quantum discord [ 62 ] and its geometric measure [ 63 ] , i have investigated the global quantum discord ( qd ) and its geometric measure ( gqd ) under the influence of external environments for different types of multipartite quantum states .
quantum systems can never be isolated from their environments completely and the interactions with the environment deteriorate the purity of the quantum states .
this gives rise to the phenomenon of decoherence [ 64 ] , which appears when a system interacts with its environment in a irreversible way .
it plays a fundamental role in the description of the quantum - to - classical transition [ 65 ] and has been successfully applied in the cavity qed [ 66 ] .
another familiar aspect , the degradation of entanglement has also been investigated recently by several authors [ 67 - 70 ] , with special attention from the non - inertial perspective .
entanglement in noninertial frames was first time introduced by alsing et al . [ 71 ] .
the subject have attracted much attention during recent years [ 72 - 80 ] .
it has also been investigated under decoherence for bipartite [ 81 - 85 ] , qubit - qutrit [ 86 ] and multipartite [ 87 ] systems . the entanglement dynamics for noninertial observers in a correlated environment is considered in ref . [ 88 ] , where it is shown that correlated noise compensates the loss of entanglement caused by the unruh effect .
recently , i have studied the decoherece dynamics of gmqd and min at finite temperature for non - inertial observers [ 89 ] . in this paper ,
i have investigated the decoherence dynamics of multipartite quantum discord and geometric quantum discord for @xmath0-type initial states in inertial and non - inertial frames .
different decoherence channels are considered parameterized by decoherence parameter @xmath4 such that @xmath5 $ ] .
the lower and upper limits of decoherence parameter represent the fully coherent and fully decohered system , respectively .
it is seen that the quantum discord is more robust than geometric quantum discord under environmental effects .
however , werner-@xmath0 type states are found to be more fragile to decoherence for higher values of @xmath3 .
the depolarizing channel heavily influences the qd and gqd as compared to the amplitude damping channel .
it is also seen that the effect of environment on qd and gqd is much stronger than that of the acceleration of the accelerated observer .
the degradation of qd and gqd happens due to unruh effect .
the quantum discord , a measure of the minimal loss of correlation in the sense of quantum mutual information , can be defined for a bipartite quantum state @xmath6 composed of subsystems @xmath7 and @xmath8 , as [ 1]@xmath9where@xmath10is the quantum mutual information and measures . here
@xmath11tr@xmath12 are the reduced density matrices and @xmath13is the von - neumann entropy .
@xmath14 $ ] is the measure of classical correlations between the two subsystems .
it is defined as the maximum information about one system that can be obtained by performing a set of projective measurements on the other subsystem and the maximum is taken over the set of projective measurements @xmath15 as@xmath16and @xmath17with @xmath18,% \text { } k=1,2\]]recently , rulli et al . [
15 ] have proposed global quantum discord ( qd ) for multipartite quantum states .
qd for an arbitrary @xmath3-partite state @xmath19 under the set of local measurement @xmath20 can be defined as@xmath21\]]where @xmath22 and @xmath23 with @xmath24 one can select the set of von - neumann measurements as @xmath25\]]@xmath26\]]one must find the measurement bases that minimize the qd by varying the angles @xmath27 and @xmath28 , which is achieved by adopting local measurements in the @xmath29 eigenbases for @xmath0-type initial states .
the geometric quantum discord for bipartite x - state system has been proposed by dakic et al [ 58 ] as@xmath30where the minimum is over the set of zero - discord states @xmath31 .
the square of hilbert - schmidt norm of hermitian operators , @xmath32tr@xmath33 $ ] . whereas the geometric quantum discord ( gqd ) for @xmath3-partite state @xmath34 is defined as @xmath35\end{aligned}\]]where
@xmath36 is a locally projective measurement on @xmath37 ....... @xmath38 here in this paper , we consider three different types of initial states as part of the following general @xmath3-qubit werner-@xmath0 initial state of the form @xmath39 where @xmath40 is @xmath41 identity operator , @xmath42 $ ] and @xmath43**(i ) * * let the three partners share a three - qubit werner-@xmath0 initial state as given by @xmath44where @xmath45**(ii ) * * a six - qubit werner-@xmath0 initial state as given by @xmath46where @xmath47**(iii ) * * let the three observers : alice , an inertial observer , bob and charlie , the accelerated observers moving with uniform acceleration , share the following maximally entangled @xmath0-type initial state@xmath48where @xmath49 and @xmath50 are vacuum states and the first excited states from the perspective of an inertial observer respectively .
let the dirac fields , as shown in refs .
[ 90 ] , from the perspective of the uniformly accelerated observers , are described as an entangled state of two modes monochromatic with frequency @xmath51 @xmath52@xmath53and the only excited state is@xmath54where @xmath55 , @xmath56 is the acceleration of @xmath57 observer .
the subscripts @xmath40 and @xmath58 of the kets represent the rindler modes in region @xmath40 and @xmath58 , respectively , in the rindler spacetime diagram ( see ref .
[ 82 ] , fig .
considering that an accelerated observer in rindler region @xmath40 has no access to the field modes in the causally disconnected region @xmath58 and by taking the trace over the inaccessible modes , one obtains the following tripartite state in rindler spacetime as given by [ 78]@xmath59\end{aligned}\]]for the sake of simplicity , the frequency subscripts are dropped and in density matrix formalism , the above state can be written as@xmath60\end{aligned}\]]in order to simplify our calculations , it is assumed that bob and charlie move with the same acceleration , i.e. @xmath61 the interaction between the system and its environment introduces the decoherence to the system , which is a process of the undesired correlation between the system and the environment .
the evolution of a state of a quantum system in a noisy environment can be described by the super - operator @xmath62 in the kraus operator representation as [ 91 ] @xmath63 where the kraus operators @xmath64 satisfy the following completeness relation @xmath65 the kraus operators for the evolution of @xmath3-partite system can be constructed from the single qubit kraus operators by taking their tensor product over all @xmath66 combinations of @xmath67 indices as @xmath68 where @xmath69 corresponds to the number of kraus operators for a single qubit channel . the single qubit kraus operators for different channels are given in table 1 .
using equations ( 7 - 11 ) along with the initial density matrices as given in equations ( 13 , 14 and 19 ) , the qd and gqd for the multipartite system under different environments can be found .
the analytical relations for gqd for all the three types of initial states are given in tables 2 , 3 and 4 respectively .
however , the analytical expressions for qd are too lengthy , therefore , these are not presented in the text and have been explained in figures .
analytical expressions for the quantum discord and geometric quantum discord are calculated for different situations .
influence of different decoherence channels such as amplitude damping , phase damping , depolarizing and phase flip , bit flip and bit - phase flip channels is investigated for @xmath0 type initial states in inertial and non - inertial frames .
the results consist of three parts ( i ) the effect of decoherence parameter @xmath4 on the quantum discord ( qd ) and geometric quantum discord ( gqd ) for three - qubit werner-@xmath0 states ( ii ) the effect of decoherence parameter @xmath4 on qd and gqd for six - qubit werner-@xmath0 states ( iii ) the dynamics of quantum discord and geometric quantum discord for three - qubit @xmath0 state in non - inertial frames influenced by different decoherence channels . in figure 1 ,
the quantum discord is plotted as a function of decoherence parameter @xmath4 for @xmath70 for different noisy channels for three - qubit and six - qubit @xmath0 states .
it is seen that the quantum discord vanishes for @xmath1 in case of three - qubit @xmath0 states and for @xmath71 for six qubit @xmath72 states .
the depolarizing channel has dominant effect on the discord as compared to the amplitude damping channel . whereas the behaviour of flipping channels such as bit flip , phase flip and bit - phase flip channels is symmetrical around 50% decoherence as expected .
it is seen that rapid sudden death of discord occurs in case of phase flip channel
. however , for bit flip channel , no sudden death happens in case of six - qubit states .
the rise and fall of quantum discord is seen for all the flipping channels .
it is shown that states with higher number of qubits ( @xmath73 ) are more prone to decoherence as compared to the states with less number of qubits ( @xmath74 ) . in figure 2 ,
the geometric quantum discord is plotted as a function of decoherence parameter @xmath4 for @xmath70 for different noisy channels , panel ( a ) three - qubit @xmath0 states and panel ( b ) six - qubit @xmath0 states .
it can be seen from the figure that the depolarizing channel heavily influences the geometric quantum discord if compared with amplitude damping channel .
it is seen that quantum discord is more robust than geometric quantum discord .
furthermore , it is seen that the vanishing of geometric quantum discord happens for both cases of the qubit states for bit - flip channel . in figure 3 ,
the geometric quantum discord is plotted as a function of decoherence parameter @xmath4 and acceleration @xmath75 ( a ) amplitude damping channel and ( b ) bit - phase flip channel , for different values of acceleration @xmath75 lower panel , respectively .
it is seen that the behaviour of bit - phase flip channel is symmetrical at @xmath76 .
it is also seen that effect of environment on multipartite geometric quantum discord is much stronger than that of the acceleration of non - inertial frames . in figure 4 ,
the quantum discord and geometric quantum discord are plotted as a function of acceleration @xmath75 for different decoherence channels at @xmath77 . to illustrate the environmental effects , a comparison for different values of decoherence parameter @xmath4
is given in sub figure ( d ) for amplitude damping channel only .
it is seen that the degradation of quantum discord and geometric quantum discord occurs due to unruh effect .
the depolarizing channel influences the discords more heavily as compared to the other channels in non - inertial frames .
it means that the depolarizing channel has the most destructive influence on the discords for multipartite states .
quantum discord ( qd ) and geometric quantum discord ( gqd ) for multipartite quantum states is investigated under the influence of external environments .
different types of initial states such as , werner-@xmath0 type three - qubit and six - qubit states are considered in inertial and non - inertial frames .
dynamics of qd and gqd is investigated under amplitude damping , phase damping , depolarizing and flipping channels .
it is seen that the quantum discord is more robust than geometric quantum discord under environmental effects .
however , werner-@xmath0 type states are found to be more fragile to decoherence for higher values of @xmath3 .
sudden death and birth of discords occur in case of flipping channels .
however , for bit flip channel , no sudden death happens for six - qubit @xmath0 states . whereas depolarizing channel heavily influences the qd and gqd as compared to the amplitude damping channel .
this implies that the depolarizing channel has the most destructive influence on the discords for multipartite states . for the case of accelerated observers
, it is seen that the effect of environment on qd and gqd is much stronger than that of the acceleration of the accelerated observer .
however , the degradation of qd and gqd happens due to unruh effect . in general , qd exhibits more robustness than gqd when multipartite systems are exposed to environment . | quantum discord is an optimal resource for the quantification of classical and non - classical correlations as compared to other related measures
. geometric measure of quantum discord is another measure of quantum correlations .
recently , the geometric quantum discord for multipartite states has been introduced by jianwei xu [ arxiv : quant / ph.1205.0330 ] .
motivated from the recent study [ ann .
phys . 327
( 2012 ) 851 ] for the bipartite systems , i have investigated global quantum discord ( qd ) and geometric quantum discord ( gqd ) under the influence of external environments for different multipartite states . werner-@xmath0 type three - qubit and six - qubit states are considered in inertial and non - inertial settings .
the dynamics of qd and gqd is investigated under amplitude damping , phase damping , depolarizing and flipping channels .
it is seen that the quantum discord vanishes for @xmath1 in case of three - qubit @xmath0 states and for @xmath2 for six qubit @xmath0 states .
this implies that multipartite states are more fragile to decoherence for higher values of @xmath3 .
surprisingly , a rapid sudden death of discord occurs in case of phase flip channel .
however , for bit flip channel , no sudden death happens for the six - qubit states . on the other hand , depolarizing channel heavily influences the qd and gqd as compared to the amplitude damping channel .
it means that the depolarizing channel has the most destructive influence on the discords for multipartite states . from the perspective of accelerated observers
, it is seen that effect of environment on qd and gqd is much stronger than that of the acceleration of non - inertial frames .
the degradation of qd and gqd happens due to unruh effect .
furthermore , qd exhibits more robustness than gqd when the multipartite systems are exposed to environment .
keywords : multipartite correlations ; gqd ; werner-@xmath0 states ; decoherence |
there are many interesting ways to associate an ellipsoid to a convex body .
if the convex body is centrally symmetric then its _ legendre ellipsoid _ is the unique ellipsoid with the same center and the same momenta of inertia with respect to axes through the center .
the _ lwner john ellipsoids _ are the unique ellipsoids of minimal ( resp .
maximal ) volume containing ( resp . contained in ) a given convex body .
the _ banach mazur ellipsoids _ are also a pair of ellipsoids that sandwich a given convex body , but with the property that they are homothetic , with homothecy factor as close to @xmath3 as possible .
see e.g. @xcite for much more information , including many connections to seemingly unrelated problems .
in this paper we consider the problem of approximating a given convex body by an ellipsoid of the same volume so to minimize the volume of the symmetric difference .
actually we consider only the two - dimensional case .
let @xmath4 be the set of all subsets @xmath5 that are compact , convex , centrally symmetric ( @xmath6 ) , and have area @xmath7 .
we endow this set with the _ symmetric difference metric _
( a.k.a .
_ frchet nikodym metric _ ) : @xmath8 let @xmath9 be the subset of @xmath4 formed by ellipses . given @xmath10 ,
consider its distance to @xmath9 , that is , @xmath11 an easy compactness argument shows that the infimum above is attained .
actually we will prove that the infimum is attained at a _
unique _ ellipse ; in other words : [ t.unique ] for every @xmath12 there exists @xmath13 such that @xmath14 for every other @xmath15 . without the assumptions of @xmath0 and @xmath1 being centrally symmetric ,
the question about uniqueness of the best approximation by ellipses was formulated by wodek kuperberg on mathoverflow @xcite in december 2016 .
[ t.unique ] follows from sharper results , which we describe next .
consider the family of ellipses whose axes are the @xmath16 and @xmath17 axes , together with the unit circle .
we parameterize this family by @xmath18 as follows : @xmath19 a function @xmath20 is called _ quasiconvex _
if for every @xmath21 , @xmath22 , @xmath23 , @xmath24 the function is called _ strictly quasiconvex _ if the inequality on the right is always strict .
our crucial result is this : [ t.main ] for every @xmath10 , the function @xmath25 is strictly quasiconvex .
let us put this result in a broader perspective .
we need a few definitions .
let @xmath26 denote the group of area preserving linear maps of @xmath27 .
this group acts on the space @xmath4 by isometries , preserving the subset @xmath9
. note that given any two elements @xmath1 , @xmath28 , we can find some @xmath29 such that , in terms of notation , @xmath30 and @xmath31 for certain @xmath21 , @xmath18 ; moreover if we ask that @xmath32 then such @xmath33 becomes uniquely determined . for each pair @xmath1 , @xmath34 of elements of @xmath9 , we associate a subset @xmath35 $ ] of @xmath9 , called a _
segment_. these sets are uniquely determined by the following requirements : 1 .
[ i.symmetry ] @xmath36 = [ e , f]$ ] ; 2 . [ i.equivariance ] @xmath37 = l[e , f]$ ] for every @xmath29 ; 3 .
if @xmath38 are real numbers then , in terms of notation , @xmath39 = \bigl\ { e_t { \;\mathord{;}\;}s \le t \le u \bigr\}$ ] .
this definition of segments is not artificial .
indeed , in [ s.app ] we exhibit a metric @xmath40 naturally associated to @xmath41 such that @xmath42 is a geodesic metric space whose geodesic segments are exactly those defined above ; moreover @xmath42 is isometric ( modulo a constant factor ) to the hyperbolic plane @xmath43 , and the action of @xmath26 on @xmath9 corresponds to the action on @xmath43 by orientation - preserving mbius transformations .
let us say that a set @xmath44 is _ convex _ if @xmath35 \subset x$ ] for every pair @xmath1 , @xmath45 .
the set of _ extreme points _ of a convex set @xmath46 is defined as : @xmath47 \subset x \
e = e _ * \text { or } f = e _ * \bigr\ } \ , .\ ] ] let us say that a function @xmath48 is _ quasiconvex _ if for every @xmath49 , the sublevel set @xmath50)$ ] is convex . if , in addition , @xmath51)$ ] then we say that @xmath52 is _
strictly quasiconvex_. [ t.main ] can be reformulated as follows : [ t.main2 ] for every @xmath10 , the function @xmath53 is strictly quasiconvex .
note that : [ t.main ] @xmath54 [ t.main2 ] @xmath54 [ t.unique ] . indeed , the first implication follows from the properties of the @xmath26 action , and the second one follows from the fact that a strictly quasiconvex function can not attain its minimum at two different points .
[ t.main ] itself will follow from more precise quantitative information on the area function , which we describe in the next .
let @xmath55 be the subset of @xmath10 satisfying the following conditions : 1 .
[ i.c2 ] the boundary @xmath56 is a @xmath57 curve ; 2 .
[ i.finite ] there is a finite set @xmath58 such that for every @xmath59 , the curves @xmath56 and @xmath60 are transverse ; 3 .
[ i.quadratic ] for every @xmath61 , every tangency between @xmath56 and @xmath60 is quadratic ( i.e , the curves have a first- but not a second - order contact ) , and in particular the number of such tangencies is finite ; 4 .
[ i.envelope ] the tangency points do not belong to the envelope hyperbola @xmath62 of the family of curves @xmath63 .
[ p.trans ] the set @xmath55 is dense in @xmath4 .
the is a simple consequence of the transversality theorem ; see [ s.trans ] .
the following is the core technical result of this paper , and its proof occupies [ s.main ] : [ p.main ] let @xmath64 , and let @xmath58 be the corresponding ( finite ) set of tangency parameters .
then the function @xmath65 defined by @xmath66 is @xmath57 except at @xmath67 . moreover : 1 .
[ i.tang ] for every @xmath61 , the left and right derivatives of @xmath68 at @xmath22 exist in @xmath69 $ ] , at least one of them is infinite , and @xmath70 .
[ i.trans ] for every @xmath59 , we have : @xmath71 for some @xmath72 .
moreover , if @xmath73 is such that @xmath74 then @xmath75 depends only on @xmath76 , and not on @xmath0 nor @xmath22 . the implies that when @xmath64 then the function @xmath25 has no local maximum and therefore is strictly quasiconvex .
more importantly , the provides infinitesimal quasiconvexity properties that are uniform with respect to @xmath0 . this uniformity , combined with the denseness provided by [ p.trans ] , will allow us to extend information from the subset @xmath55 to the whole space @xmath4 ( see [ s.implication ] ) and so we will obtain [ t.main ] .
let @xmath77 denote the projective space of @xmath27 , i.e. the set of all lines through the origin .
let @xmath78\in { \mathbb{p}}^1 $ ] denote the line determined by a nonzero vector @xmath79 .
if @xmath80 is a smooth @xmath3-dimensional submanifold , let @xmath81 be the set of pairs @xmath82)$ ] such that @xmath83 and @xmath84 is tangent to @xmath85 at @xmath86 .
recalling notation , let @xmath87 .
also let @xmath88 , where @xmath89 .
[ l.submanifolds ] @xmath90 and @xmath91 are closed smooth submanifolds of @xmath92 of respective dimensions @xmath93 and @xmath3 .
the proof of [ l.submanifolds ] is postponed .
let @xmath94 , the additive group of real numbers mod @xmath95 .
a _ regular parametrization _ of a smooth simple closed curve @xmath80 is a map @xmath96 that is a smooth diffeomorphism onto @xmath85 . in that case , let @xmath97 denote the map @xmath98)$ ] , which is a smooth diffeomorphism onto @xmath99 .
[ l.transversalities ] suppose @xmath10 has smooth boundary , and @xmath96 is a regular parametrization of it .
if @xmath100 is transverse to both submanifolds @xmath91 and @xmath90 then @xmath64 .
let us also postpone the proof of this , and use it to obtain the following : [ p.transsl2r ] if @xmath10 has smooth boundary then there is an open dense subset @xmath101 of @xmath26 such that if @xmath102 then @xmath103 .
let @xmath104 be the set of pairs @xmath82)$ ] such that @xmath86 and @xmath84 are linearly independent .
note that @xmath91 , @xmath90 are subsets of @xmath105 .
the group @xmath26 acts on @xmath105 in the obvious way : @xmath106 ) = ( lu , [ lv])$ ] . this action is smooth , transitive , and faithful ; in particular @xmath26 and @xmath105 are diffeomorphic .
let @xmath96 be a regular parametrization of @xmath56 .
then @xmath100 takes values in @xmath105 .
the map @xmath107 defined by @xmath108 is a submersion .
therefore , by the transversality theorem @xcite ( or see ( * ? ? ?
* theorem 2.7 ) for a more precise version ) , the set @xmath109 formed by those @xmath33 such that @xmath110 is transverse to @xmath91 and to @xmath90 is open and dense in @xmath26 .
take @xmath102 . noting that @xmath111 , it follows from [ l.transversalities ] that @xmath103 .
this yields the denseness of @xmath55 in @xmath4 : given @xmath10 , we initially perturb it w.r.t .
the symmetric difference metric so the boundary becomes smooth : for example , we can take an inscribed polygon , smoothen the corners , and inflate it to normalize the area . then by [ p.transsl2r ] we can apply a element of @xmath26 close to the identity and so obtain the desired approximation in @xmath55 .
we are left to prove the lemmas .
the _ qr decomposition _ comes in handy : there is a diffeomorphism @xmath112 given by : @xmath113 changing coordinates under @xmath114 , the left action of the diagonal subgroup corresponds to translation of the first coordinate .
let @xmath105 be as in the proof of [ p.transsl2r ] .
define a diffeomorphism @xmath115 by @xmath116)$ ] , where @xmath117 is the canonical basis of @xmath27 .
so the map @xmath118 allows us to put global coordinates @xmath119 on @xmath105 .
note that , in the coordinates just described , @xmath120 is given by equations @xmath121 , @xmath122 .
so , applying the diagonal subgroup , we conclude that @xmath90 is the surface @xmath123 .
analogously , @xmath91 corresponds to @xmath122 and @xmath124 .
this proves [ l.submanifolds ] .
let @xmath10 and suppose that @xmath56 has a regular parametrization @xmath125 such that @xmath100 is transverse to both @xmath91 and @xmath90 .
it is clear that each parameter @xmath126 for which @xmath127 corresponds to a point where @xmath56 is tangent to one of the ellipses @xmath60 , say @xmath128 . by transversality
, the number of such parameters is finite , so we obtain condition ( [ i.finite ] ) in the definition of @xmath55 .
choose a regular parametrization @xmath129 of @xmath128 such that @xmath130 .
now suppose that @xmath56 and @xmath128 have a _
second - order _ contact at the point @xmath131 .
then two curves @xmath100 and @xmath132 have a first - order contact ( i.e. are tangent ) at @xmath126 . since @xmath132 is an immersion whose image is contained in @xmath90 , we conclude that @xmath100 is _ not _ transverse to @xmath90 , which is a contradiction .
this proves condition ( [ i.quadratic ] ) in the definition of @xmath55 .
since @xmath91 is @xmath3-dimensional , @xmath100 is transverse to @xmath91 iff its image does not intersect @xmath91 : this is equivalent to condition ( [ i.envelope ] ) , concluding the proof that @xmath64 .
in this , we will prove [ p.main ] up to part ( [ i.tang ] ) , and also deduce some formulas for the derivatives that will be the basis for the proof of part ( [ i.trans ] ) .
let @xmath64 . by applying a suitable element of @xmath26 that preserves the @xmath16 and @xmath17 axes , it is sufficient to consider @xmath121 .
first , assume that @xmath133 , so the curves @xmath56 and @xmath134 are transverse .
then they intersect at @xmath135 points , which we list in counterclockwise order as @xmath136 , , @xmath137 .
extend this to a periodic two - sided sequence @xmath138 and lift it to an increasing two - sided sequence @xmath139 in @xmath140 such that @xmath141 and @xmath142 for each @xmath143 . actually , since @xmath0 is centrally symmetric , @xmath144 . shifting indices by @xmath3 if necessary
, we assume that the following condition holds : if the curve @xmath56 is traversed counterclockwise , then it exits the unit disk @xmath145 at the points @xmath146 with @xmath143 even , and enters it at the points @xmath146 with @xmath143 odd : see [ f.notation ] .
let @xmath147 denote the non - oriented angles of intersection ; note that @xmath148 since @xmath0 is centrally symmetric .
so @xmath149 is a two - sided sequence with period @xmath150 . and the circle @xmath134 . ] [ l.formulas ] the function @xmath68 is @xmath57 at a neighborhood of @xmath151 and @xmath152 \ , .
\label{e.2nd}\end{aligned}\ ] ] in polar coordinates @xmath153 , the ellipse @xmath60 has equation : @xmath154 similarly , the curve @xmath56 is represented by some equation @xmath155 , where @xmath156 is a positive @xmath57 function on the circle @xmath157 which satisfies @xmath158 . since @xmath56 and @xmath134 are transverse , @xmath159 vanishes exactly on the points @xmath160 , , @xmath161 , and @xmath162 for each @xmath143 .
let @xmath163 . by the implicit function theorem , for @xmath22 sufficiently close to @xmath151 , the function
@xmath164 vanishes on @xmath135 points @xmath165 , , @xmath166 ; moreover each @xmath167 is @xmath57 and satisfies @xmath168 and @xmath169 then : @xmath170 by leibniz integral rule , @xmath171 \ , , \label{e.leibniz1}\\ a''(t ) & = \sum_{j=1}^{2n } ( -1)^j \left [ \int_{\xi_j(t)}^{\xi_{j+1}(t ) } f_{tt}(\theta , t ) { \,\mathrm{d}}\theta + f_t(\xi_{j+1}(t),t ) \xi_{j+1}'(t ) - f_t(\xi_j(t),t ) \xi_j'(t ) \right ] \notag \\ & = \sum_{j=1}^{2n } ( -1)^j \left [ \int_{\xi_j(t)}^{\xi_{j+1}(t ) } f_{tt}(\theta , t ) { \,\mathrm{d}}\theta- 2 f_t(\xi_j(t),t ) \xi_j'(t ) \right ] \notag \\ & = \sum_{j=1}^{2n } ( -1)^j \left [ \int_{\xi_j(t)}^{\xi_{j+1}(t ) } f_{tt}(\theta , t ) { \,\mathrm{d}}\theta + \frac{2 [ f_t(\xi_j(t),t)]^2}{f_\theta(\xi_j(t),t ) } \right ] \ , .
\label{e.leibniz2}\end{aligned}\ ] ] in particular , @xmath68 is a @xmath57 function in a neighborhood of @xmath151 , as claimed .
now we consider @xmath172 .
a computation gives @xmath173 plugging into , @xmath174 proving .
analogously , from we obtain : @xmath175 \\ & = \sum_{j=1}^{2n } \left [ \frac{(-1)^j \sin 4 \xi_j}{2 } + \frac{1 + \cos 4\xi_j}{|g'(\xi_j)| } \right ] \ , , \end{aligned}\ ] ] where in the last step we have used . since @xmath176 , we obtain .
it follows from [ l.formulas ] that the function @xmath68 is @xmath57 except at the points @xmath22 in @xmath67 ( the set of tangency parameters ) .
consider such a point .
as before , we can assume @xmath172 .
define functions @xmath177 and @xmath178 as in the proof of the above .
fix numbers @xmath179 with @xmath180 such that the tangencies between @xmath56 and the unit circle @xmath134 occur at the points @xmath181 . then
for each @xmath182 we have then @xmath183 , @xmath184 , @xmath185 , and @xmath186 ; these last two properties follow respectively from conditions ( [ i.quadratic ] ) and ( [ i.envelope ] ) of the definition of @xmath55 . in order to correct the formula to take the tangencies into account , for each @xmath182
we need to add a certain correction term @xmath187 , which we describe next .
consider for instance the case where @xmath188 and @xmath189 , which we call type @xmath190 .
for each @xmath22 sufficiently close to zero , consider the equation @xmath191 for @xmath192 in a neighborhood of @xmath193 : it has no solution @xmath192 for @xmath194 , exactly one solution @xmath193 for @xmath172 , and exactly two solutions @xmath195 for @xmath196 .
then the correction term is : @xmath197 and @xmath198 if @xmath199 . for @xmath196 close to @xmath151 , the width @xmath200 is @xmath201 , and the correction term @xmath187 is @xmath202 .
so this tangency forces the left derivative @xmath203 to be @xmath204 .
the other cases are similar .
the upshot is : @xmath205 therefore the statement ( [ i.tang ] ) of [ p.main ] holds .
we assume @xmath133 and keep the notation from [ ss.derivatives ] . before going into the the proof of part ( [ i.trans ] ) of [ p.main ] itself
, we will establish some preliminary inequalities .
it is convenient to reparameterize the sequence @xmath206 differently .
for each @xmath207 , let : @xmath208 so @xmath209 , @xmath210 , and @xmath211 .
[ l.angle_bound ] for each @xmath182 we have @xmath212 .
[ f.angle_bound ] shows how to bound @xmath213 .
the bound for @xmath214 is analogous . .
] next , we want some bounds on the parameters @xmath215 . shifting indices if necessary , we assume that @xmath216 if @xmath217 , we fix @xmath218 such that @xmath219 .
note that @xmath220 and in particular @xmath221 .
the following uses the other hypothesis from part ( [ i.trans ] ) of [ p.main ] , namely that @xmath222 is not too close to @xmath151 nor to @xmath95 .
[ l.restr ] for every @xmath73 there exists @xmath223 , not depending on @xmath0 , such that if @xmath224 then : @xmath225 and , if @xmath217 , @xmath226 note that @xmath0 contains the disk of radius @xmath227 centered at the origin ( see [ f.inner_disk ] ) , and in particular : @xmath228 so @xmath229 implies @xmath230 , proving the first inequality in .
now consider : if this inequality does not hold then @xmath231 and @xmath232 are both approximately @xmath233 .
then we can show that @xmath234 is contained in the union of four small regions ( see [ f.4_corners ] ) , and so @xmath222 is again close to @xmath151 .
the second inequality in is trickier . consider regions @xmath235 and @xmath236 as in [ f.pogoball ] , such that : @xmath237 in particular , @xmath238 the areas of the regions @xmath239 and @xmath240 are functions of the corresponding intersection angles : @xmath241 so @xmath242 where @xmath243 .
on the other hand , the triangles @xmath244 have an angle @xmath245 with opposite side @xmath246 .
so the area of @xmath244 can not exceed the area of an isosceles triangle with angle @xmath247 and opposite side @xmath248 , that is , @xmath249 so @xmath250 yields : @xmath251 in particular : @xmath252 and therefore : @xmath253 now suppose @xmath231 is close to @xmath7 ( otherwise there is nothing to show ) . there are two cases : * if holds then @xmath254 is small , which by implies that @xmath255 is small , and so yields that @xmath222 is small . *
if holds then @xmath256 is small , that is , @xmath254 is close to @xmath257 ; then implies that @xmath258 is close to @xmath7 , and yields that @xmath222 is close to @xmath95 . as a result
, the assumption @xmath259 forbids @xmath231 to approach @xmath7 .
this completes the proof of the second inequality in and of the .
we will now come back to the formulas obtained in [ l.formulas ] and rewrite them in terms of the new parameters ; we will also use [ l.angle_bound ] to obtain a convenient lower bound for the second derivative .
[ l.new ] @xmath260 \ , .
\label{e.new_2nd}\end{aligned}\ ] ] we can rewrite as @xmath261 so follows from the identity @xmath262 analogously , rewriting as @xmath263 \ , .\ ] ] by [ l.angle_bound ] , @xmath264 .
so , using the identity @xmath265 we obtain : @xmath266 \ , .\ ] ] let us manipulate the quantity between square brackets . for simplicity of writing ,
we omit the @xmath182 indices : @xmath267 & = \underbrace{\sin 2\omega}_{2 \sin \omega \cos \omega } \cos 2\sigma + \cot \tfrac{1}{2}\omega \ , \bigl [ \underbrace{(-1 + \cos 2\omega)}_{-2\sin^2 \omega}\cos 2\sigma + \underbrace{(1 + \cos 2\sigma)}_{2 \cos^2 \sigma } \bigr ] \\ & = 2 \sin \omega \underbrace{\bigl ( \cos \omega - \cot \tfrac{1}{2}\omega \ , \sin \omega\bigr)}_{1 } \underbrace{\vphantom{\bigl(\bigr ) } \cos 2\sigma}_{\cos^2 \sigma - \sin^2 \sigma } + \ 2 \cot \tfrac{1}{2}\omega \cos^2 \sigma \\ & = 2\sin\omega \sin^2 \sigma + 2 \left ( \cot \tfrac{1}{2}\omega - \sin\omega \right ) \cos^2 \sigma \ , , \end{aligned}\ ] ] yielding .
the proof is a case - by - case analysis ; in most of the cases we will show that @xmath268 is positive and away from zero , but in a few cases the conclusion is that @xmath269 is away from zero .
all the estimates on those derivatives will be obtained from [ l.new ] , which will not be explicitly mentioned .
let us introduce some notation : @xmath270 so the fundamental inequality can be rewritten as : @xmath271 note that @xmath272 is a convex combination of the two functions @xmath273 and @xmath274 , which are plotted in [ f.2_functions ] .
we will use this fact repeatedly to obtain bounds . note that the abscissa of the crossing between the two graphs is @xmath275 , that @xmath276 , and that @xmath277 recall that @xmath231 is the biggest of all angles @xmath215 s and so it is the only angle that can be bigger than @xmath257 ; therefore the sum in contains at most one negative term . and @xmath52 .
] * case 1 .
* @xmath278 . then @xmath279 so we obtain in the remaining cases , we assume @xmath280 . * case 2 . * @xmath281 .
then , by the first inequality in , @xmath282 in the remaining cases , we assume @xmath283 . * case 3 . * @xmath284 . then , by the first inequality in , @xmath285 in the remaining cases , we assume @xmath286 . recall from [ ss.geometric ] that @xmath232 is the second biggest of the @xmath215 s .
* case 4 . *
so : @xmath288 now , @xmath289 and the function @xmath52 is decreasing on the interval @xmath290 $ ] ( actually , it is decreasing on a slightly bigger interval ) ,
so : @xmath291 now , using we obtain : @xmath292 and we are done in this case . in the remaining cases , we assume @xmath293 .
define the following numbers : @xmath294 * case 5 .
* @xmath295 .
then @xmath296 and we are done .
in the final and most interesting case , we assume @xmath297 .
* let us establish two upper estimates for @xmath298 ; the first is : @xmath299 and the second is : @xmath300 define : @xmath301 which by , satisfies @xmath302 .
note that @xmath303 that is , @xmath304 .
let @xmath305 .
we can rewrite @xmath306 as : @xmath307 \ , .\ ] ] since @xmath308 and the function @xmath309 is positive and decreasing on the interval @xmath310 , the assumption @xmath311 yields two other inequalities : @xmath312
so , on one hand , @xmath313 on the other hand , recalling that @xmath314 and @xmath315 , @xmath316 by cauchy
schwarz inequality , @xmath317 this inequality together with allows us to show that @xmath269 is not too close to zero : @xmath318 this concludes the proof of part ( [ i.trans ] ) of [ p.main ] .
to begin , let us refine the consequences of [ p.main ] .
consider @xmath64 and let @xmath319 .
trivially , @xmath320 as @xmath321 .
let us say that @xmath18 is a _ critical point _
if @xmath322 either does not exist ( i.e. , @xmath323 ) or is zero .
it follows from [ p.main ] that there are only finitely many critical points , and that there exists exactly one of them , say @xmath324 , such that @xmath68 is decreasing on @xmath325 $ ] and increasing on @xmath326 .
given @xmath73 , let @xmath327 this is a ( possibly empty ) compact interval .
let @xmath328 be given by part ( [ i.trans ] ) of [ p.main ] , and consider : @xmath329 it follows from [ p.main ] that @xmath330 is either empty or a compact interval with the same left endpoint as @xmath331 ; in the latter case , the restriction of @xmath68 to @xmath330 is @xmath57 .
analogous facts hold for the sets : @xmath332 we use this information to obtain the following : [ l.no_plateau ] given @xmath333 and @xmath334 , there exists @xmath335 such that for every @xmath64 , the function @xmath319 has the following property : for every @xmath336 , if @xmath337 then @xmath338 .
let @xmath73 and @xmath339 be given .
let @xmath72 be given by [ p.main ] and let @xmath340
. now fix @xmath64 and consider the intervals @xmath341 and @xmath342 as in the preceding discussion .
fix @xmath336 such that @xmath343 $ ] .
then @xmath344 .
suppose @xmath345 , since the other case is analogous .
assume that @xmath346 , otherwise there is nothing to prove .
then the interval @xmath347 $ ] is contained in @xmath331 .
now consider two cases : the first case is @xmath348 .
then @xmath349 almost everywhere in @xmath350 .
note that the function @xmath68 , which may have points of non - differentiability , still obeys the fundamental theorem of calculus .
so we have @xmath351 , completing the proof in this case . in the remaining case ,
@xmath352 $ ] for some @xmath21 . then , recalling that the restriction of @xmath68 to @xmath330 is @xmath57 , we have : @xmath353 \ \rightarrow \
a'(t ) = \underbrace{a'_+ ( t_0)}_{\ge 0 } + \int_{t_0}^{t } \underbrace{\vphantom{a'_+ ( t_0 ) } a''(s)}_{>\delta } { \,\mathrm{d}}s \ge \delta ( t - t_0 ) \ , , \ ] ] while @xmath354 \
\rightarrow \
a'(t ) \ge \delta \ge \delta ( t - t_0 ) \text { as well}\ ] ] ( using that @xmath355 ) .
so : @xmath356 fix @xmath10 and let @xmath319 . if @xmath357 is another element of @xmath4 , then by the triangle inequality the corresponding function @xmath358 satisfies : @xmath359 by [ p.trans ] , we can approximate @xmath0 by @xmath357 in @xmath55 , and by [ p.main ] the corresponding function @xmath360 is ( strictly ) quasiconvex .
therefore @xmath68 is the uniform limit of quasiconvex functions and so it is quasiconvex .
assume for a contradiction that @xmath68 is not stricly quasiconvex .
then @xmath68 has a plateau , that is , a non - trivial interval @xmath361 $ ] , say with @xmath355 , where @xmath68 is constant , say equal to @xmath362 . note that @xmath363 .
fix a positive @xmath364 , and let @xmath365 be given by [ l.no_plateau ] . by [ p.trans ] ,
there exists @xmath366 with @xmath367 . letting @xmath368 , we have
@xmath369 this contradicts the conclusion of [ l.no_plateau ] .
the is proved .
coming back to the uniqueness [ t.unique ] , let us briefly comment on the problem of now to actually determine the ellipse @xmath370 that best approximates a given @xmath10 . if @xmath0 has any non - trivial symmetry ( more precisely , there is some @xmath371 such that @xmath372 ) then @xmath370 must have the same symmetry , and therefore becomes uniquely determined .
for example , the best approximation to a regular @xmath150-gon is the unit disk . in the absence of symmetries ,
the information provided by [ t.main2 ] should be helpful .
for example , if one manages to find a local minimum of the area function then this is automatically the solution .
actually , in most cases it should be sufficient to find a critical point , due to the following result ( which follows from the proof of [ p.main ] ) : [ c.scholium ] let @xmath10 and @xmath373 .
suppose that the boundaries of @xmath0 and @xmath374 intersect at finitely many points , and that @xmath56 is @xmath57 and transverse to @xmath375 on the neighborhood of each of these points .
then the function @xmath53 is @xmath57 at a neighborhood of @xmath374 .
if moreover its gradient vanishes at @xmath374 then @xmath374 is the global minimum .
note that the gradient is easy to compute under the assumptions of the [ c.scholium ] , as we have seen in [ ss.derivatives ] .
let us remark that quasiconvexity is a very useful property for the design of optimization algorithms : see for instance @xcite .
there are several other natural questions one can ask concerning the metric space @xmath376 : what is its diameter , what elements of @xmath4 are most distant to @xmath9 , what is that maximum , etc . as in @xcite , we ask for the validity of the uniqueness result in higher dimensions .
the space of ellipsoids also has a natural affine structure ( see [ ss.ellipsoids ] ) , so we also ask whether strict quasiconvexity holds .
this appendix is not needed for the rest of the paper , but it justifies our definition of segments in the space @xmath9 . it also has interest on its own . given any metric space @xmath377 , we can define the ( possibly infinite ) _ length _ of a continuous curve @xmath378 \to m$ ] as @xmath379 where the supremum is taken over all choices of points @xmath380 .
if @xmath381 is path - connected then we define , for every @xmath16 , @xmath382 @xmath383 where @xmath52 runs over all the continuous curves from @xmath16 to @xmath17 .
note that @xmath384 and @xmath40 satisfies the usual axioms of a metric , except that it may be infinite .
( actually , it may be identically @xmath385 : consider for example the koch curve . ) if @xmath40 is everywhere finite and it induces the same topology as @xmath41 , then we call it the _ intrinsic metric _ induced by @xmath41 .
a curve @xmath52 that attains the infimum in ( if it exists ) is a called a ( parameterized ) _ geodesic segment _ joining @xmath16 to @xmath17 .
see ( * ? ? ?
* chapter 2 ) for much more information on this subject .
now consider the metric space @xmath386 where , as before , @xmath9 is the set of ellipses with center at the origin and area @xmath7 , and @xmath41 is the symmetric difference metric . a geometric description of the associated length of curves will be given later ( see [ p.footprint ] ) .
let @xmath389 be the upper - half plane , endowed with the riemannian metric obtained by multiplying the euclidian inner product by @xmath390 .
let @xmath391 be the corresponding distance function .
this space has constant curvature @xmath392 , and its geodesics are either euclidian half - lines or euclidian half - circles orthogonal to the real axis .
moreover , the group @xmath26 acts on @xmath43 by isometries via mbius transformations : @xmath393 the action is transitive , and the stabilizer of @xmath394 is @xmath395 , the group of rotations . since the action of @xmath26 on @xmath9 is also transitive , and the stabilizer of the unit disk @xmath145 is
also @xmath395 , there exists an unique bijection @xmath396 that sends @xmath182 to @xmath145 and that commutes with the group action ( i.e. , @xmath397 ) .
initially , consider an arbitrary pair of distinct numbers @xmath22 , @xmath401 . then
@xmath402 consists on four points @xmath403 , @xmath404 .
consider the following four lines : the two coordinate axes , the line containing @xmath403 , and the line containing @xmath404 .
these four lines split the region @xmath405 into eight `` slices '' .
we claim that all slices have the same area . in the case that @xmath406 , this claim is clearly true by symmetry . applying the diagonal subgroup of @xmath26
, we conclude that the claim always holds .
now consider @xmath407 .
then four of the slices are sectors of the unit disk @xmath145 with central angles @xmath408 , and therefore all eight slices have area @xmath409 .
so : @xmath410 yielding the announced formula . pulling back the symmetric difference metric @xmath41 by the map @xmath398 we obtain a new metric on @xmath43 : @xmath411 for any pair of points @xmath412 , @xmath413 there exists @xmath414 such that @xmath415 and @xmath416 for some @xmath18 .
note that @xmath417 and that @xmath418 .
so it follows from [ l.dist_formula ] that @xmath419 in particular , @xmath420 this property clearly implies that @xmath421 induces the intrinsic metric @xmath422 .
follow .
[ [ geometric - interpretation - of - the - length - with - respect - to - the - symmetric - difference - metric ] ] geometric interpretation of the length with respect to the symmetric difference metric ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ [ p.footprint ] suppose that @xmath423 \to { \mathcal{e}}$ ] is a continuous curve .
then the length of @xmath34 is given by the area integral @xmath424 where @xmath425 { \;\mathord{;}\;}(x , y ) \in
\partial f(t ) \}$ ] . in informal terms
, @xmath426 measures the area of @xmath427 } \partial f(t)$ ] with multiplicities ; thus we call it the _ footprint length _ of the path @xmath34 .
to illustrate , [ f.footprint ] shows the case where @xmath34 is a parametrized segment @xmath428 $ ] .
[ r.locally_flat ] unlike @xmath430 , geodesic segments are not unique in the space @xmath376 : for example if @xmath22 is sufficiently small then there exist many curves in @xmath4 from @xmath145 to @xmath431 with footprint length equal to @xmath432 .
let @xmath433 be space of ellipsoids in @xmath434 with center at the origin and the same volume as the unit ball , endowed with the symmetric difference metric @xmath435 .
there is an associated intrinsic metric @xmath436 ; moreover geodesic segments always exist and are unique modulo reparametrization .
there is a bijection @xmath437 between the symmetric space @xmath438 and the space @xmath433 that sends the basepoint to the unit ball and that commutes with the action of @xmath439 .
consider @xmath440 endowed with its unique @xmath439-invariant riemannian metric ( which has nonpositive curvature ) .
then @xmath437 is _ not _ an isometry , but it is affine : it preserves geodesic segments .
the pull - back of @xmath436 to @xmath440 is the distance function induced by a @xmath439-invariant finsler non - riemannian structure .
the lengths of curves have a `` footprint '' characterization similar to [ p.footprint ] . | we prove that for every planar centrally symmetric convex body @xmath0 , among all ellipses @xmath1 with the same center and the same area as @xmath0 , there exists a unique one for which the area of the symmetric difference @xmath2 is minimal .
we obtain this result by showing that this area is a strictly quasiconvex function of the ellipse @xmath1 , with respect to a natural affine structure on the set of ellipses . |
early - type galaxies ( ellipticals and lenticulars ) are typically dominated by an old stellar population and contain at most a small fraction of their total mass in cold gas and dust .
recent surveys of nearby galaxies have detected co emission in 5% of the elliptical galaxies in the atlas@xmath6 sample @xcite and submillimeter emission from cold dust in 24% of the elliptical galaxies in the herschel reference survey @xcite .
unfortunately there are relatively few early - type galaxies that have been observed and detected in both dust and gas tracers .
@xcite measured gas - to - dust mass ratios of @xmath7 in a sample of 7 elliptical galaxies .
@xcite measure gas - to - dust mass ratios for 8 s0 galaxies that are consistent with or higher than the value of 100 - 150 typically found in spiral galaxies .
is a luminous e6(pec ) galaxy at a distance of 23.9 mpc @xcite , which underwent a merger and a burst of star formation 6 - 8 gyr ago @xcite .
its 60 @xmath8 m iras luminosity of @xmath9 l@xmath1 places it at the high luminosity end of the sauron sample @xcite .
however , ngc 4125 is undetected in either hi or co @xcite ; tight upper limits correspond to @xmath10 m@xmath1 ( 2@xmath11 , 8.7@xmath12 beam ) and @xmath13 m@xmath1 ( 2@xmath11 , 21@xmath14 beam ) when using a co - to - h@xmath15 conversion factor of @xmath16 h@xmath15 @xmath17 ( k km s@xmath18)@xmath18 @xcite .
it has significant x - ray emission from hot gas , with at most a few percent contribution by an agn @xcite .
it was selected as the `` normal '' elliptical galaxy for the very nearby galaxies survey with the _ herschel space observatory _
( pi : c. d. wilson ) . however , as we describe in this letter , the interstellar medium ( ism ) of ngc 4125 appears to be anything but normal .
ngc 4125 was observed in scan map mode over a @xmath19 area with the _ herschel space observatory _ with the photodetector array camera and spectrometer ( pacs , * ? ? ?
* obsids 1242188202 , 1242188203 ) and the spectral and photometric imaging receiver ( spire , * ? ? ?
* obsid 1242199156 ) . the data processing and calibration for the spire data
are described in @xcite . for the pacs photometry
, the level-1 data were downloaded from the herschel science archive using hipe 9.0 and the latest calibration ( pacs_cal_41_0 ) and mapped using scanamorphos v21 @xcite .
ngc 4125 was observed in the [ cii ] 158 @xmath8 m , [ nii ] 122 @xmath8 m , and [ oi ] 63 @xmath8 m emission lines ( obsids 1342222063 , 1342222064 , 1342222065 ) using pacs in chop / nod mode with a single pointing covering @xmath20 .
the maps were processed as described in @xcite using hipe and the pacsman package @xcite .
the 250 @xmath8 m image of ngc 4125 is shown overlaid on the k - band image from 2mass @xcite in figure [ fig - kcomp ] .
the total continuum flux of ngc 4125 was measured using an elliptical aperture @xmath21 in diameter ( 62% of the @xmath22 optical size ) with position angle 82@xmath23 ; the sky background was measured in an elliptical annulus with inner and outer diameters of 216@xmath14 and 360@xmath14 . for the spire data , in calculating the total flux , we adopt beam areas appropriate for emission going as @xmath24 of 432 , 766 and 1573 square arcseconds at 250 , 350 , and 500 @xmath8 m , respectively , and apply the multiplicative color corrections appropriate for extended sources @xcite . for the pacs data ,
we adopt the color corrections appropriate for a 20 k blackbody @xcite . for both sets of maps ,
the uncertainties are calculated as the quadrature sum of the instrumental uncertainty in each pixel and the uncertainty due to the sky confusion .
calibration uncertainties of 7% for spire @xcite and 5% for pacs
@xcite are added in quadrature to these measurement uncertainties in table [ tbl - phot ] .
we have detected weak [ cii ] and [ nii ] emission ( figure [ fig - pacs ] ) . the peak [ cii ] flux in an 18@xmath14 diameter aperture is @xmath25 w m@xmath26 and the total flux summed in a 36@xmath14 diameter aperture is @xmath27 w m@xmath26 . for [ nii ] , the peak flux is @xmath28 w m@xmath26 and the total flux is @xmath29 w m@xmath26 .
the [ oi ] 63 @xmath8 m line is not detected , with a 2@xmath11 upper limit to the total flux of @xmath30 w m@xmath26 .
calibration uncertainties are 30% for these spectroscopic data .
one unusual aspect of the dust emission in ngc 4125 is that the emission is spatially extended and follows the general shape of the stellar emission ( figure [ fig - kcomp ] ) .
while previous optical extinction maps have shown a central dust lane aligned with the major axis @xcite , the _ herschel _ images trace dust over a significantly larger area and the dust emission even follows the @xmath4 profile that is a common feature of ellipticals .
fitting the global dust emission from 160 to 500 @xmath8 m with a modified blackbody with dust opacity going as @xmath31 gives an average dust temperature of @xmath32 k ( figure [ figbb ] ) .
the dust mass can be calculated as @xmath33 m@xmath1 , where @xmath34 is the distance to the galaxy in mpc , @xmath35 is the dust temperature in k , @xmath36 is the 250 @xmath8 m flux in jy , and @xmath37 is the dust emissivity at 250 @xmath8 m in @xmath38 g@xmath18 . as our default , we assume a standard graphite plus silicate dust emissivity @xmath39 @xmath40 g@xmath18 @xcite , which corresponds to @xmath41 and has also been used for previous analyses of elliptical galaxies @xcite .
we obtain a global dust mass for ngc 4125 of @xmath42 m@xmath1 ( table [ tbl - prop ] ) and a mass in the peak resolution element of @xmath43 m@xmath1 . using the same distance and emissivity ,
our dust mass is a factor of 4 larger than that of @xcite , who obtained a warmer dust temperature due to a lack of fluxes longward of 160 @xmath8 m .
the dust mass is rather sensitive to our assumptions about the grain properties .
if we instead adopt the mixture of amorphous carbon and silicates explored for the large magellanic cloud by @xcite , for which @xmath44 and @xmath45 @xmath40 g@xmath18 , we obtain a dust temperature of @xmath46 k and a total dust mass of @xmath47 m@xmath1 . performing the fit with @xmath48 allowed to vary between 1 and 2.5 returns
a fit with @xmath49 and a temperature of @xmath50 k. combining this temperature with @xmath51 @xmath40 g@xmath18 gives similar dust masses to the amorphous carbon model of @xcite .
we also note , however , that single temperature fits to spiral galaxies usually yield smaller dust masses than more realistic multiple temperature models @xcite . fitting the data with two components with @xmath52 ( figure [ figbb ] )
gives temperatures of 17 and 51 k and a cold dust mass of @xmath53 m@xmath1 .
although the 500 @xmath8 m flux lies somewhat above the fits , the excess emission is likely due to two faint ( @xmath54 mjy ) point - like sources to the west of ngc 4125 ( figure [ fig - kcomp ] ) .
these sources contribute 20% of the total flux at 500 @xmath8 m but only 10% at 250 @xmath8 m and may be background galaxies .
we estimate a stellar mass of @xmath55 m@xmath1 from the 3.6 @xmath8 m flux in @xcite and the median mass - to - light ratio of 1.00 @xmath56 @xmath57 from @xcite .
the dust to stellar mass ratio of @xmath58 is towards the high end of the values seen for nearby elliptical galaxies @xcite .
comparing the global dust mass with the sum of the hi and h@xmath15 upper limits @xcite and including a factor of 1.36 to account for helium @xcite gives a global gas - to - dust ratio of @xmath59 . because the areas covered by the three different measurements ( hi , co , dust ) are very different , another relevant comparison is between the peak dust mass ( 18@xmath14 beam ) and the co upper limit ( 21@xmath14 beam ) . comparing these two values gives a limit on the gas - to - dust mass ratio of @xmath60 .
these @xmath61 upper limits are factors of 7 - 12 smaller than the typical gas - to - dust mass ratio of 150 seen in our own and other galaxies .
thus , there appears to be a deficit of neutral gas to go along with the cold dust in ngc 4125 .
we can also estimate the gas mass associated with the observed [ cii ] emission .
we assume that the [ cii ] emission comes from the ionized gas phase ( @xmath62 k ) via collisional excitation with electrons .
we note that the [ nii]/[cii ] ratio is consistent with a low - density ionized medium @xcite . assuming the electron density , @xmath63 , is much less than the critical density @xmath64 @xmath65 @xcite , the column density is given by @xmath66 with @xmath67 in units of erg s@xmath18 @xmath17 sr@xmath18 .
assuming a c@xmath68 abundance relative to hydrogen of @xmath69 , the mass corresponding to the total flux in the map is @xmath70 m@xmath1 .
solving for the case where the ionized medium uniformly fills the 18@xmath14 ( 2.1 kpc ) radius volume gives a lower limit to @xmath63 of 0.11 @xmath65 and an ionized hydrogen mass of @xmath71 m@xmath1 .
combining this mass estimate with our dust mass and again accounting for helium with a factor of 1.36 gives a total ( ionized plus neutral ) gas - to - dust mass ratio of @xmath72 for @xmath52 or @xmath73 for @xmath44 .
this analysis suggests that a significant fraction of the gas in ngc 4125 is warm ionized gas that produces [ cii ] emission rather than cold neutral gas .
the dusty ism of ngc 4125 has some unusual properties .
the dust has a similar radial profile to the stellar light out to at least 7 kpc ( 1@xmath12 ) radius ( figure [ fig - kcomp ] ) .
its total dust mass is among the highest seen in elliptical galaxies @xcite and is only a factor of 3 times smaller than the dust mass in the elliptical galaxy centaurus a @xcite .
however , unlike cen a , where the neutral gas - to - dust mass ratio is 100 and the gas traces a clearly rotating disk @xcite , in ngc 4125 the corresponding gas is not neutral but in a warmer ionized or even x - ray emitting phase .
these unusual properties suggest an internal origin for the ism of ngc 4125 .
the ratio @xmath74 is towards the high end of the range seen in ellipticals , as is the total stellar mass @xmath75 @xcite .
the optical colors of ngc 4125 suggest that it underwent a burst of star formation 6 - 8 gyr ago @xcite .
long - slit spectroscopy combined with simple stellar population models indicate super - solar metallicity and also point to a recent dissipational merger event and stellar ages as young as 8 gyr @xcite .
many of these younger stars would now be in the red giant , carbon star , or asymptotic giant branch ( agb ) phase when stars shed large quantities of gas and dust .
the material that is lost from these stars has a normal gas - to - dust mass ratio .
however , if the dust from the star survives for a longer time ( i.e. 50 myr , * ? ? ?
* ) than the gas survives in the neutral phase , the observed _ neutral _ gas to dust mass ratio could be reduced , because the majority of the gas would be in an ionized or x - ray emitting phase . assuming a dust lifetime of 50 myr , the stars in ngc 4125 would need to produce dust at a rate of @xmath76 m@xmath1 yr@xmath18 to explain the observed dust mass .
this rate is much higher than the typical rate seen in passive elliptical galaxies @xcite .
@xcite derived total ( gas plus dust ) mass loss rates from a few @xmath77 to a few @xmath78 m@xmath1 yr@xmath18 for a collection of agb stars and proto - planetary and planetary nebulae in our own galaxy , with a median mass loss rate of @xmath79 m@xmath1 yr@xmath18 .
@xcite find that extreme agb stars are responsible for @xmath80 of the dust produced by evolved stars in the smc . assuming a normal gas - to - dust mass ratio of 150 , ngc 4125 would need to contain roughly @xmath81 such stars ( 0.001% of its total stellar mass ) to produce the required dust .
given the total mass of ngc 4125 , such a population of stars seems quite reasonable .
in contrast , if the dust in ngc 4125 was acquired via an interaction with another galaxy , the lack of neutral gas suggests that the interaction must have been extremely recent .
we would need to be viewing ngc 4125 at the precise time when most of the gas has been ionized but significant cold dust remains .
it is hard to understand how such a large mass of dust could survive for much more than 50 myr without the high column densities which would shield both the dust and the neutral gas .
one question which arises is why we have not discovered more galaxies like ngc 4125 .
first , such galaxies may be intrinsically rare in the local universe .
the herschel reference survey @xcite contains all elliptical galaxies in a portion of the northern sky with @xmath82 mag and @xmath83 mpc ; ngc 4125 meets the distance and luminosity criteria but falls just outside the upper declination limit of the survey . adding ngc 4125 to the 18 elliptical galaxies from @xcite
, we can estimate that the rate for ngc 4125-like galaxies may be @xmath84% .
however , not all the galaxies in the herschel reference survey have atomic and molecular gas measurements and so some could be similar to ngc 4125 .
second , ngc 4125 is a luminous galaxy , which makes it easier to detect a given mass fraction of dust . for internal processes , a higher stellar mass would correlate with a higher dust mass .
the presence of a younger stellar population than in passively evolving ellipticals could also serve to increase the dust to stellar mass ratio .
given the sensitivities of typical _ herschel _ maps , a similar galaxy which was 1/3 as massive would appear as a point source at 250 @xmath8 m at the same distance as ngc 4125 , while a galaxy 1/10 as massive could not be distinguished easily from the _ herschel _ confusion limit .
finally , it is hard to know how critical the exact age and mass fraction of the younger stellar population might be in producing this particular configuration of dust without neutral gas . with stellar lifetimes of 1.4 gyr for a 2 m@xmath1 star and 5.5 gyr for a 1.25 m@xmath1 star @xcite , the exact age of the young population may not be a critical factor , as long as a young enough population forms a sufficient fraction of the stellar mass .
understanding the frequency and origin of galaxies like ngc 4125 may have important implications for understanding red galaxies at larger redshifts .
a galaxy like ngc 4125 would be rather difficult to interpret if it was observed at much larger distances .
its relatively strong submillimeter emission could be interpreted as indicating the presence of a significant cold ism @xcite , while its rest - frame 70 @xmath8 m emission could be taken as an indicator of on - going star formation .
for example , using the star formation rate calibration of @xcite , the 70 @xmath8 m flux in the central 18@xmath14 could be interpreted as a star formation rate of 0.15 m@xmath1 yr @xmath18 . for a typical gas depletion time of 2 gyr @xcite
, such a star formation rate would imply an h@xmath15 mass of @xmath85 m@xmath1 , a mass which is clearly ruled out by the existing co upper limits .
at some point in the past , the progenitors of elliptical galaxies must have contained a significant ism in order to form their current population of stars .
if the ism in ngc 4125 has an internal origin linked to the age of its most recent star formation episode , then we might expect galaxies like ngc 4125 to be more common at larger redshifts , when the overall stellar population of elliptical galaxies was , of necessity , younger .
recently , observations of moderate redshift galaxies have identified optically red galaxies which contain significant infrared emission . @xcite and
@xcite have studied large samples of early - type ( elliptical and lenticular ) galaxies from the h - atlas @xcite survey .
while these galaxies typically have larger dust masses and lower stellar masses than ngc 4125 , it would be interesting to follow up on the handful of massive galaxies with red @xmath86 colors . @xcite have identified a sample of dust - obscured galaxies with high spitzer mid - infrared to optical flux ratios .
the @xmath87 colors of some of these galaxies show they are clearly forming stars @xcite , but many are not detected in @xmath88 and so could instead be passive , post - star formation galaxies . _
herschel _ surveys have also identified a number of 250 @xmath8 m sources with red optical colors @xcite . while @xcite conclude that most of these are highly obscured galaxies , some of them have @xmath86 colors that are similar to ngc 4125 .
if galaxies like ngc 4125 , where the far - infrared emission is not related to star formation and neutral gas in the usual manner , become more common at higher redshift , this would have significant implications for our understanding of high redshift galaxies and galaxy evolution .
we thank the referee for comments that helped improve this paper .
this research of cdw is funded by the canadian space agency and the natural sciences and engineering research council of canada .
mps has been funded by the agenzia spaziale italiana ( asi ) under contract i/005/11/0 .
pacs has been developed by a consortium of institutes led by mpe ( germany ) and including uvie ( austria ) ; ku leuven , csl , imec ( belgium ) ; cea , lam ( france ) ; mpia ( germany ) ; inaf - ifsi / oaa / oap / oat , lens , sissa ( italy ) ; iac ( spain ) . this development has been supported by the funding agencies bmvit ( austria ) ,
esa - prodex ( belgium ) , cea / cnes ( france ) , dlr ( germany ) , asi / inaf ( italy ) and cicyt / mcyt ( spain ) .
spire has been developed by a consortium of institutes led by cardiff university ( uk ) and including univ .
lethbridge ( canada ) ; naoc ( china ) ; cea , lam ( france ) ; ifsi , univ .
padua ( italy ) ; iac ( spain ) ; stockholm observatory ( sweden ) ; imperial college london , ral , ucl - mssl , ukatc , univ .
sussex ( uk ) ; and caltech , jpl , nhsc , univ .
colorado ( usa ) .
this development has been supported by national funding agencies : csa ( canada ) ; naoc ( china ) ; cea , cnes , cnrs ( france ) ; asi ( italy ) ; mcinn ( spain ) ; snsb ( sweden ) ; stfc ( uk ) ; and nasa ( usa ) .
this research has made use of the nasa / ipac extragalactic database ( ned ) which is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
m contours overlaid on the @xmath89-band image from 2mass .
contour levels are 10 , 20 , 30 , 40 , 60 , 80 , 100 , 150 , and 200 mjy beam@xmath18 .
the two point - like sources to the west could be background sources and may be responsible for the slight excess in the 500 @xmath8 m emission ( figure [ figbb ] ) .
( bottom ) radial profiles of the 250 @xmath8 m ( solid ) , 160 @xmath8 m ( dashed ) and k - band brightness ( dot - dashed ) from 2mass .
the dotted line shows the 250 @xmath8 m beam profiled measured in the same annuli as the radial profile .
[ fig - kcomp ] ] 158 @xmath8 m spectrum of ngc 4125 .
the yellow solid line shows the fit to the data ; the dashed yellow line indicates the intrinsic velocity resolution of the instrument .
( bottom ) central [ nii ] 122 @xmath8 m spectrum of ngc 4125 .
[ fig - pacs],title="fig : " ] 158 @xmath8 m spectrum of ngc 4125 .
the yellow solid line shows the fit to the data ; the dashed yellow line indicates the intrinsic velocity resolution of the instrument .
( bottom ) central [ nii ] 122 @xmath8 m spectrum of ngc 4125 .
[ fig - pacs],title="fig : " ] m with @xmath52 ( solid ) , @xmath44 ( dotted ) and @xmath49 ( dashed ) overlaid .
( bottom ) a two temperature component fit to the data from 24 to 500 @xmath8 m with @xmath52 .
[ figbb],title="fig : " ] m with @xmath52 ( solid ) , @xmath44 ( dotted ) and @xmath49 ( dashed ) overlaid .
( bottom ) a two temperature component fit to the data from 24 to 500 @xmath8 m with @xmath52 .
[ figbb],title="fig : " ] lcl property & value & notes + @xmath90 & @xmath42 m@xmath1 & @xmath52 , graphite + silicate grains + @xmath91 & @xmath92 & 2@xmath11 upper limit + @xmath93 & @xmath94 & 2@xmath11 upper limit + @xmath95 & @xmath96 & assumes uniform ionized medium ( [ gd ] ) + @xmath75 & @xmath55 m@xmath1 & ... + @xmath97 & @xmath58 & ... + | data from the _ herschel space observatory _ have revealed an unusual elliptical galaxy , ngc 4125 , which has strong and extended submillimeter emission from cold dust but only very strict upper limits to its co and hi emission .
depending on the dust emissivity , the total dust mass is @xmath0 m@xmath1 . while the neutral gas - to - dust mass ratio is extremely low ( @xmath2 ) , including the ionized gas traced by [ cii ] emission raises this limit to @xmath3 .
the dust emission follows a similar @xmath4 profile to the stellar light and the dust to stellar mass ratio is towards the high end of what is found in nearby elliptical galaxies .
we suggest that ngc 4125 is currently in an unusual phase where evolved stars produced in a merger - triggered burst of star formation are pumping large amounts of gas and dust into the interstellar medium . in this scenario ,
the low neutral gas - to - dust mass ratio is explained by the gas being heated to temperatures @xmath5 k faster than the dust is evaporated .
if galaxies like ngc 4125 , where the far - infrared emission does not trace neutral gas in the usual manner , are common at higher redshift , this could have significant implications for our understanding of high redshift galaxies and galaxy evolution . |
rooted trees , and more generally digraphs , are widely used to represent evolutionary relationships in biology @xcite .
such a digraph typically has a set @xmath1 of labelled leaves ( vertices of outdegree 0 ) which corresponds to the collection of present - day species under study .
the arcs in the digraph are directed away from a single ` root ' vertex which represents the evolutionary ancestor of the species in @xmath1 .
the remaining vertices of the digraph are usually unlabelled ; they represent hypothetical ancestral species , and an arc from @xmath2 to @xmath3 indicates that ancestral species @xmath2 contributes directly to the genetic makeup of @xmath3 .
the simplest phylogenetic networks are trees , and these have traditionally been used when evolution is described purely by the formation of each new species from an existing one .
however , processes of reticulate evolution ( in particular the formation of hybrid species , and lateral gene transfer ) mean that a new species can have genetic contributions from more than one ancestral species , which is the reason digraphs are increasingly seen as a desirable model in molecular systematics @xcite . more precisely , if one considers the evolution of a particular gene , its evolution will generally be described by a tree , but reticulate processes mean that different genes can have different evolutionary tree structures , that can only be adequately reconciled by fitting them into a phylogenetic network .
this raises a fundamental question in computational phylogenetics : given a tree and a phylogenetic network , both with leaf set @xmath1 , is there an efficient algorithm to determine whether the tree ` fits inside ' the network ( in a sense we define precisely below ) ?
in general this question is np - hard , even for certain restricted classes of phylogenetic networks .
however , we show that , for particular classes , there exist polynomial - time algorithms which also provide a recipe for finding an explicit embedding of the tree in the phylogenetic network . in addition , we consider the computational complexity of deciding whether a subset of @xmath1 is a cluster of some tree that sits inside a given network .
the structure of this article is as follows . after giving formal definitions in the next section
, we present our polynomial - time algorithms for the ` tree containment ' problem in section [ sec : polytime ] , discuss the ` cluster containment ' problem in section [ sec : cluster ] , show the computational intractability of both problems for certain classes of networks in section [ sec : hard ] , and finish with an open problem in section [ sec : open ] .
consider a set @xmath1 of taxa .
a _ rooted phylogenetic network _ ( _ network _ for short ) on @xmath1 is a directed acyclic ( simple ) graph with a single _ root _
( vertex with indegree 0 ) , _ leaves _ ( vertices with outdegree 0 ) bijectively labeled by @xmath1 and no vertices with indegree and outdegree one .
we identify each leaf with its label and refer to the directed edges ( arcs ) as _
edges_. a vertex is called a _ reticulation vertex _ ( _ reticulation _ for short ) if it has indegree at least two and a _ tree - vertex _ otherwise .
an edge is called a _ tree - edge _ if it ends in a tree - vertex and it is called a _ reticulation - edge _ otherwise .
tree - path _ is a directed path that contains only tree - edges . a network with no reticulations is said to be a _ rooted phylogenetic tree _ ( _ tree _ for short ) .
a network is said to be _ binary _ if each reticulation has indegree two and outdegree one and all of the vertices have outdegree at most two
. given two vertices @xmath4 ( of some tree or network ) , we use @xmath5 to denote that there is a directed path from @xmath6 to @xmath7 . if @xmath5 and @xmath8 , we write @xmath9 . for a vertex @xmath3
, we define @xmath10 .
if @xmath11 is an edge , then we say that @xmath2 is a _ parent _ of @xmath3 and @xmath3 is a _ child _ of @xmath2 .
if two vertices @xmath4 have a common parent , then they are said to be _
an edge @xmath12 of a network @xmath13 is a _ cut - edge _ if removal of @xmath12 disconnects the undirected graph underlying @xmath13 .
a cut - edge is _ trivial _ if it ends in a leaf .
given a tree @xmath14 and network @xmath13 , we say that @xmath13 _ displays _
@xmath14 if there is a subgraph @xmath15 of @xmath13 that is a subdivision of @xmath14 ( i.e. @xmath15 can be obtained from @xmath14 by replacing edges by directed paths ) . for a tree @xmath14 on @xmath1 and a subset @xmath16
, we use @xmath17 to denote the tree on @xmath18 that is displayed by @xmath14 .
these concepts are illustrated in fig .
[ fig : leo ] . .
this network is normal , but not binary and has one reticulation vertex . _
middle _ : one of the two rooted phylogenetic trees on @xmath19 that are displayed by this network .
_ right _ : if we add either of the dashed edges , the resulting network is no longer normal ; adding the steeper , left - most arrow violates condition ( r3 ) while adding the right - most arrow results in a network that violates the tree - child property for the parent of @xmath20 and @xmath21 . ] in the first part of this paper we consider the following fundamental decision problem : lp.85 + _ instance : _ & a set @xmath1 of taxa , a rooted phylogenetic network @xmath13 on @xmath1 and a rooted phylogenetic tree @xmath14 on @xmath1 .
+ _ question : _ & does @xmath13 display @xmath14 ?
+ this problem is known to be np - complete @xcite for the general class of rooted , binary phylogenetic networks . here
we study the restriction of this problem to some established classes of rooted phylogenetic networks .
we then consider the complexity of a related ` cluster containment ' problem . a phylogenetic network @xmath13 is said to be _ regular _ if for any two distinct vertices @xmath22 of @xmath13 1 .
2 . @xmath24 if and only if @xmath25 ; 3 .
there is no edge @xmath11 if there is also a directed path from @xmath2 to @xmath3 of length greater than one .
a phylogenetic network @xmath13 is said to be _ tree - child _
( see e.g. @xcite ) if each vertex of @xmath13 either is a leaf or has a child that is a tree - vertex .
it follows immediately that from each vertex of a tree - child network there exists a tree - path to a leaf .
a network is said to be _ normal _ if it is tree - child , has no vertices of outdegree 1 , and in addition condition ( r3 ) above holds .
it has been shown that any normal network is regular , i.e. that automatically conditions ( r1 ) and ( r2 ) hold @xcite .
this section describes our main results , polynomial - time algorithms for tree containment restricted to normal networks and to binary tree - child networks .
we first give an algorithm for normal networks in section [ sec : normal ] , because this algorithm is simpler and will be used as a subroutine for the algorithm for binary tree - child networks in section [ sec : treechild ] .
we show that , given a normal phylogenetic network @xmath13 and a phylogenetic tree @xmath14 , one can decide in polynomial time whether @xmath13 displays @xmath14 .
we propose the following algorithm .
an example is in figure [ fig : normalalg ] .
* algorithm * locatenormal + for each reticulation @xmath26 of @xmath13 , do the following . 1 .
find a leaf @xmath27 that can be reached from @xmath26 by a tree - path .
2 . construct a set @xmath18 consisting of @xmath27 and , for each parent @xmath28 of @xmath26 , a leaf that can be reached from @xmath28 by a tree - path .
if @xmath27 does not have a sibling in @xmath17 , report that @xmath13 does not display @xmath14 .
4 . if @xmath27 does have a sibling in @xmath17 , let @xmath29 be such a sibling and @xmath30 the reticulation from which @xmath29 can be reached by a tree - path .
delete all edges entering @xmath26 except for @xmath31 .
when there are no reticulations left , the network has been transformed into a phylogenetic tree @xmath15 . check if @xmath15 is a subdivision of @xmath14 .
if it is , report that @xmath13 displays @xmath14 .
otherwise , report that @xmath13 does not display @xmath14 . in ( a ) and normal network @xmath13 in ( b ) .
dotted edges are removed from @xmath13 ; in ( b ) because @xmath32 and @xmath20 are siblings in @xmath33 , in ( c ) because @xmath21 and @xmath34 are siblings in @xmath35 , and in ( d ) because @xmath34 and @xmath12 are siblings in @xmath36 . ]
[ thm : normal ] given a normal phylogenetic network @xmath13 on @xmath1 and a phylogenetic tree @xmath14 on @xmath1 , algorithm locatenormal decides in polynomial time whether @xmath13 displays @xmath14 .
moreover , if @xmath13 displays @xmath14 , locatenormal finds the unique subtree of @xmath13 that is a subdivision of @xmath14 .
we first show that , for any reticulation @xmath26 , the leaves in the set @xmath18 described by the algorithm ( which exist by the tree - child property ) are always distinct . to show this ,
assume that two such tree - paths end in the same leaf .
since tree - paths can not combine in a reticulation , the only remaining possibility is that , for two parents @xmath37 of @xmath26 , the tree - path from @xmath38 to a leaf passes through @xmath39 .
however , this implies the existence of a directed path @xmath40 of length greater than one , contradicting ( r3 ) , since there is also an edge @xmath41 .
we now prove that locatenormal reports that @xmath13 displays @xmath14 if and only if this is the case .
if the algorithm reports that @xmath13 displays @xmath14 , then it is clear that @xmath13 indeed displays @xmath14 , since the algorithm checks if the constructed subtree @xmath15 of @xmath13 is a subdivision of @xmath14 .
now suppose that @xmath13 displays @xmath14 .
then there exists a subtree @xmath15 of @xmath13 that is a subdivision of @xmath14 .
take any reticulation @xmath26 and @xmath18 as in the algorithm .
observe that the root of @xmath15 is the root of @xmath13 by property ( r1 ) .
thus , by this observation and the fact that there is a tree - path from @xmath26 to a leaf , @xmath26 and exactly one of the edges entering @xmath26 are included in @xmath15 .
let @xmath42 be the edge entering @xmath26 that is included in @xmath15 and @xmath43 the leaf in @xmath18 that can be reached from @xmath44 by a tree - path .
then @xmath43 is a sibling of @xmath27 in @xmath45 ( which is isomorphic to @xmath17 ) and hence @xmath46 .
it follows that locatenormal deletes all reticulation - edges not included in @xmath15 , thus reconstructing @xmath15 .
moreover , since there is no choice which reticulation - edges to delete , the subtree @xmath15 is unique .
the running - time is clearly polynomial in the size of the input .
moreover , the number of vertices of a normal network grows at most quadratically with the number of leaves @xcite .
hence , the running - time is also polynomial in the number of leaves .
this subsection shows that , given a binary tree - child phylogenetic network @xmath13 and a phylogenetic tree @xmath14 , one can decide in polynomial time whether @xmath13 displays @xmath14 .
we partition the reticulations of @xmath13 into four types .
+ * type i. * there is no tree - path between the parents of the reticulation . + * type ii .
* there is a singe edge connecting the parents of the reticulation . +
* type iii .
* there is a tree - path between the parents of the reticulation that contains a vertex that is the parent of two tree - vertices .
+ * type iv .
* there is a tree - path between the parents of the reticulation that contains at least one internal vertex but no vertex that is the parent of two tree - vertices .
note that in a normal network all reticulations are of type i , because of restriction ( r3 ) .
thus , the difficulty in generalizing the algorithm for normal networks in section [ sec : normal ] is that the algorithm for tree - child networks below also has to take reticulations of types ii , iii and ( especially ) iv into account . * algorithm * locatetreechild + repeat the following four steps until none is applicable . +
* if there is a reticulation of type i , proceed as for normal networks . + * step 2 . *
if there is a reticulation of type ii , choose one of its two incoming edges arbitrarily and remove it .
+ * step 3 .
* if there is a reticulation @xmath26 of type iii , do the following . 1 .
let @xmath38 and @xmath39 be the parents of @xmath26 such that @xmath47 . pick a vertex @xmath3 that lies on the path from @xmath38 to @xmath39 and is the parent of two tree - vertices .
let @xmath21 be the child of @xmath3 that does not lie on the path between the parents of @xmath26 .
let @xmath48 and @xmath49 be leaves that can be reached from @xmath50 and @xmath21 respectively by tree - paths .
if @xmath51 and @xmath27 are siblings in @xmath52 , remove the edge @xmath41 from @xmath13 .
3 . if @xmath51 and @xmath49 are siblings in @xmath52 , remove the edge @xmath53 from @xmath13 .
4 . otherwise ,
report that @xmath13 does not display @xmath14
. * step 4 .
* if there is a reticulation of type iv , do the following . 1 .
if @xmath13 contains a nontrivial cut - edge @xmath11 , let @xmath54 be the set of taxa reachable from @xmath3 . find a vertex @xmath55 of @xmath14 with @xmath56 ( if there is no such vertex , report that @xmath13 does not display @xmath14 ) .
construct @xmath57 from @xmath14 by deleting all descendants of @xmath55 and labeling @xmath55 by a new taxon @xmath58 .
construct @xmath59 from @xmath13 by deleting all descendants of @xmath3 and labeling @xmath3 by @xmath49 .
construct @xmath60 by restricting @xmath13 to the vertices and edges reachable from @xmath3 .
run the algorithm recursively for @xmath61 and @xmath62 . if @xmath59 displays @xmath57 and @xmath60 displays @xmath63 , report that @xmath13 displays @xmath14
otherwise , report that @xmath13 does not display @xmath14 .
2 . contract all vertices with indegree and outdegree 1 .
find a leaf @xmath64 that has a reticulation as sibling . construct a tree - path @xmath65 as follows .
initialise @xmath65 as @xmath64 and , as long as the first vertex @xmath3 of @xmath65 has a reticulation as sibling , add the parent of @xmath3 to @xmath65 . construct a set @xmath18 by including @xmath64 and , for each reticulation that has a parent on @xmath65 , a leaf that can be reached from this reticulation by a tree - path .
if @xmath64 does not have a sibling in @xmath17 , report that @xmath13 does not display @xmath14 .
4 . if @xmath64 does have a sibling @xmath29 in @xmath17 , let @xmath66 be the reticulation from which @xmath29 can be reached by a tree - path .
let @xmath67 and @xmath68 be the parents of @xmath66 such that @xmath69 . remove the edge @xmath70 from @xmath13 . when there are no more reticulations left , check if the resulting phylogenetic tree @xmath15 is a subdivision of @xmath14 .
if it is , report that @xmath13 displays @xmath14 .
otherwise , report that @xmath13 does not display @xmath14 .
[ thm : treechild ] given a binary tree - child phylogenetic network @xmath13 on @xmath1 and a phylogenetic tree @xmath14 on @xmath1 , algorithm locatetreechild decides in polynomial time whether @xmath13 displays @xmath14 .
if there exist reticulations of type i or iii , it can be shown analogously to the proof of theorem [ thm : normal ] that steps 1 and 3 of locatetreechild either resolve these reticulations in the only possible way , or conclude that @xmath13 does not display @xmath14 .
it is furthermore clear that step 2 correctly deals with reticulations of type ii ( because @xmath13 is binary ) and step 4.1 with nontrivial cut - edges .
thus , from now on , we assume that all reticulations are of type iv and that there are no nontrivial cut - edges .
we first show that a leaf @xmath64 as described by step @xmath71 indeed exists , i.e. we show that there is a leaf that has a sibling which is a reticulation .
consider a tree - path @xmath72 of maximum length . because @xmath72 is a tree - path
, it ends in a tree - vertex @xmath64 , which must be a leaf because @xmath13 is tree - child and @xmath72 is of maximum length .
the single parent of @xmath64 is a tree - vertex because , if it were a reticulation , @xmath72 would consist of just @xmath64 but , by the existence of a type iv reticulation , there exists a tree - path of length at least two .
thus , leaf @xmath64 has a sibling , which has to be a reticulation because otherwise it would be possible to extend @xmath72 to a longer tree - path .
this shows the existence of a leaf @xmath64 as described in step @xmath71 .
we now show that the algorithm reports that @xmath13 displays @xmath14 precisely when this is the case .
the algorithm checks if the constructed subgraph @xmath15 is a subdivision of @xmath14 , so it will clearly never report that @xmath13 displays @xmath14 if this is not the case .
so assume that @xmath13 does display @xmath14 .
it remains to show that the algorithm will construct some subdivision of @xmath14 in @xmath13 .
observe that , apart from @xmath64 , all vertices on the path @xmath65 have two children : one reticulation and one tree - vertex ( which lies on @xmath65 ) .
let @xmath73 be the set of all reticulations with a parent on @xmath65 .
thus , @xmath18 consists of @xmath64 and , for each reticulation in @xmath73 , a leaf that can be reached from this reticulation by a tree - path . by the choice of @xmath64 , the path @xmath65 contains at least one other vertex , and so the set @xmath18 contains at least two leaves .
we claim that all parents of reticulations in @xmath73 lie on @xmath65 .
assume that this is not the case .
then there is a reticulation @xmath74 with two parents @xmath75 such that @xmath76 and one of @xmath75 is not on @xmath65 while the other one is .
since @xmath65 and the path @xmath76 both contain only tree - edges , we must have that @xmath77 is on @xmath65 while @xmath78 is not .
however , then the path @xmath79 contains a vertex that is the parent of two tree - vertices , which means that @xmath74 is of type iii .
this is a contradiction because the algorithm has already resolved all reticulations of type iii in step 3 .
thus , apart from @xmath64 , @xmath65 consists precisely of the parents of reticulations in @xmath73 . since @xmath13 displays @xmath14
, there exists a subtree @xmath15 of @xmath13 that is a subdivision of @xmath14 .
we next claim that @xmath64 has a sibling in @xmath45 .
let @xmath80 be the last vertex on the path @xmath65 for which the reticulation - edge leaving @xmath80 is included in @xmath15 ( such a vertex exists because one of the edges entering the sibling of @xmath64 is included in @xmath15 ) .
let @xmath29 be the leaf in @xmath18 that can be reached from the reticulation - child of @xmath80 by a tree - path .
clearly , @xmath29 and @xmath64 are siblings in @xmath45 ( which is isomorphic to @xmath17 ) .
thus , the algorithm removes the edge @xmath70 from @xmath13 , with @xmath66 the reticulation - child of @xmath80 and @xmath67 equal to either @xmath80 or the other parent of @xmath66 .
first suppose that @xmath67 is not equal to @xmath80 .
thus , @xmath81 and @xmath67 is the other parent of @xmath66 . in this case
, the algorithm correctly removes the edge @xmath70 that is not in @xmath15 and we are done .
now assume that @xmath82 and thus that the algorithm removes the edge @xmath70 that is in @xmath15 .
we replace in @xmath15 the edge @xmath70 by the edge @xmath83 .
the resulting subgraph of @xmath13 is again a subdivision of @xmath14 , because none of the reticulation - edges leaving vertices below @xmath82 were included in @xmath15 ( by the choice of @xmath80 ) .
see figure [ fig : treechildproof ] for an illustration .
we conclude that the algorithm correctly constructs a subdivision of @xmath14 in @xmath13 .
the running - time is clearly polynomial in the size of the input .
moreover , the number of reticulations in a tree - child network is at most @xmath84 and it can be shown by induction on the number @xmath85 of reticulations that the number of edges of a binary tree - child network is at most @xmath86 . hence , the running - time is also polynomial in the number of leaves . .
a subdivision of the tree @xmath14 in the network @xmath13 is indicated by solid lines .
replacing edge @xmath70 by @xmath83 gives another subdivision of @xmath14 in @xmath13 , because @xmath70 is the last edge leaving the path @xmath65 ( the vertical path ) that is used by the subdivision . ]
a _ cluster _ is a strict subset of @xmath1 .
there are two different ways of seeing clusters in networks .
let @xmath3 be a vertex of a network @xmath13 on @xmath87 .
the _ hardwired cluster _ of @xmath3 , denoted @xmath88 , is the set containing all taxa that can be reached from @xmath3 .
a cluster is said to be a hardwired cluster of @xmath13 if it is the hardwired cluster of some vertex of @xmath13 .
a cluster @xmath89 is said to be a _ softwired cluster _ of @xmath3 if @xmath54 equals the set of all taxa that can be reached from @xmath3 when , for each reticulation @xmath26 , exactly one incoming edge of @xmath26 is `` switched on '' and the other incoming edges of @xmath26 are `` switched off '' .
notice that each vertex has exactly one hardwired cluster and one or more softwired clusters .
we say that @xmath54 is a _ softwired cluster _ of a network @xmath13 if it is a softwired cluster of some vertex of @xmath13 .
the softwired clusters of @xmath13 can elegantly be characterized as follows .
a cluster @xmath89 is a softwired cluster of @xmath13 if and only if there exists a tree @xmath14 such that @xmath54 is a hardwired cluster of @xmath14 and @xmath14 is displayed by @xmath13 .
we are interested in the following decision problem .
lp.85 + _ instance : _ & a set @xmath1 of taxa , a rooted phylogenetic network @xmath13 on @xmath87 and a cluster @xmath90 .
+ _ question : _ & is @xmath54 a softwired cluster of @xmath13 ? + this problem is known to be np - complete @xcite for general binary phylogenetic networks . however , it has a polynomial - time algorithmic solution if we restrict @xmath13 to tree - child networks @xcite
. we will show below ( theorem [ thm : reghard ] ) that , if we extend the class of normal networks to regular networks , the problem cluster containment becomes np - hard , even if we add further structural restrictions to this class of networks .
but first we describe yet another class of networks for which both problems tree containment and cluster containment have polynomial - time algorithms .
we finish this section by observing that both tree containment and cluster containment are polynomial - time solvable for the class of binary level-@xmath0 networks .
a binary network is _ biconnected _ if it contains no cut - edges .
a biconnected subgraph @xmath91 of a binary network @xmath13 is said to be a _ biconnected component _ if there is no biconnected subgraph @xmath92 of @xmath13 that contains @xmath91 .
a binary phylogenetic network is a _
@xmath93 network _
if each biconnected component has at most @xmath0 reticulations .
[ obs : levelk ] tree containment and cluster containment are polynomial - time solvable when restricted to binary level-@xmath0 networks , for any fixed @xmath0 .
we may assume that the network contains no nontrivial cut - edges , because step 4.1 of algorithm locatetreechild can be applied until there are no nontrivial cut - edges left .
a level-@xmath0 network with no nontrivial cut - edges contains at most @xmath0 reticulations .
thus , we can loop through all @xmath94 ways of selecting one incoming edge for each of the @xmath0 reticulations . for each of the resulting trees
, we check whether it is a subdivision of the input tree ( in case of the tree containment problem ) or check whether the input cluster is a cluster of the obtained tree ( in case of the cluster containment problem ) .
a phylogenetic network is said to be _ tree - sibling _ ( see e.g. @xcite ) if each reticulation has a sibling that is a tree - vertex .
a phylogenetic network @xmath13 is said to be _ time - consistent _ ( see e.g. @xcite ) if it is possible to assign each vertex @xmath3 of @xmath13 a `` time stamp '' @xmath95 such that for each edge @xmath11 of @xmath13 : 1 .
@xmath96 if @xmath11 is a tree - edge and 2 .
@xmath97 if @xmath11 is a reticulation - edge .
[ thm : reghard ] tree containment and cluster containment are both np - complete when restricted to tree - sibling , time - consistent , regular phylogenetic networks .
we reduce from the tree containment and cluster containment problems on general networks , which were shown to be np - complete by kanj et al .
@xcite .
let @xmath1 be a set of taxa , and let @xmath13 and @xmath14 be a network and tree on @xmath1 , respectively .
we will modify @xmath13 to a tree - sibling , time - consistent , regular network @xmath98 on a set of taxa @xmath99 and show that a cluster @xmath89 is a softwired cluster of @xmath98 if and only if it is a softwired cluster of @xmath13 . since this modification can be carried out in polynomial time and the size of @xmath98
is polynomial in the size of @xmath13 , this shows np - hardness of cluster containment on tree - sibling , time - consistent , regular networks .
furthermore , we will ( in polynomial time ) modify @xmath14 to a tree @xmath15 on @xmath100 and show that @xmath98 displays @xmath15 if and only if @xmath13 displays @xmath14 , thus showing np - hardness of tree containment on this restricted class of networks .
the construction of @xmath98 and @xmath15 relies on repeatedly applying the following operation . for a vertex @xmath3 of @xmath13 ,
define hangleaves@xmath101 as making the following changes to @xmath13 , @xmath14 and @xmath1 .
we add two new taxa @xmath102 to @xmath1 .
let @xmath26 be the root of @xmath13 .
we add to @xmath13 leaves @xmath102 , a new root @xmath66 , an internal vertex @xmath28 and edges @xmath103 and @xmath104 .
let @xmath105 be the root of @xmath14 .
we add to @xmath14 the leaves @xmath102 , a new root @xmath106 , an internal vertex @xmath107 and edges @xmath108 and @xmath109 .
see figure [ fig : modification ] .
modifies @xmath13 and @xmath14 .
vertices @xmath26 and @xmath105 are the roots of @xmath13 and @xmath14 respectively . ]
we now describe how we transform @xmath13 into a tree - sibling , time - consistent , regular network by repeated applications of hangleaves .
first , we make the network regular by doing the following for each pair @xmath22 of distinct vertices of @xmath13 . 1 .
if @xmath110 , then apply hangleaves@xmath111 and hangleaves@xmath101 .
2 . if @xmath25 but there is no directed path from @xmath2 to @xmath3 , then apply hangleaves@xmath101 .
if there exist two distinct directed paths from @xmath2 to @xmath3 , one of which is an edge , then subdivide this edge with a single vertex @xmath112 and apply hangleaves@xmath113 .
for any two vertices @xmath22 of @xmath13 , operations 1 , 2 and 3 make sure that properties ( r1 ) , ( r2 ) and ( r3 ) ( respectively ) of a regular network are satisfied .
furthermore , by the definition of hangleaves , these properties are also satisfied for newly added vertices .
it follows that the obtained network is regular .
call this network @xmath114 .
the next step is to make the network time - consistent . for each reticulation - edge @xmath11 of @xmath114 ,
subdivide @xmath11 by a new vertex @xmath112 and apply hangleaves@xmath113 .
let @xmath115 be the resulting phylogenetic network .
we claim that @xmath115 is time - consistent .
let @xmath116 be a labeling of the vertices of @xmath114 such that @xmath117 for each edge @xmath11 .
this is possible because @xmath114 is acyclic .
we specify a label @xmath118 ( the time - stamp ) for each vertex @xmath3 of @xmath115 as follows .
each vertex of @xmath115 that is also a vertex of @xmath114 gets the same label as in @xmath114 , i.e. @xmath119 for all @xmath120 .
now consider a reticulation - edge @xmath11 of @xmath114 .
such an edge corresponds to two edges @xmath121 and @xmath122 of @xmath115 .
now label vertex @xmath112 the same as vertex @xmath3 , i.e. @xmath123 .
observe that @xmath122 is a reticulation - edge and now satisfies restriction ( tc2 ) of time - consistency .
furthermore , we have @xmath124 and so @xmath121 , which is a tree - edge , satisfies restriction ( tc1 ) of time - consistency .
it remains to label the vertices that have been added by hangleaves .
this can easily be done in such a way that the restrictions of time - consistency are satisfied .
namely , we give @xmath64 and @xmath28 the same label as @xmath3 , give @xmath66 any label that s smaller than @xmath125 and @xmath43 any label that is greater than @xmath126 ( processing vertices in the same order as in which they have been added by applications of hangleaves ) .
finally , we make the network tree - sibling . for each reticulation
@xmath26 of @xmath115 , that has not been added by hangleaves , we do the following .
observe that , as a result of the modifications that made @xmath115 time - consistent , @xmath26 has two siblings , both of which are reticulations added by two different applications of hangleaves .
pick any of the two siblings and call it @xmath64 .
let @xmath3 be the common parent of @xmath64 and @xmath26 .
subdivide edge @xmath104 by a new vertex @xmath112 and apply hangleaves@xmath113 .
reticulation @xmath26 now has a sibling that is a tree - vertex , namely @xmath112 .
moreover , all reticulations @xmath64 added by applications of hangleaves have a sibling @xmath43 that is a tree - vertex .
hence , the resulting network is tree - sibling
. let @xmath98 be this resulting network and @xmath15 the resulting tree .
we claim that @xmath98 is not only tree - sibling , but also still regular and time - consistent . to see that @xmath98 is regular , observe that it has been obtained from the regular network @xmath114 by repeatedly subdividing an edge by a new vertex @xmath112 and applying hangleaves@xmath113 .
it can easily be checked that a regular network remains regular after such a modification . to see that @xmath98 is also time - consistent , observe that it has been constructed from time - consistent network @xmath115 by repeatedly subdividing an edge @xmath104 by a new vertex @xmath112 and applying hangleaves@xmath113 . using that @xmath64 is a leaf added by hangleaves ,
it can easily be checked that a time - consistent network remains time - consistent after such a modification .
thus , @xmath98 is a tree - sibling , time - consistent , regular network .
it remains to show that ( i ) a cluster @xmath89 is a softwired cluster of @xmath98 if and only if it is a softwired cluster of @xmath13 and ( ii ) @xmath98 displays @xmath15 if and only if @xmath13 displays @xmath14 . the crux to showing these things is that @xmath98 and @xmath15 have been obtained from @xmath13 and @xmath14 by subdividing edges and applying hangleaves . by this observation ,
( i ) is clear . to see
( ii ) , let @xmath127 , @xmath128 and @xmath129 be the result of a single application of hangleaves@xmath101 to @xmath130 .
we claim that @xmath127 displays @xmath128 if and only if @xmath13 displays @xmath14 . first note
that an embedding of @xmath14 in @xmath13 can easily be extended to an embedding of @xmath128 in @xmath127 by adding to the embedding all new vertices , a path from @xmath66 to the root of the embedding and edges @xmath131 , @xmath132 , @xmath133 .
now consider an embedding of @xmath128 in @xmath127 .
since @xmath64 and @xmath43 are siblings in @xmath128 , this embedding necessarily contains the newly added vertices and edges except for the edge @xmath104 .
thus , the restriction of the embedding of @xmath128 in @xmath127 to an embedding of @xmath14 in @xmath127 does not contain any of the newly added vertices and edges and is thus an embedding of @xmath14 in @xmath13 .
thus , @xmath127 displays @xmath128 if and only if @xmath13 displays @xmath14 . by recursively applying this argument
, it follows that @xmath98 displays @xmath15 if and only if @xmath13 displays @xmath14 .
for a vertex @xmath3 and a leaf @xmath64 of some phylogenetic network , we say that @xmath3 is a _ stable ancestor _ of @xmath64 if all directed paths from the root to @xmath64 pass through @xmath3 .
a network is said to be _ reticulation - visible _ if each reticulation is a stable ancestor of some leaf .
recently , it was shown that cluster containment is polynomial - time solvable for reticulation - visible networks @xcite .
this class of networks contains , but is more general than , the class of tree - child networks .
thus , the tantalizing question remaining open after this work is whether tree containment is also polynomial - time solvable for reticulation - visible networks .
note that tree containment can not simply be solved by checking if each cluster of the input tree @xmath14 is a softwired cluster of the input network @xmath13 ( using an algorithm for cluster containment ) .
this approach fails because , even if all clusters of @xmath14 are softwired clusters of @xmath13 , and even if @xmath13 is reticulation - visible , it might be that @xmath13 does not display @xmath14 , see @xcite .
thus , there is no obvious reduction from cluster containment to tree containment or vice versa .
l. nakhleh , evolutionary phylogenetic networks : models and issues . in : _ the problem solving handbook for computational biology and bioinformatics _ , l. heath and n. ramakrishnan ( editors ) .
springer , to appear . | _ phylogenetic trees _ and _ networks _ are leaf - labelled graphs that are used to describe evolutionary histories of species
. the tree containment problem asks whether a given phylogenetic tree is embedded in a given phylogenetic network .
given a phylogenetic network and a cluster of species , the cluster containment problem asks whether the given cluster is a cluster of some phylogenetic tree embedded in the network .
both problems are known to be np - complete in general . in this article
, we consider the restriction of these problems to several well - studied classes of phylogenetic networks .
we show that tree containment is polynomial - time solvable for normal networks , for binary tree - child networks , and for level-@xmath0 networks . on the other hand , we show that , even for tree - sibling , time - consistent , regular networks , both tree containment and cluster containment remain np - complete .
algorithms , computational complexity , phylogenetic trees , phylogenetic networks |
in recent years there has been tremendous progress in the computation of baryon masses within lattice qcd .
still , it remains necessary to extrapolate the calculated results to the physical pion mass ( @xmath0 mev ) in order to make a comparison with experimental data . in doing
so one necessarily encounters some non - linearity in the quark mass ( or @xmath1 ) , including the non - analytic behavior associated with dynamical chiral symmetry breaking .
we recently investigated this problem for the case of the nucleon magnetic moments @xcite .
it is vital to develop a sound understanding of how to extrapolate to the physical pion mass .
chiral symmetry is dynamically broken in qcd and the pion alone is a near goldstone boson .
it is strongly coupled to baryons and plays a significant role in @xmath2 and @xmath3 self - energies .
the one - loop pion induced self - energies of the @xmath2 and @xmath3 are given by the processes shown in fig .
[ se - fig ] . in the standard heavy baryon limit ,
the analytic expression for the pion cloud correction to the masses of the @xmath2 and @xmath3 are of the form @xcite @xmath4 where @xmath5 @xmath6 @xmath7 here @xmath8 , @xmath9 is the axial charge of the nucleon , @xmath10 is the pion energy and @xmath11 , @xmath12 , @xmath13 are the @xmath14 , @xmath15 , @xmath13 form factors associated with the emission of a pion of three - momentum @xmath16 .
the form factors reflect the finite size of the baryonic source of the pion field and suppress the emission probability at high virtual pion momentum . as a result
, the self - energy integrals are not divergent .
the leading non - analytic ( lna ) contribution of these self - energy diagrams is associated with the infrared behavior of the corresponding integrals ; i.e. the behavior as @xmath17 . as a consequence ,
the leading non - analytic behavior does not depend on the details of the form factors .
indeed , the well known results of chiral perturbation theory @xcite are reproduced even when the form factors are approximated by @xmath18 . of course , our concern with respect to lattice qcd is not so much the behavior as @xmath19 , but the extrapolation from high pion masses to the physical pion mass . in this context
the branch point at @xmath20 , associated with transitions of @xmath21 or @xmath22 , is at least as important as the lna behavior near @xmath23 .
heavy quark effective theory suggests that as @xmath24 the quarks become static and hadron masses become proportional to the quark mass . in this spirit
, corrections are expected to be of order @xmath25 where @xmath26 is the heavy quark mass .
the presence of a cut - off associated with the form factor acts to suppress the pion induced self energy for increasing pion masses , as evidenced by the @xmath1 in the denominators of eqs .
( [ se - nn - integral - eqn ] ) , ( [ se - nd - integral - eqn ] ) and ( [ se - dn - integral - eqn ] ) . while some @xmath1 dependence in the form factor is expected , this is a second - order effect and does not alter the qualitative feature of the self - energy corrections tending to zero as @xmath27 in the heavy quark limit . rather than simplifying our expressions to just the lna terms
, we retain the complete expressions @xcite , as they contain important physics that would be lost by making a simplification .
we note that keeping the entire form is not in contradiction with @xmath28pt .
however , as one proceeds to larger quark masses , differences between the full forms and the expressions in the chiral limit will become apparent , highlighting the importance of the branch point and the form factor reflecting the finite size of baryons . as a result of these considerations ,
we propose to use the analytic expressions for the self - energy integrals corresponding to a sharp cut - off in order to incorporate the correct lna structure in a simple three - parameter description of the @xmath29 dependence of the @xmath2 and @xmath3 masses . in the heavy quark limit
hadron masses become proportional to the quark mass .
hence we can simulate a linear dependence of the baryon masses on the quark mass , @xmath30 , in this region , by adding a term involving @xmath31 .
the functional form for the mass of the nucleon suggested by this analysis is then : @xmath32 while that for the @xmath3 is : @xmath33 the use of a sharp cut - off , @xmath18 , as a form factor may seem somewhat unfortunate given that phenomenology suggests a dipole form factor better approximates the axial - vector form factor .
however , the sensitivity to such model - dependent issues is shown to be negligible in fig .
[ se - fit ] .
there , the self - energy contribution @xmath34 for a 1 gev dipole form factor ( solid curve ) is compared with a sharp cut - off form factor combined with the standard @xmath35 terms of ( [ n - form - eqn ] ) or ( [ d - form - eqn ] ) .
optimizing @xmath36 , @xmath37 and @xmath38 provides the fine - dash curve of fig . [ se - fit ] .
differences are at the few mev level indicating negligible sensitivity to the actual analytic structure of the form factor . here
we have focused on the pion self - energy contribution to the @xmath2 and @xmath3 form factors .
only the pion displays a rapid mass dependence as the chiral limit is approached .
other mesons participating in similar diagrams do not give rise to such rapidly changing behavior and can be accommodated in the @xmath35 terms of ( [ n - form - eqn ] ) or ( [ d - form - eqn ] ) . moreover
, the form factor suppresses the contributions from more massive intermediate states including multiple pion dressings .
other multi - loop pion contributions renormalize the vertex and hence we use the renormalized coupling @xmath39 as a measure of the pion - nucleon coupling .
we consider two independent dynamical - fermion lattice simulations of the @xmath2 and @xmath3 masses .
we select results from cp - pacs s @xcite @xmath40 and @xmath41 simulations at @xmath42 , and ukqcd s @xcite @xmath43 simulations at @xmath44 .
figure [ 3fits - n - fig ] displays fits of ( [ n - form - eqn ] ) to the lattice data . in order to perform fits in which @xmath36 is unconstrained
, it is essential to have lattice simulations at light quark masses approaching @xmath45 gev@xmath46 .
it is common to see the use of the following @xmath28pt - motivated expression for the mass dependence of hadron masses , @xmath47 the result of such a fit for the @xmath2 is shown as the dashed curve in fig .
[ 3fits - n - fig ] .
the coefficient of the @xmath48 term in a three parameter fit is @xmath49 .
this disagrees with the coefficient of @xmath50 known from @xmath28pt ( which is correctly incorporated in ( [ n - form - eqn ] ) and illustrated as the solid and dash - dot curves of fig .
[ 3fits - n - fig ] ) by almost an order of magnitude .
this clearly indicates the failings of ( [ naive - eqn ] ) . the dotted curve of fig .
[ 3fits - n - fig ] indicates the leading non - analytic term of the chiral expansion dominates from the chiral limit up to the branch point at @xmath51 mev , beyond which @xmath28pt breaks down .
the curvature around @xmath52 , neglected in previous extrapolations of the lattice data , leads to shifts in the extrapolated masses of the same order as the departure of lattice estimates from experimental measurements . | consideration of the analytical properties of pion - induced baryon self - energies leads to new functional forms for the extrapolation of light baryon masses .
these functional forms reproduce the leading non - analytic behavior of chiral perturbation theory , the correct heavy - quark limit and have the advantage of containing information on the extended structure of hadrons .
the forms involve only three unknown parameters which may be optimized by fitting to present lattice data .
recent dynamical fermion results from cp - pacs and uk - qcd are extrapolated using these new functional forms .
we also use these functions to probe the limit of the chiral perturbative regime and shed light on the applicability of chiral perturbation theory to the extrapolation of present lattice qcd results . |
experimental advances in nanofabrication and in laser cooling and trapping have turned optomechanical systems into viable laboratories for the observation of quantum mechanics at macroscopic scales .
non - equilibrium cooling of small movable mirrors using laser - driven cavities has been demonstrated experimentally by a number of groups @xcite .
likewise , theory has shown that in principle these methods should be able to lower the mirror to its quantum mechanical ground state @xcite .
the preparation of that state of the mirror is an important first step in exploring characteristic features of quantum mechanics such as superposition @xcite and entanglement in macroscopic systems @xcite .
squeezed states have also attracted much attention , due to their favorable quantum noise properties @xcite . squeezed states of light
are expected to find applications in precision measurements @xcite and optical communications @xcite . in a parallel development
, the squeezing of classical noise in mechanical oscillators has been demonstrated in optomechanical cavities @xcite , ion traps @xcite optical lattices @xcite and other systems @xcite .
quantum squeezing of phonons has been achieved in ion traps @xcite and in crystals @xcite . proposals to realize squeezed states of nanomechanical oscillators in the quantum regime
have been made involving two - mirror cavities @xcite , parametric mixing in solid state circuits @xcite , microwave coupling to a charge qubit @xcite , quantum non - demolition measurements @xcite and the parametric modulation of a mechanical spring @xcite .
their application to gravitational interferometry has also been discussed @xcite .
other nonclassical states such as schrodinger ` cats ' have been proposed using movable cavity mirrors @xcite .
this article shows how to realize a squeezed state of a mechanically moving mirror in a high finesse optical cavity .
previous proposals to achieve this goal have relied on the mathematical analogy between an optical resonator with a moving mirror and a kerr medium , and the mechanism of squeezing has been _
parametric driving_. here we invoke _ compression _ as an alternative route to squeezing @xcite . in that scheme squeezing of the mirror motion
relies on coupling it dispersively with the cavity , a possibility that has recently been pointed out @xcite and analyzed in detail @xcite . to provide a complete discussion we consider not one but two modes of the cavity , the moving mirror being coupled dispersively to one of the modes and dissipatively to the other
this configuration was recently proposed as an efficient cooling and trapping configuration for semi - transparent mirrors @xcite ; here we show that this configuration also allows for displacing and squeezing the error ellipse of the oscillating mirror in phase space . the remainder of the paper is organized as follows .
section [ sec : ham ] introduces the physical system and its model hamiltonian , section [ sec : ev ] discusses the corresponding evolution operator and the resulting displacement and squeezing assuming that the moving mirror starts from its quantum mechanical ground state .
section [ sec : therm ] discusses the effects of squeezing in the presence of noise and damping .
section [ sec : con ] supplies a conclusion and an outlook .
we consider a high finesse cavity with two perfectly reflecting fixed end mirrors , and a partially reflective movable middle mirror as shown in fig.1 . the middle mirror is assumed to execute small harmonic oscillations @xmath0 about its equilibrium position .
it couples dissipatively ( linearly in @xmath0 ) to a cavity mode @xmath1 of frequency @xmath2 and dispersively ( quadratically in @xmath0 ) to a second mode @xmath3 of frequency @xmath4 .
the hamiltonian @xmath5 modelling the system is derived in ref .
@xcite , and is given explicitly by @xmath6 where @xmath7 is the oscillation frequency of the middle mirror , @xmath8 with @xmath9 the equilibrium position of the moving mirror of transmissivity @xmath10 , @xmath11 , @xmath12 and @xmath13 , and @xmath14 where @xmath15 .
the frequencies of the modes @xmath1 and @xmath3 can be chosen such that @xmath16 are either positive or negative . in the case of @xmath17
this corresponds to the use of trapping and anti - trapping modes , respectively @xcite .
for @xmath18 we have @xmath19,\ ] ] and for @xmath20 @xmath21.\ ] ] similarly , @xmath22 for @xmath23 and @xmath24 for @xmath25 .
the first two terms in the hamiltonian @xmath5 describe the energies of the optical modes , the next two the energy of the oscillating mirror , and the last two the dissipative and dispersive coupling energies .
the bosonic modes obey the commutation relations @xmath26=1 $ ] and @xmath27=1 $ ] , and the dynamical variables of the oscillating mirror follow the canonical commutation relation @xmath28=i\hbar$ ] .
the hamiltonian ( [ eq : hq ] ) indicates that for low values of @xmath29 the spring potential energy dominates the anti - trapping due to radiation pressure , hence the middle mirror still behaves as a harmonic oscillator , but of lower frequency .
however , for @xmath30 increasing @xmath31 leads to a point @xmath32 where the mirror behaves as a free particle . for even higher values of @xmath31 radiation pressure - induced anti - trapping dominates and the mirror behaves like an inverted harmonic oscillator @xcite .
we do not consider that regime in this paper .
this is consistent with the assumption of small mirror displacements @xmath0 used to derive the hamiltonian ( [ eq : hsc ] ) , as well as with requirements of stability . in the following we consider a semiclassical version of the hamiltonian @xmath5 valid for situations where
the optical modes can be treated classically . in that case @xmath33 and expressing the mirror displacement in terms of raising and lowering operators @xmath34 we have @xmath35 where we have removed a constant energy @xmath36 and @xmath37 in the semiclassical hamiltonian @xmath38 we have also introduced the operators @xmath39 which together with @xmath40 and @xmath41 form the basis of the so - called two - photon lie algebra @xcite , with @xmath42&= & \pm k_\pm , \,\,\,\ , \left [ k_- , k_+\right ] = 2 k_0 , \nonumber \\ \left [ k_-,c \right ] & = & \left [ k_+,c^\dagger\right ] = 0 , \nonumber \\ \left [ k_-,c^\dagger \right ] & = & c , \,\,\,\ , \left [ k_0,c^\dagger \right ] = c^\dagger/2.\end{aligned}\ ] ] as is well known , the operators @xmath43 and @xmath44 form two sub - algebras , the associated operators forming the generators of coherent states and of squeezed states , respectively .
the hamiltonian ( [ eq : hsc ] ) has previously been studied in some detail in the context of molecular translational - vibrational interactions @xcite and laser - plasma scattering @xcite and very recently in the context of atomic vapors inside resonators @xcite .
see also @xcite for additional discussions of this model .
using the lie - algebraic symmetries of @xmath38 , the associated evolution operator can be disentangled as @xcite @xmath45=e^{i\delta } d(\nu)r(\phi)s(\kappa),\ ] ] where @xmath46 is an unimportant overall phase , and @xmath47 is a displacement operator , with @xcite @xmath48\ ] ] and @xmath49^{1/2}.\ ] ] in the bound oscillator regime , i.e. for @xmath50 , we have @xmath51 , and we can choose @xmath52 without loss of generality . that parameter largely determines the time scale of the mirror dynamics ; in the absence of squeezing @xmath53 it is just the harmonic oscillator period . the factor in parentheses in eq .
( [ eq : chi ] ) quantifies the mismatch from the condition @xmath54 [ eq . [ eq : crit ] ] which demarcates the regimes of qualitatively different physical behaviors in the system . from eqs .
( [ eq : complexnu ] ) and ( [ eq : chi ] ) @xmath55 can both increase or decrease the characteristic time scale of the displacement as well as its magnitude .
the displacement in phase space is given by the absolute value of @xmath56 .
a plot of @xmath57 versus time for typical experimental parameters is shown in fig.[fig : displace ] .
near the first minimum i.e. for @xmath58 the displacement is linear in time to lowest order , i.e. @xmath59 + \mathcal{o}\left[t^5\right ] \right|,\ ] ] and the effects of squeezing come in at third order .
interestingly , by adjusting the squeezing such that @xmath60 , which is still in the @xmath61 regime , the third - order time dependence of the displacement can be removed .
qualitatively similar behavior can be seen near every minimum in fig .
[ fig : displace ] .
we note that in the absence of squeezing @xmath53 , @xmath62 while for large squeezing @xmath63 , @xmath64 for typical parameters we have @xmath65 .
therefore for both small and large squeezing a coherent mechanical state of the middle mirror of relatively large amplitude can be produced starting from the ground state .
returning to the various components of the evolution operator @xmath66 we observe that @xmath67 is a rotation operator , and @xmath68 is a squeezing operator , with @xmath69 as expected that operator does not depend on @xmath70 , i.e. displacement does not affect squeezing .
it turns out that the rotation angle @xmath71 in eq .
( [ eq : rot ] ) is exactly opposite the angle at which the squeeze operator tilts the error ellipse of the moving mirror in phase space @xcite , i.e. @xmath72.\ ] ] the two rotations therefore cancel each other out and @xmath71 effectively plays no role in the dynamics .
it is in fact intuitively clear that the effects of rotation should cancel out , i.e. the axes of the final error ellipse should be aligned along @xmath73 and @xmath0 in phase space .
this is because eq .
( [ eq : hq ] ) stipulates that position is the only quadrature of the oscillating middle mirror that can be squeezed or anti - squeezed , the latter situation corresponding to momentum squeezing .
figure [ fig : squeeze ] shows @xmath74 versus time for typical experimental parameters . as can be seen from that plot the squeezing first grows linearly in time .
this can be confirmed by analytically expanding eq .
[ eq : modkappa ] for the case of @xmath75 @xmath76+\mathcal{o}\left[t^5\right]\right|.\ ] ] we note that the third - order time dependence can be removed for @xmath77 . actually under this condition it can readily be seen from eq .
( [ eq : modkappa ] ) that all higher orders vanish and squeezing is purely linear in time : @xmath78 .
however that case corresponds to @xmath79 , a situation where the mirror does not behave as a bound harmonic oscillator .
if the middle mirror is prepared in its quantum mechanical ground state , the squeezing operator ( [ eq : skappa ] ) produces a squeezed vacuum @xcite . from fig .
[ fig : squeeze ] , the maximum value of @xmath80 is approximately 4 , which implies a maximum squeezing of @xmath81 , or log@xmath82db of squeezing .
however the placement of a macroscopic nano - oscillator in its ground state has not yet been achieved experimentally , so we also consider thermal states of the middle mirror .
they are characterized by a thermal phonon number given by the bose distribution @xmath83^{-1},\ ] ] where @xmath84 is the mirror equilibrium temperature and @xmath85 is boltzmann s constant .
any realistic model should also include the damping of the mirror .
we estimate these effects by including noise and damping in the heisenberg equations of the mirror in a manner consistent with the fluctuation - dissipation theorem .
this produces the corresponding quantum langevin equations from eq .
( [ eq : hq ] ) in a standard way .
setting @xmath86 for simplicity and concentrating therefore solely on the squeezing part of the hamiltonian @xmath38 the quantum langevin equations turn out to be @xmath87 where @xmath88 is the damping constant of the mirror and @xmath89 represents brownian noise with average zero and fluctuations correlated as @xmath90 .
\nonumber \\\end{aligned}\ ] ] for a high mechanical quality factor , the brownian force becomes delta - correlated in the time domain @xcite . in fourier space the correlation can then be written as @xmath91 by setting the time derivatives equal to zero the steady - state solutions to eq .
( [ eq : qle ] ) can easily be found to be @xmath92 linearizing all operators in eq .
( [ eq : qle ] ) as sums of a semiclassical steady - state value and a small quantum fluctuation ( i.e. @xmath93 ) we obtain linear dynamical equations for the fluctuations . using fourier transforms and eq .
( [ eq : fcorr ] ) we can solve the fluctuation equations to obtain @xmath94 , etc .
we can therefore also find the ( equal - time ) correlation function for the position @xmath95 which is independent of time since the noise process we have considered is stationary [ eq . ( [ eq : brownian ] ) ] .
this result for the position uncertainty has followed from a linear response analysis , however it agrees to first order with results from more sophisticated computations @xcite .
for example in the absence of squeezing @xmath53 , and at high temperatures , @xmath96 , @xmath97 on the other hand for @xmath98 and low temperatures @xmath99 , @xmath100 which is just the square of the oscillator length of the ground state of the moving mirror . the results in eqs .
( [ eq : hot ] ) and ( [ eq : cold ] ) agree with an earlier and more rigorous derivation @xcite . using eq .
( [ fluctuations ] ) in the presence of squeezing @xmath101 and defining a position uncertainty @xmath102 in terms of the ground state oscillator length we find @xmath103^{1/2 } \sim \left(\frac{k_b t_{e}}{\hbar c_s } \right)^{1/2},\ ] ] where the last expression has been written in the limit of high temperature and high squeezing .
@xmath102 needs to be lower than 1 for squeezing to be present , i.e. the fluctuations in the position need to be smaller than the ground state uncertainty in position . for millikelvin temperatures and hundreds of milliwatts of laser power , @xmath104 and about 8db of mechanical squeezing can be obtained , which is still considerable .
in conclusion we have considered a partially reflective vibrating mirror coupled dispersively to an optical mode of a high finesse cavity .
we have shown that quantum squeezing of the mechanical motion of the mirror can be achieved in this way .
we have described the unitary dynamics of the oscillator in some detail and shown that the squeezing remains non - negligible in the presence of noise and damping .
clearly the squeezing field itself can be employed in a time - dependent fashion although we have not investigated such a scenario .
it was also shown that the oscillator can be displaced by a second field to which it is coupled dissipatively .
the dynamics of the displacement can be influenced by the squeezing field , although the converse is not true .
it would be interesting to explore the effects of the fully quantum mechanical hamiltonian [ eq . ( [ eq : hq ] ) ] without making the semiclassical approximation of eq .
( [ eq : semic ] ) .
this may lead to highly non - classical states of the mirror - field system as found in the case of purely dissipative coupling @xcite .
we are currently also working on generalizing the present proposal to the case of multiple mirrors in the same cavity .
this work is supported in part by the us office of naval research , by the national science foundation and by the us army research office .
we thank h. uys and o. dutta for useful discussions .
| we consider a small partially reflecting vibrating mirror coupled dispersively to a single optical mode of a high finesse cavity .
we show this arrangement can be used to implement quantum squeezing of the mechanically oscillating mirror . |
we performed density functional theory ( dft ) calculations considering various full potential all electron codes , such as wien2k @xcite , elk @xcite , and fplo @xcite using the generalized gradient approximation functional in its pbe form @xcite , and verified that the results agree reasonably well among different codes .
such comparison is particularly important because the codes implement the spin - orbit coupling in slightly different ways , employing usually unimportant , but in principle unequal approximations . in the _
non - relativistic _ calculations the core electrons were treated fully relativistically and the valence electrons non - relativistically ( scalar relativistic approximation ) . in
the _ fully relativistic _
calculations , _ i.e. _ with inclusion of spin - orbit coupling , all electrons were treated fully relativistically .
we considered the c@xmath114 crystal structure as given in ref . and shown in fig .
[ structure ] .
we used projective wannier functions as implemented in the fplo basis @xcite to determine a tight - binding ( tb ) representation for the ir @xmath25 bands . in figure
[ fig : wannierbs ] we show the dft band structure together with the bands corresponding to the wannier representation and the tb bands derived from this representation . in fig .
[ fig : wannier ] we present the projective wannier functions for the @xmath25 orbitals of one ir site .
the wannier functions exhibit the typical shape of the @xmath25 functions at the ir site .
besides , they show a clear asymmetry due to na as well as tails on the o sites . in order to analyze the contribution to the non - relativistic band structure of the various tight - binding hopping parameters and its relation to the quasi - molecular orbital ( qmo ) picture , we present in fig . [ fig : restrictedbs ] the band structure that results if we restrict the tight - binding hamiltonian to first neighbors ( top left ) , up to second nearest neighbors ( top right ) , up to third nearest neighbors ( bottom left ) , and without restriction ( bottom right ) .
one can see that already the second neighbors model provides a good semiquantitative description of the band formation . in the next figure [ fig : qmobs ]
we show the tight - binding band structures within the qmo model . in these calculations
we have included the on - site trigonal splitting ( the top left panel ) , adding the nearest neighbors @xmath115 hopping ( top right ) , then the second nearest neighbors @xmath116 hopping ( bottom left ) and , finally , including also the third nearest neighbors hopping between the like orbital , which also proceeds through na and does not take an electron out of the corresponding qmo ( bottom right ) .
the small dispersion that arises for nearest neighbors is due to deviations from the perfect octahedral environment of iridium . upon inclusion of second nearest neighbors ,
as mentioned in the main text , the upper doublet and singlet merge to form one three - band manifold . in figure
[ projection ] , we show projections of the total density of states of na@xmath0iro@xmath1 onto the quasi - molecular orbitals specified in table [ 1 ] of the main text.the eigenvector matrix @xmath117 ( with @xmath118 ) is a unitary transformation that rotates the atomic ir @xmath2 orbitals into the qmo orbital space . @xmath66 and @xmath62 states are perfectly degenerate in the nonrelativistic case ( figure [ projection ] ( a ) ) .
when spin - orbit coupling is turned on ( figure [ projection ] ( b ) ) , interestingly , the three upper bands are no more equivalent in this sense , with the central band being mostly @xmath59 , and the other two mostly @xmath62 .
importantly , there is hardly any mixing between the lower three bands and the upper three bands , emphasizing the fact that the low - energy physics is nearly exclusively defined by the upper three qmos , and their mutual interaction , whether with or without spin - orbit . at the same time
, one can , alternatively , project the same bands onto the relativistic orbitals , @xmath119 and @xmath120 , and , as observed before@xcite , the upper two bands have more @xmath119 character than @xmath120 character , but , for instance , at the @xmath121 point , only slightly so ( more at some other points ) .
thus , even though the so effects are considerable , they are not strong enough to reduce the problem to a two @xmath119 model .
the ferromagnetic state shows in the absence of so an energy gain of nearly 80 mev per ir with respect to the non - magnetic solution and about half this value against competing antiferromagnetic states ( zigzag and stripy phases ) ; the simple nel state is much higher in energy .
inclusion of so changes the energetics considerably , as described in the main text , with the zigzag antiferromagnetic ordering becoming competitive with the ferromagnetic one , and lower in energy than the stripy phase .
we deliberately do not discuss the calculated energies in detail , because the energy differences involved are on the order of one mev per atom , which is beyond the accuracy of the density functional theory itself , and on the border of the technical accuracy of existing band structure codes . in fig .
[ doss ] we show the density of states for some magnetic orderings considered in our fully relativistic calculations .
note that the zigzag ordering preserves the nonmagnetic pseudogap at the fermi level , while the stripy ordering destroys it .
finally some considerations about the hubbard @xmath21 are at place . in fact , there are two ways of defining @xmath21 in this case . as usually , the actual value of @xmath21 depends on which orbitals it is being applied to . for instance
, it is well known that in fe pnictides the appropriate value of @xmath21 acting on the wannier functions combining fe @xmath14 and as @xmath46 states is more than twice smaller that that acting on actual atomic @xmath14 orbitals since the screening effects change depending on the basis of active states considered . in molecular solids , such as fullerides , the atomic value of @xmath21 often appears completely irrelevant , and the physically meaningful value of @xmath21 is the ( much smaller ) energy of coulomb repulsion of two electrons placed on two molecular orbitals . in the case of na@xmath0iro@xmath1 one has a choice of using an atomic @xmath122 1.5 - 2 ev , realizing that the results will be strongly affected by the fact that electrons are localized not on individual ions , but on individual qmos , or of constructing @xmath21 in the qmo basis . the former way is readily available in such formalisms as lda+u but it may be a poor choice for the description of a system based on quasi - molecular orbitals .
p. blaha , k. schwarz , g. k. h. madsen , d. kvasnicka , and j. luitz 2001 wien2k , _ an augmented planewave+localorbitals program for calculating crystal properties _
( karlheinz schwarz , techn .
universitt wien , austria ) .
s. k. choi , r. coldea , a. n. kolmogorov , t. lancaster , i. i. mazin , s. j. blundell , p. g. radaelli , yogesh singh , p. gegenwart , k. r. choi , s .- w .
cheong , p. j. baker , c. stock , j. taylor , phys .
lett . * 108 * , 127204 ( 2012 ) . | contrary to previous studies that classify na@xmath0iro@xmath1 as a realization of the heisenberg - kitaev model with dominant spin - orbit coupling , we show that this system represents a highly unusual case in which the electronic structure is dominated by the formation of quasi - molecular orbitals ( qmos ) , with substantial quenching of the orbital moments .
the qmos consist of six atomic orbitals on an ir hexagon , but each ir atom belongs to three different qmos .
the concept of such qmos in solids invokes very different physics compared to the models considered previously . employing density functional theory calculations and model considerations
we find that both the insulating behavior and the experimentally observed zigzag antiferromagnetism in na@xmath0iro@xmath1 naturally follow from the qmo model .
high interest in the recently synthesized hexagonal iridates @xcite is due to the hypothesis @xcite that the electronic structure in these materials is dominated by the spin - orbit ( so ) interaction . in this case , the ir @xmath2 bands are most naturally described by relativistic atomic orbitals with the effective angular moment , @xmath3 and @xmath4 in this approximation , the splitting between the 3/2 and 1/2 states is larger than their dispersion .
the upper band @xmath5 is half filled and ir atoms can be described as localized ( @xmath6 @xmath7 @xmath8 ) magnetic moments @xcite with the exchange interaction strongly affected by so coupling . in particular , this picture leads to a very appealing framework known as heisenberg - kitaev model @xcite , with highly nontrivial physical properties .
however , experimental evidence for the @xmath9 scenario is lacking @xcite . in this letter , based on _ ab initio _ density functional theory ( dft ) calculations and model considerations , we show that this picture does not apply to the actual na@xmath0iro@xmath1 .
instead , this system represents a highly unusual case where the formation of electronic structure is dominated by quasi - molecular orbitals ( qmos ) , which involve six ir atoms arranged in a hexagon .
what distinguishes this picture from molecular solids is that there is no associated spatial clusterization , but each ir atom ( via its three @xmath2 orbitals ) participates in three different qmos , yet in the first approximation there is no inter - qmo hopping the thus formed bands are dispersionless .
such an electronic structure calls for a new approach .
there is no known recipe for handling its magnetic properties , or adding coulomb correlations , for instance .
while we will not present a complete theory of spin dynamics and correlations in the qmo framework , we will outline the general directions and most important questions , in the expectation that this will stimulate more theoretical and experimental work and eventually generate more insight . yet , the key observable features of na@xmath0iro@xmath1 : small magnetic moment , unusual zigzag antiferromagnetism , and mott - enhanced insulating behavior , are naturally consistent with the qmo framework .
the main crystallographic element of na@xmath0iro@xmath1 ( see si ) is an ir@xmath10 ( 5@xmath11 ) honeycomb layer with a na@xmath12 ion located at its center .
each ir is surrounded by an o octahedron , squeezed along the cubic [ 111 ] ( hexagonal @xmath13 ) axis .
therefore , ir @xmath14-states are split into an upper @xmath15 doublet and a lower @xmath2 triplet .
the [ 111 ] squeezing further splits the @xmath2 levels into a doublet and singlet ; initially this effect was neglected @xcite , however , it was later included @xcite ( and overestimated ) to explain the observed deviations from the heisenberg - kitaev model . in the previous works , after identifying the @xmath16 splitting it was assumed that the energy scales are @xmath17 @xmath18 ( @xmath19 , @xmath20 ) @xmath18 @xmath21 , where @xmath22 is the @xmath14-electron band width , @xmath23 the effective hopping parameter , @xmath19 the hund s rule coupling , @xmath20 the so parameter , and @xmath21 the on - site coulomb repulsion . in this limit , the electrons are localized and the system is a mott insulator . while @xmath20 @xmath24 0.4 - 0.5 ev for @xmath25 ions , the bandwidth for 5@xmath14 orbitals is 1.5 - 2 ev and @xmath26 ev , @xmath27 ev , reduced compared to typical @xmath28 ev and @xmath29 ev for @xmath30 electrons .
many - body renormalization may narrow the bands by a factor @xmath31 however , given that in ir @xmath32 , it is unrealistic to expect a large renormalization . therefore , the usual starting point @xmath33 is not valid here , rather , the system is close to an itinerant regime .
@xmath34 , one can not justify reducing the description of na@xmath0iro@xmath1 ( and possibly other iridates ) to an effective @xmath35 model , decoupled from the other @xmath36 states .
thus , the first step ( usually skipped ) is to understand the non - relativistic band structure .
we have therefore performed dft calculations ( see si ) initially without so effects ( see fig .
[ bandso ] , solid purple lines ) . inverting the band structure results ( see si )
, we obtained the corresponding tight - binding hamiltonian .
the leading channel ( by far ) is the nearest neighbor ( nn ) o - assisted hopping between unlike orbitals ( see fig .
[ fig : structure2 ] ) .
this was also correctly identified previously @xcite .
there are three different types of nn ir - ir bonds ; for one ( we name it @xmath37 bond ) ( see fig .
3 ) this hopping is only allowed between @xmath38 and @xmath39 orbitals , for the next ( @xmath40 ) between @xmath39 and @xmath41 orbitals and for the third bond ( @xmath42 ) between @xmath41 and @xmath38 . in our calculations
this hopping , @xmath43 ( the prime indicates that the hopping is @xmath44 o ) is about 270 mev .
perturbatively , this term is proportional to @xmath45 where @xmath46 stands for the o @xmath46 states . ref .
@xcite pointed out another ( next nearest neighbors , nnn ) o - assisted term , which we find to be @xmath47 mev .
jackeli and khalliulin @xcite invoked another nn hopping process , between like orbitals pointing directly to each other . despite the short ir - ir distance ,
these matrix elements are surprisingly small , @xmath48 mev .
finally , some authors @xcite addressed the trigonal squeeze , which creates non - zero matrix elements between the same - site @xmath2 orbitals .
the main feature of the calculated non - relativistic band structure ( see fig .
[ bandso ] ) is formation of a singly degenerate ( not counting spins ) band state at @xmath49 ev , a doubly degenerate one at @xmath50 ev , and a three - band manifold between @xmath51 and @xmath52 ev . this clear separation , of the order of 0.3 ev , can not be related to the trigonal squeeze , as this can only split the 6 @xmath2 bands ( there are two ir per cell ) into a doublet and quartet . in order to understand this , we start with the dominant hopping , the nn o - assisted @xmath43 .
let us consider an electron on a given ir site in a particular orbital state , say , @xmath53 the site has three nn neighbors .
as discussed above , this electron can hop , with the amplitude @xmath54 to a neighboring state of @xmath39 symmetry , located at a particular nn site . from there
, it can hop further into a @xmath41 state on the next site , and so on ( see figs .
[ fig : structure2 ] and [ qmo ] ) . at each site ,
the electron has only one bond along which it can hop .
following the electron around , we see that after six hops it returns to the same state and site from where it started . _
this means that in the nn _
@xmath43 _ approximation every electron is fully localized within 6 sites forming a hexagon . _ such a state could be called a molecular orbital , except that there are no spatially separated molecules on which electrons are localized .
each ir belongs to three hexagons , and each ir - ir bond to two .
thus , three different @xmath2 orbitals on each ir site belong to three different quasi - molecular orbitals ( qmo ) and these qmos are fully localized in this approximation ( fig .
[ qmo ] ) .
six qmos localized on a particular hexagon form six levels , listed in table [ table ] , grouped into the lowest b@xmath55 singlet , the highest a@xmath56 singlet , and two doublets e@xmath56 and e@xmath57 .
the energy separation between the lowest and the highest level is @xmath58 which is close to the calculated total non - relativistic @xmath2 band width .
[ c]l|l|lsymmetry & eigenenergy & eigenvector(s ) + @xmath59 & @xmath60 & @xmath61 + @xmath62 & @xmath63 & @xmath64 + ( twofold ) & & @xmath65 + @xmath66 & @xmath67 & @xmath68 + ( twofold ) & & @xmath69 + @xmath70 & @xmath71 & @xmath72 we now add the o - assisted nnn hopping @xmath73 .
here there are several such paths .
however , the dominant hopping takes advantage of the diffuse na @xmath74 orbital ( see fig . [
fig : structure2 ] ) , and is proportional to @xmath75 .
it connects unlike nnn @xmath2 orbitals that belong to the same qmo , and therefore retains the complete localization of individual qmos .
it does shift the energy levels though , as shown in table [ table ] .
the upper singlet and doublet get closer and the lower bands move apart providing the average energy separations of @xmath76 , @xmath77 , and @xmath78 ev among the calculated non - relativistic subbands ( at @xmath79 the upper two levels merge ; in reality , @xmath80 . given that the subband widths are 0.20.3 ev , obviously , the upper doublet and singlet merge to form one three - band manifold .
several effects contribute to the residual dispersion of the qmo subbands .
the trigonal splitting plays a role , albeit smaller than often assumed : the trigonal hybridization is @xmath81 mev ( the splitting being @xmath82 ) .
this may seem surprising , given the large distortion of the o octahedra , however , in triangular layers several factors of different signs contribute to @xmath83 and strong cancellations are not uncommon@xcite .
trigonal splitting , combined with various nn and nnn hoppings not accounted for above , all of them on the order of 20 mev , trigger subband dispersions of 200300 mev ( see si for further discussion ) .
we shall now address the so interaction .
the corresponding bands and density of states ( dos ) are shown in fig .
[ bandso ] .
the lowest two subbands hardly exhibit any so effect , even though the spin - orbit parameter @xmath20 in ir is @xmath24 0.4 - 0.5 ev , larger than both the subband widths and subband separation .
however , a simple calculation shows that not only are the orbital momentum matrix elements between the qmos on the same hexagon zero ( this follows from the quenching of the orbital momentum in the qmo states ) , but they also vanish between the like qmos , located at the neighboring hexagons , such as @xmath84 furthermore , at @xmath85 the matrix elements between the two lowest subbands , @xmath70 and @xmath86 vanish because of different parities ; away from the @xmath85 point the effect of so increases , in the first approximation , as @xmath87 , where * a * ,
* b * and * c * are the three vectors connecting the centers of the hexagons , as can be worked out by applying the @xmath88 operator to the corresponding qmos . the situation becomes more complex in the upper manifold , where three bands , @xmath59 and two @xmath62 , come very close .
even though the diagonal matrix elements , as well as nondiagonal elements at @xmath85 still vanish , the fact that @xmath59 and @xmath62 are nearly degenerate in energy induces a considerable so effect at all other * k*-points ( which grows linearly with @xmath89 as @xmath90 note that deviations from the minimal model ( @xmath91 ) and so coupling with the lower @xmath66 states also affect the bands at @xmath92 .
we also remind that the orbital moment of the individual electronic states can only be finite if the qmos mix ( which is the case ) , and the direction of the orbital moment is different in different parts of the brillouin zone : along one of the three cardinal in - plane directions it is parallel to the cubic @xmath93 , along another to @xmath94 , @xmath95 .
since the spin moment tends to be parallel to the orbital moment , so is competing with the hund s rule coupling and suppresses the tendency to magnetism .
let us now discuss the effect of the hubbard correlations .
it was initially conjectured that na@xmath0iro@xmath1 was a mott insulator .
this seems counterintuitive , since similar @xmath96 ru and rh compounds are correlated metals , and more diffuse 5@xmath14 orbitals have a smaller hubbard @xmath97 ev and stronger hybridization .
it is hard to justify that this @xmath21 can drive a 5/6 filled band of a similar width into an insulating state .
recently another , more logical concept has gained currency : on the lda level na@xmath0iro@xmath1 is a semimetal , barely missing being a semiconductor , and a small hubbard @xmath21 just helps to enhance the already ( spin - orbit driven ) existing gap .
indeed , in our calculations the minimal gap is @xmath988 mev , but the average direct gap is 150 mev , consistent with the optical absorption@xcite .
the minimal direct ( optical ) gap is 50 mev , so it is plausible that it is somewhat enhanced by correlation effects . in order to include the effect of an onsite hubbard @xmath21 in the qmo basis
, a @xmath99 has to be applied to each qmo@xcite , with a residual coulomb repulsion between neighboring qmos , @xmath100 ( note that two qmos overlap on two sites ) .
overall , we expect that the effect of the coulomb repulsion in our system is similar to that in a single - site two - orbital hubbard model at half filling ( the upper qmo band is half - filled ) and @xmath101 mev . in this case , since @xmath102 does not compete with one - electron hopping any more , one should expect that the gap will be enhanced by a considerable fraction of @xmath102 , which is consistent with the experiment .
thus , hubbard correlations are of no qualitative importance , and only moderately enhance the existing gap .
since all electrons are fully delocalized over six sites , any model assuming localized spins ( whether heisenberg or kitaev ) is difficult to justify .
on the other hand , the qmos are not magnetically rigid objects and neighboring qmos overlap on 2 out of 6 sites , which makes a model with magnetic moments localized on qmos equally unsuitable @xcite .
we will consider therefore magnetism in the itinerant approach . in the non - relativistic case
, the non - magnetic dos shows a high peak at @xmath103 due to @xmath62 and @xmath59 merging and rather flat band dispersion ( see fig . [ bandso ] ) .
such a system is unstable against ferromagnetism ( fm ) and the peak is easily split gaining exchange energy ( 1 @xmath104ir ) with little loss of kinetic energy .
the resulting fm state is half - metallic ( fig .
[ dos ] ) ( see si ) . turning on the so interaction has a drastic effect on magnetism .
so competes with the hund s rule that favors all onsite orbitals to be collinear .
the spin moment is then reduced from @xmath105 @xmath8 to @xmath106 @xmath8/ir for ferro- , and @xmath107 @xmath108/ir for the zigzag and stripe antiferromagnetic ( afm ) arrangements ( see si ) .
the orbital moment is parallel to the spin one , reminiscent of the @xmath5 state , and is roughly equal in magnitude and not twice larger , as it should be for @xmath5 .
the energy gain for the fm case drops to a few mev / ir @xcite , and the zigzag pattern evolves as the most favorable afm state .
qualitatively , two closely competing ground states emerge from the relativistic dft calculations : ferromagnetic and zigzag . in the context of an itinerant picture
, we can argue as follows .
so creates a pseudogap at the fermi level in the non - magnetic calculations ( see fig .
[ bandso ] ) .
this gains one - electron energy and any afm arrangement that destroys this pseudogap incurs a penalty . from the three
considered afm states only zigzag preserves ( even slightly enhances ) the pseudogap ( see si ) . that gives this state an immediate energetical advantage and leads to the energy balance described above .
two notes are in place : first , all the above holds in lda+u calculations with a reasonable atomic @xmath21 ( we have checked @xmath21 up to 3.8 ev ) . the role of @xmath21 in these systems - as stated previously- is merely enhancing the existing so - driven gap .
second , if the dos indeed plays a decisive role in magnetic interactions , it is unlikely that they can be meaningfully mapped onto a short - range exchange model , heisenberg or otherwise . summarizing ,
our dft calculations demonstrate that na@xmath0iro@xmath1 is close to an itinerant regime .
the electronic structure of this system is naturally described on the basis of quasi - molecular orbitals centered each on its own hexagon .
this makes this , and similar materials rather unique .
proceeding from this description one can understand the main properties of na@xmath0iro@xmath1 , including its unique zigzag magnetic ordering with small magnetic moment . however , the main goal of our work is not a complete understanding of the magnetic properties of na@xmath0iro@xmath1 .
we realize that this understanding is still incomplete and that full explanation of the weak antiferromagnetism , as well as of the magnetic response in this compound remains a challenge .
rather , we lay out the framework in which this challenge has to be resolved .
we demonstrate that both the simplified ( but correct ) tb model proposed in previous studies @xcite , and full _ ab initio _ calculations provide a framework that is best described by the quasi - molecular orbitals .
this is an as yet unexplored concept ( as opposed to molecular orbitals or atomic orbitals ) , and there are many open questions about how to treat correlations , magnetic response _ etc .
_ in this framework , however , it appears to be the only way to reduce the full 12 atomic orbitals ( @xmath2 or their relativitsic combinations ) problem to a smaller subspace ( 3@xmath109 ) qmos .
i.i.m . acknowledges many stimulating discussions with radu coldea and his group , and with alexey kolmogorov , and is particularly thankful to radu coldea for introducing him to the world of quasihexagonal iridates .
h.o.j .
, r.v . and d.kh .
acknowledge support by the deutsche forschungsgemeinschaft through grants sfb / tr 49 and for 1346 ( h.o.j . and r.v . ) and sfb 608 and for 1346 ( d.kh . ) .
h.o.j .
acknowledges support by the helmholtz association via ha216/emmi
. 99 i. felner and i.m .
bradaric , physica b * 311 * , 195 ( 2002 ) .
h. kobayashi , m. tabuchi , m. shikano , h. kageyama and r. kanno , j. mater .
chem .
* 13 * , 957 ( 2003 ) .
y. singh and p. gegenwart , phys .
rev .
b * 82 * , 064412 ( 2010 ) .
g. jackeli , and g. khaliullin , phys .
rev .
lett .
* 102 * , 017205 ( 2009 ) .
a. shitade , h. katsura , j. kune , x .-
l .
qi , s .- c .
zhang , and n. nagaosa , phys .
rev .
lett . * 102 * , 256403 ( 2009 ) . note that in this case the spin magnetic moment is 1/3 @xmath8 and the orbital moment is 2/3 @xmath110 j. chaloupka , g. jackeli , and g. khaliullin , phys
. rev .
lett . * 105 * , 027204 ( 2010 ) .
j. reuther , r. thomale , and s. trebst , phys .
rev .
b * 84 * , 100406(r ) ( 2011 ) .
while in recent xas measurements ( j.p .
clancy _
et al . _
arxiv:1205.6540 ) a non - zero branching ratio was reported in na@xmath0iro@xmath1 , this observation only indicates that there is a substantial correlation between the spin and the orbital moments direction , @xmath111 , but this fact _ per se _ does not tell us that relativistic atomic orbitals @xmath112 form a good basis for describing the electronic structure . as we show in the letter , because of an accidental degeneracy of the three top _ molecular _ orbitals , the effect of so is substantial , but are not necessarily described in terms of a particular @xmath113 . c. h. kim , h.s .
kim , h. jeong , h. jin , and j. yu , phys .
rev .
lett . * 108 * , 106401 ( 2012 ) .
s. bhattacharjee , s .- s .
lee and y. b. kim , new j. phys .
* 14 , * 073015 ( 2012 ) .
s. k. choi , r. coldea , a. n. kolmogorov , t. lancaster , i. i. mazin , s. j. blundell , p. g. radaelli , y. singh , p. gegenwart , k. r. choi , s .- w .
cheong , p. j. baker , c. stock , and j. taylor , phys .
rev .
lett . * 108 * , 127204 ( 2012 ) .
d. pillay , m. d. johannes , i. i. mazin , o. k. andersen , phys .
rev .
b * 78 * , 012501 ( 2008 ) .
r. comin , g. levy , b. ludbrook , z .- h .
zhu , c.n .
veenstra , j.a .
rosen , yogesh singh , p. gegenwart , d. stricker , j.n .
hancock , d. van der marel , i.s .
elfimov , a. damascelli , arxiv:1204.4471 ( unpublished ) .
compare to mott - hubbard transition in fullerides : o. gunnarsson .
_ alkali - doped fullerides : narrow - band solids with unusual properties_. world scientific , 2004 .
spin dynamics can be in principle always mapped onto a localized spin model of a sufficient range , but this can be a dangerous exercise : compare the fe pnictides , where such mapping led to an unphysically drastic temperature dependence of the heisenberg exchange parameters , until it was realized that proper mapping requires a strong biquadratic term ( a.l .
wysocki , k. d. belashchenko , and v.p .
antropov , nature physics * 7 * , 485 ( 2011 ) ) .
incidentally , the ferromagnetic state in relativistic calculations acquires substantial anisotropy . depending on the polarization direction , the energy gain over the nonmagnetic state varies by a factor of two .
this is consistent with the experimentally observed substantial anisotropy of the uniform susceptibility in the paramagnetic state @xcite . |
let @xmath2 be the canonical imbedding as a chart .
real algebraic sets in @xmath3 are not in general real algebraic sets in @xmath4 .
the zariski closure of the image of an algebraic set ( under @xmath5 ) usually has extra components at infinity .
an algebraic subset of @xmath6 which remains an algebraic set in @xmath4 is called a _ projectively closed algebraic set _ ( @xcite ) .
not every algebraic set is projectively closed . in general , isotoping a submanifold of the projective space @xmath4 to an algebraic subset is a much harder problem than the corresponding problem in the affine case @xmath7 . in this paper
we produce a transcendental submanifold of @xmath4 in the sense of @xcite .
that is , we find a smooth submanifold of @xmath4 which is isotopic to a nonsingular projective algebraic subset , but which can not be isotoped to the real part of any complex nonsingular algebraic subset of @xmath8 .
this results generalizes the affine examples of @xcite to the projective case . we want to thank msri for giving us opportunity to work together
let @xmath9 be a real ( or complex ) algebraic set defined over @xmath10 , and let @xmath11 ( or @xmath12 ) .
then we can define algebraic homology groups @xmath13 to be the subgroup of @xmath14 generated by the compact real ( or complex ) algebraic subsets of @xmath9 ( cf .
we define @xmath15 to be the poincar duals of the groups @xmath13 when defined . by the resolution of singularities theorem ( @xcite ) , @xmath13 is also the subgroup generated by the classes @xmath16)$ ] where @xmath17 is an entire rational function
, @xmath18 is a compact nonsingular real ( or complex ) algebraic set and @xmath19 $ ] is the fundamental class of @xmath18 .
therefore even when @xmath9 is real , we can define @xmath20 to be the subgroup generated by @xmath16)$ ] where @xmath21 is an entire rational function from an oriented compact nonsingular real algebraic set and @xmath19 $ ] is the fundamental class of @xmath18 .
now let @xmath22 be a nonsingular projective real algebraic set of dimension @xmath23 , and suppose its complexification @xmath24 is nonsingular .
let @xmath25 denote the inclusion .
define @xmath26 to be the subgroup of @xmath27 generated by irreducible complex algebraic subsets defined over @xmath10 with @xmath28-dimensional real parts .
in other words it is generated by the complexification of @xmath28-dimensional real algebraic subsets of @xmath9 in @xmath29 .
again by the resolution theorem , @xmath30 is generated by the classes @xmath31)$ ] , where @xmath32 is an irreducible nonsingular complex projective algebraic set defined over @xmath10 with nonempty real part and @xmath33 is a regular map defined over @xmath10 .
let @xmath34 denote the poincar dual of @xmath35 .
define @xmath36 let @xmath37 to be the mod @xmath38 reduction of @xmath39 ( under the obvious coefficient homomorphism @xmath40 ) .
define the natural subgroup of @xmath41 @xmath42 recall that theorem a ( b ) of @xcite relates these groups to each other : for all @xmath28 the following hold : @xmath43 .
let @xmath44 be a closed smooth submanifold of a nonsingular algebraic set @xmath9 .
the problem of whether @xmath45 is isotopic to a nonsingular algebraic subset of @xmath9 is an old one . if we allow stabilization ( replacing @xmath9 by @xmath46 for sufficiently large @xmath28 ) the problem becomes solvable ( @xcite and @xcite for the affine case , @xcite for the projective case , and @xcite for the general case where there are obstructions ) .
if we do nt allow stabilization the problem becomes much harder , in which the complexification of @xmath9 begins to play an important role .
in @xcite it was shown that every closed smooth @xmath47 can be isotoped to a nonsingular ( topological ) component of an algebraic subset of @xmath48 .
more generally in @xcite it was shown that any immersed submanifold of @xmath3 can be isotoped to a nonsingular algebraic subset of @xmath48 if and only if @xmath45 is cobordant through immersions to an algebraic subset of @xmath48 . here
we need a very special case of these theorems , the one which allows us to isotop some submanifolds of @xmath0 ( the examples in the next section ) to projectively closed algebraic subsets .
obviously the following lemmas hold in more general contexts , but to make the examples of the main theorem in the next section transparent , we chose to state them in this special form , which is enough to prove the theorem .
every closed codimension one submanifold of @xmath49 can be approximated by a nonsingular projectively closed algebraic subset .
this for example is proven in @xcite ( also theorem 2.8.2 of @xcite ) , but can also be seen from seifert s original proof @xcite by noting that the highest degree terms of the polynomial he constructs are a constant times @xmath50 ( clearly the zeros of such polynomials are projectively closed algebraic sets ) .
let @xmath51 be closed smooth manifolds with @xmath45 separating @xmath52 .
let @xmath53 be a smooth imbedding .
then @xmath54 is isotopic to a nonsingular projectively closed subvariety of @xmath49 .
in particular , by viewing @xmath54 as a submanifold of @xmath55 via the natural inclusion @xmath56 , @xmath54 is isotopic to a nonsingular projective algebraic subvariety of @xmath55 . by @xcite
we can isotop @xmath57 to a nonsingular topological component @xmath58 of a real algebraic subvariety of @xmath49 .
let @xmath59 be one of the codimension zero components of @xmath60 , and let @xmath61 be the boundary of a small tubular neighborhood of @xmath62 in @xmath49 .
by lemma 2 above we can isotop @xmath63 to a projectively closed nonsingular algebraic subvariety @xmath64 of @xmath49 . then @xmath65 gives the desired variety .
@xmath9 is projectively closed since @xmath64 is projectively closed .
in @xcite we constructed smooth submanifolds @xmath47 which are isotopic to nonsingular algebraic subsets of @xmath66 , but not isotopic to the real parts of nonsingular complex algebraic subsets of @xmath1 .
so , this means that either every algebraic model of @xmath45 in @xmath67 develops extra components at infinity when its zariski closure is taken in @xmath68 ( i.e. it is not projectively closed ) , or @xmath45 admits nonsingular algebraic models in @xmath69 but the complexifications all such models necessarily contain singular points in @xmath70 .
the following theorem eliminates the first possibility , hence it gives genuine topological obstructions to moving smooth submanifolds of @xmath4 to nonsingular algebraic sets in the strong sense ( i.e. no singularities in complexification ) .
there are closed smooth submanifolds @xmath71 which can be approximated ( via a small isotopy ) by nonsingular subvarieties of @xmath72 , but they can not be isotoped to the real parts of nonsingular complex algebraic subvarieties of @xmath8 defined over @xmath10 . by @xcite for any @xmath73 there is an @xmath74 such that we have an imbedding @xmath75 .
let @xmath76 , where @xmath77 .
let @xmath78 in @xmath69 , so by lemma 2 above @xmath45 can be isotoped to a nonsingular projectively closed algebraic subset @xmath79 of @xmath69 where @xmath80 .
we claim that when @xmath81 when @xmath74 is even , and @xmath82 when @xmath83 , @xmath9 can not be the real part of a nonsingular complex algebraic subset @xmath84 of @xmath1 ( defined over @xmath10 ) .
suppose such a @xmath84 exists . by a version of the lefschetz hyperplane theorem due to larsen @xcite , for @xmath85
, the restriction induces an isomorphism @xmath86 in particular when @xmath87 , the group @xmath88 lies in the the image of @xmath89 under restriction .
but since @xmath9 lies in a chart @xmath66 of @xmath0 , from the diagram we conclude that @xmath91 for @xmath92 . on the other hand by @xcite
the stiefel - whitney classes of @xmath9 are represented by real algebraic subsets since @xmath9 is a nonsingular real algebraic set .
hence , when @xmath74 is even @xmath93 is algebraic , also when @xmath83 then @xmath94 is algebraic , where @xmath95 is the generator of @xmath96 .
so when @xmath74 is even and @xmath81 , or when @xmath83 and @xmath97 we get a contradiction to theorem 1 above , for example when @xmath74 is even @xmath98 we should point out that the above mentioned theorem of @xcite actually implies that there are imbeddings @xmath75 for the pairs @xmath99 with @xmath74 even and @xmath81 , or @xmath83 and @xmath82 .
this is what is used in the proof .
recall that on any nonsingular real algebraic variety structure @xmath9 of a smooth manifold @xmath45 , the stiefel - whitney classes , and mod @xmath38 reductions of pontryagin classes of @xmath45 are algebraic in @xmath9 ( @xcite , @xcite ) .
so the examples in the above theorem generalize many directions .
for example we can take @xmath100 to be any closed smooth manifold which admits a separating imbedding into a closed manifold @xmath101 , such that @xmath102 with @xmath103 , and @xmath104 such that @xmath105 and @xmath95 lies in the subring of the cohomology group generated by stiefel - whitney and pontryagin classes . | in this paper we give examples of closed smooth submanifolds of @xmath0 which are isotopic to nonsingular projective subvarieties of @xmath0 but they can not be isotopic to the real parts of nonsingular complex projective subvarieties of @xmath1 . |
the increased imaging and energy resolution capabilities of the new generation x - ray observatories like _ xmm - newton _ and _ chandra _ , provide a better understanding of the x - ray emission from nearby galaxies . a combination of a number of discrete sources , hot interstellar gas , and a probable active galactic nucleus
are thought to produce the observed x - ray emission from such galaxies ( fabbiano , 1989 ) .
large number of discrete sources have already been detected by _ xmm - newton _ and _ chandra _ in the disks of nearby galaxies such as m 31 ( pietsch 2005 ; kong et al .
2002a , 2003b ) , m 33 ( pietsch 2004 ; plucinsky et al . 2008 )
m 81 ( swartz et al . 2003 ) , and m 101 ( pence et al . 2001 ; jenkins et al . 2004 ) .
these sources are typically high and low mass x - ray binaries ( xrbs ) , supernova remnants ( snrs ) , bright super soft sources ( sss ) , and a number of very luminous sources , so - called ultra - luminous x - ray sources ( ulxs ) . for a study of discrete x - ray source population in the nearby spiral galaxies ,
a sample of three of them ( ngc 4395 , ngc 4736 , and ngc 4258 ) were chosen for this work .
all three have a low activity seyfert nucleus . their low foreground absorption ( 1.20 - 1.43)@xmath310@xmath4 @xmath5 ( dickey @xmath6 lockman 1990 ) also made them good targets for exploring such discrete x - ray sources .
ngc 4395 is classified as one of the nearest and least luminous ( l@xmath7 @xmath2 5@xmath310@xmath8 erg s@xmath9 ) type 1 seyfert galaxies ( filippenko @xmath6 sargent 1989 ) .
this dwarf galaxy appears to harbor a central black - hole , likely to be in the @xmath2 10@xmath10 -10@xmath11 m@xmath12 range , significantly lower than more luminous seyferts with black hole masses typically in the range @xmath2 10@xmath13 -10@xmath14 m@xmath12 ( vaughan et al . 2005 ) .
its nucleus has been the subject of several x - ray studies since the _ rosat _ observations ( moran et al . 1999 ; lira et .
al 1999 ) .
they have reported that its nuclear x - ray source displays a large amplitude variability on short time scales and has a soft x - ray spectrum .
also , five x - ray sources , one being significantly brighter than nucleus , were detected within @xmath23@xmath15 of its nucleus .
later , observations by _ chandra _ and _ xmm - newton _ confirmed the variable nature of its active nucleus ( moran et al .
2005 ; vaughan et al . 2005 ) .
our next target , ngc 4736 ( m94 ) , is classified as a low luminosity galaxy with the closest sample of a low ionization nuclear emission - line ( liner ) nucleus ( flippenko @xmath6 sargent 1985 ) .
two bright sources were detected by _
_ observations in its nuclear area ( cui et al .
one of them was consistent with the nucleus of the galaxy and almost one - third of the total observed flux in ( 0.1 - 2 ) kev arose as the emission from the compact source in the center with an extended distribution of hot gas .
12 discrete sources ( including galactic nucleus itself ) were identified by _
rosat hri _ data in its optical disk and ring regions ( roberts et al .
the galaxy was also observed with the _ bepposax _ in ( 0.1 - 100 kev ) band and the _ chandra acis - s _ for a high resolution study of its nuclear region ( pellegrini et al .
they concluded that the liner activity was due to a low luminosity agn at the center .
eracleous et al . (
2002 ) had already examined the high concentration of luminous sources in the nucleus of ngc 4736 by _
they suggested that the galaxy was in a recent starburst phase where the emission was dominated by dense clusters of x - ray binaries . our third target , ngc 4258 ( m106 ) is a bright nearby type 1.9 seyfert galaxy .
it is also known by its anomalous arms discovered by h@xmath16 imaging ( courtes et al . 1993 ) .
water maser line emission of rotating gas near its center indicated a mass of 3.6@xmath310@xmath17m@xmath12 in a disk of radius ranging from 0.12 to 0.25 pc ( miyoshi et al .
_ rosat _ observations of ngc 4258 resolved the anomalous spiral arms forming the boundary of diffuse emission .
15 x - ray point sources associated with the galaxy were identified by _ rosat _ observations ( pietsch et al .
1994 ; cecil et al . 1995 ; vogler @xmath6 pietsch 1999 ) .
they could not detect the seyfert nucleus of the galaxy as a point source .
the first detection of active nucleus was reported by _
asca _ observations .
it was modeled by an absorbed power - law spectrum of photon index @xmath18 @xmath21.8 .
the unabsorbed ( 2 - 10 ) kev luminosity of the source was _
l@xmath19_= 4@xmath310@xmath8 ergs@xmath9 obscured by a column density of n@xmath20 @xmath2 1.5@xmath310@xmath21 @xmath5 .
an fe k@xmath16 emission line was detected with an equivalent width [email protected] kev ( makishima et al . 1994 ) .
ngc 4258 was also observed by _ xmm - newton _ , which showed a hard nuclear point source that could be modeled by a highly absorbed power - law spectrum with no narrow fe k@xmath16 emission ( pietsch @xmath6 read 2002 ) .
a series of observations were also made by _
chandra _ to study its various properties ( young @xmath6 wilson 2004 ) .
they examined the spectrum of the low - luminosity active galactic nucleus , which was well described by a hard x - ray power - law and a variable luminosity with a constant thermal soft x - ray component .
they also discussed the first detection of extragalactic fe absorption line at 6.9 kev , with a very inclined accretion disk varying on a timescale of 6 ks .
yang et al . (
2007 ) studied _ chandra _ and _ xmm - newton _ spectra of its anomalous arms and active jets .
they confirmed that the x - ray emission from these arms has a thermal nature , suggesting that currently active jets could be responsible for heating the gas in the arms . in the present work
, we will investigate the spectral and timing properties of the bright point - like non - nuclear sources in these three galaxies , first as observed by _ xmm - newton _ with a more extended database compared with the previous studies and later , by _
chandra _ to supplement their spectral and timing properties .
general properties of the galaxies are listed in table 1 .
we will restrict ourselves to the off - nucleus luminous sources with sufficient signal to noise ratio ( s / n ) to perform spectral fitting . source locations with respect to the galaxy centers and their cross identifications ( if any ) have been listed in table 2
. the organization of paper will be as follows ; in section 2 , we outline the data reduction and source selection criteria . in section 3 , the spectral fitting and timing analysis techniques are described and , in section 4 , detailed discussion of spectral and temporal results for individual sources in the galaxies are presented .
the overall properties of their x - ray emission are discussed in section 5 .
the data used for the analysis were retrieved from the _ xmm - newton _ public archive .
_ xmm - newton _ observatory ( jansen et al . 2001 ) has three epic cameras at its focal plane : one uses the pn - ccd ( strder et al .
2001 ) and the other two mos - ccd detectors ( turner et al . 2001 ) .
instruments were operated in the full - frame imaging mode and used thin or medium filters for pn and mos .
we analyzed the pipeline - processed data by using the standard software tools of _ xmm - newton science analysis system ( xmm - sas ) v 10.0 _
( gabriel et al .
2004 ) . in the _ sas
_ , x - ray events corresponding to pattern@xmath23 4 ( singles and doubles ) were then selected with flag==0 option for the pn camera , pattern@xmath23 12 were used for the mos cameras .
source detection was performed using the standard maximum likelihood technique as implemented by _
edetect@xmath24chain_. each source inside the _ d@xmath0 _ ellipse was checked to exclude false or doubtful sources , after the automatic selection procedure in _
sas_. the log of observations , including the dates , ids , exposure times , and the type of epic camera for each target , are given in table 3 . in the analysis , we simultaneously used data from the epic pn and the epic mos for each galaxy to determine the spectra of the sources ( with the exception of the mos data for ngc 4395 , which has poor statistics ) .
note also that the epic pn has more source counts due to its better sensitivity when compared with the mos detectors .
figures 1a , 2a , and 3a show the combined exposure - corrected epic pn , mos1 and mos2 true color images of the presently analyzed archival data for the sample galaxies
. inclination corrected _
d@xmath0 _ ellipses are also shown for comparison and identification .
the detected point sources inside a _ d@xmath0 _ ellipse were specified as off - nuclear objects .
we numbered them as xmm - n , n representing source number with decreasing pn count rate .
figures 1b , 2b , and 3b show _ xmm - newton _ point source positions overlaid on a dss1 optical blue image of each galaxy . detected x - ray source parameters in each galaxy
are summarized in tables 4 , 5 , and 6 respectively . in each table
the source i d ( col.1 ) , its position ( col.2 & 3 ) , the likelihood of existence ( where a likelihood value of 10 corresponding to a detection probability above 4@xmath25 ) ( col.4 ) , integrated ( pn + mos1 + mos2 ) count rates with errors ( col.5 ) , flux with the error ( col.6 ) , and luminosity with the error ( col.7 ) are given in the ( 0.2 - 12 ) kev band .
the luminosities were calculated from the flux values using the distances indicated in table 1 . to study the spectral and temporal properties of the non - nuclear point sources , we complemented our _ xmm - newton _ analysis with the _ chandra _ data of each target galaxy . a separate observation log of the _ chandra _ data is given in table 7 for our galaxies .
we retrieved the data from the public archive and analyzed with _
ciao v4.3 _ software package using the caldb v4.4.2 calibration database .
all data were filtered to 0.3 - 10 kev energy range .
the detection of the x - ray point sources was performed using the wavdetect task within the same package .
we have run the task over the 1 , 2 , 4 , 8 and 16-pixel wavelet scales , since these scales are appropriate for detecting point - like sources . after a visual examination for possible false detections
, we have only taken into account the sources with detection significance of 3.5@xmath25 or better and with net counts higher than 6 .
it is well known that , _ chandra _ has significantly better angular resolution than _ xmm - newton_. as a result , the detected _ xmm - newton _ point sources ( one or two ) in the central region of galaxies may have overlapped several _
point sources .
since in the present analyses , we examined only the non - nuclear sources , the sources located in the central regions of 0.3@xmath15 radius were excluded in all three galaxies .
we have started our _ chandra _ data analysis with ngc 4395 and the detected 11 point sources in its _
d@xmath0 _ region .
only one source in the central region of the galaxy has been excluded . from the remaining 10 ,
several have coincided with corresponding _ xmm - newton _ sources .
four of them were examined individually in section 5.1 due to sufficient statistics provide by _ xmm - newton _ observations .
these four _ chandra _ source positions are offset within 2@xmath26 of _ xmm - newton _ source positions center in the galaxy . when the same procedure was carried out for the other two galaxies we obtained the following : for ngc 4736 , 43 point sources in the _
d@xmath0 _ region were detected .
15 sources which were in the central region have been excluded . from the remaining 28 several overlapped by _
xmm - newton _ sources . among are them , three sources coincided with _
xmm - newton _ sources , which were examined individually in section 5.2 .
_ xmm - newton _ and _ chandra _ source positions of these three agree within 2@xmath26 in ngc 4736 . for ngc 4258
, 42 point sources were detected in the _
d@xmath0 _ region .
five point sources in the central region have been removed . among the remaining ,
six sources have coincided with corresponding _ xmm - newton _ sources , and they were examined individually in section 5.3 .
_ xmm - newton _ and _ chandra _ source positions of these six did agree within @xmath2 @xmath271@xmath26 ( except xmm-8 with two _ chandra _ sources within its 15@xmath26 _ xmm - newton _ error circle ) .
to perform a spectral analysis for each luminous source , corresponding data were extracted from a circular region with a radius ( 15@xmath26 - 20@xmath26 ) for pn , mos1 and mos2 depending on the presence of nearby sources .
background events were extracted from a source - free zone normalized to the source extraction area .
the source spectra with the instrument responses and ancillary files were generated using the _ sas v 10.0_. the spectral analyses of the x - ray data have been performed using _
xspec v.12.5_. we grouped the pn and mos spectral energy channels to have at least 20 - 30 counts per bin for a good statistical quality in the spectra .
initially we fitted the spectra with four different one - component models : blackbody ( bbody ) , power - law ( pl ) , diskblackbody ( diskbb ) , and bremsstrahlung ( bremss ) .
in addition , we applied the following two - component models : powerlaw+blackbody ( pl+bbody ) , powerlaw+disk blackbody ( pl+diskbb ) , and powerlaw+mekal ( pl+mekal ) .
we also used a photoelectric absorption model as part of the fit .
the fits were applied to data in the 0.2 - 10 kev range .
the epic pn and mos spectra were fitted simultaneously for each object only in galaxies ngc 4736 and ngc 4258 . for ngc 4395
we used only the epic pn data due to low statistical quality of the mos data .
we used a constant parameter which was set free to account for the normalization differences in the flux calibration of the three epic cameras . to compare the spectral shapes of sources between _ xmm - newton _ and _ chandra _ observations , we fitted the same spectral models to all data .
each data set was chosen based on a longer exposure time to provide sufficient counts for spectral fitting .
standard _ ciao _
routines were used to reprocess event lists ; then source and background spectra were extracted to construct so called the rmf and arf files .
we have fitted the data of 13 _ chandra _ sources in the three galaxies , however , only two of them ( xmm-3 and xmm-6 in ngc 4258 ) were above the threshold ( @xmath28 300 counts ) for spectral analysis .
our results are consistent with the _ xmm - newton _ analysis therefore , we have discussed only the spectral parameters of these two sources in ngc 4258 in section 5.3 .
x - ray variability of sources in all three galaxies were tested by deriving their short - term ( based on the observation with the longest exposure available ) and long - term light curves . for short - term variability
, the data from epic pn and mos cameras were added to improve s / n . after excluding the high background flaring , the corrected exposure times were binned in the range from 200 to 4500 s and each bin had at least 20 counts after background subtraction .
we applied a @xmath29 test to search for large amplitude variations with respect to the constant count rate hypothesis .
short - term variability of point sources for each galaxy in our list can be followed from figures 4 , 5 , and 6 respectively . in table 8
we present the results of @xmath29 tests together with _
p(var ) _ the probability values for those cases exceeding the
95% taken as the limit for variability . results are such that , in ngc 4395 one out of 5 , in ngc 4736 one out of 3 of the sources are variable while in ngc 4258 variable sources increased to 7 out of 8 .
kolmogorov - smirnov
( k - s ) tests were also applied to search for small amplitude variations in the source count rates ; however , nothing significant was found . for long - term variability , in addition to _ xmm - newton _ and _ chandra _ , _
rosat _ data from the literature were also added to the analysis , all in the 0.5 - 2 kev energy range . in this application , to calculate _ xmm - newton _ fluxes , data were fitted with their best fitting models and fluxes calculated after the spectra of the sources determined . for _ chandra _ source fluxes , even though the data for 11 out of 13 sources were below the threshold ( as noted above ) , for comparison purposes , we have tried to fit the data with the same best fitting _ xmm - newton _ model with no constrained on _ chandra _ data fit parameters .
we have succeeded in getting flux values for only two sources in ngc 4395 ( xmm-5 and xmm-6 ) , two sources in ngc 4736 ( xmm-2 and xmm-12 ) , and five sources in ngc 4258 ( xmm-3 , xmm-6 , xmm-16 , xmm-17 , xmm-21 ) within the same energy range ( 0.5 - 2 ) kev . in our timing analysis , we used the longer exposure time if the data were obtained within a year . to derive fluxes for _ rosat _ observations , count rates from available literature
were used .
the absorbed pl continuum slope value was taken from the _ xmm - newton _ observation and the standard _
webpimms _ was used to calculate flux .
resulting long - term light curves for sources in all three galaxies are given in figures 7a , 7b , and 7c .
for the transient ulx source in ngc 4258 , short and long - term light curves are shown in figures 8a , 8b . here , the total number of points of observation for the long time interval ( 18 month to 12 years ) looks rather sparse .
however , there are noticeable indications of long term variability for a number of sources studied .
our variability search for individual sources in each galaxy will be futher discussed in the following sections .
in this section , we will discuss observations of each galaxy individually , highlighting the spectral and timing properties of x - ray luminous point sources in turn .
* * this galaxy was observed in 2002 and 2003 ( see table 2 ) . within the _
d@xmath0 _ region , we have detected 29 point sources as given in table 3 .
our detection limiting flux in the 0.2 - 12 kev energy band is @xmath2 1.2 @xmath310@xmath30 erg @xmath5s@xmath9 for sources inside the optical disk .
the best fitting model spectra for 5 sources were obtained by using only the pn data since the mos data had rather poor statistics . in our further analysis
, we obtained the best - fitting spectral parameters for one - component and two - component models for the 5 sources .
our results are summarized in tables 9 and 10 respectively , together with unabsorbed fluxes and luminosities in the energy range 0.2 - 10 kev .
the best fitting model spectra of these sources are given in figure 9 .
we will elaborate on their individual spectral characteristics below : + * xmm-2 * : this is the brightest object in the galaxy and is about 2.9@xmath15 away from the galaxy center .
it is located nearby a giant h ii region complex .
the distance between the position of the source xmm-2 and the center of the near hii region is no less than @xmath2 500 pc .
this calculation takes into account a deprojected distance using the inclination of the galaxy .
( distribution of h ii regions in this galaxy was studied by cedres & cepa ( 2002 ) in detail ) .
the source spectrum obtained using _ xmm - newton _ data is not adequately fitted with any one - component model .
the best - fitting two - component model is an absorbed pl + diskbb ( @xmath18 = 3.11 , _
t@xmath31 _ = 0.19 kev ) .
but a pl+mekal model gives almost identical fit ( @xmath32= 1.2 ) with similar parameters .
it has an unabsorbed luminosity of
_ l@xmath19 _ = 2.2 @xmath3 th10@xmath33 erg s@xmath34 which is within the standard ulx limit .
both the short - term and long - term light curves do not show evidence of variability . however , its spectral shape and high luminosity lead us to conclude that it is a ulx .
this source was claimed earlier to be a ulx by stobbart et al .
xmm-2 was also defined as a ulx with the best - fitted two - component model , bb+pl ( @xmath18 = 3.44 , _ kt _ = 0.14 kev ) , using 2000 and 2002 archival data by winter , mushotzky @xmath6 reynolds ( 2006 ) ( hereafter wmr2006 ) .
we think that this ulx source needs deeper observations to decide on its spectral and temporal characteristics better . + * xmm-5 * : this is the most notable source in this galaxy .
we derived a source spectrum and fitted this spectrum with a vpshock model ( another _ xspec _ model ) of a non - equilibrium , ionization plasma emission as suitable for snrs ( borkowski et al .
the resulting spectral parameters are _ kt _ = 3.98 kev with an unabsorbed luminosity of _
l@xmath19 _ @xmath2 1.1 @xmath310@xmath35 erg s@xmath9 .
the light curve for xmm-5 shows no variability on both long and short time - scales ; this is expected since it is an snr candidate , the only one in our entire source catalog with a possible extended nature . upon this unique possibility , we also conducted an optical spectroscopic observation to confirm and to clarify its nature . in order to differentiate snrs from typical h ii regions
, we used the well known criterion [ s ii]/h@xmath16 @xmath36 0.4 proposed by mathewson @xmath6 clark ( 1973 ) .
the spectral data were obtained by tfosc ( tug faint object spectrograph and camera ) instrument mounted on rtt150 ( russian - turkish telescope at antalya , turkey ) in april 2009 .
optical spectrum of the source is given in fig .
10 . in table 11 observed line intensities , the e@xmath37 , and h@xmath16 intensity values are listed . as these optical observations reveal , this source is indeed an snr rather than a ulx claimed by swartz et al .
( 2004 ) . in a more recent study of this galaxy , snr nature of xmm-5 was also discussed by leonidaki et al .
( 2010 ) using a higher resolution _
chandra _ archival data . + * xmm-6 and xmm-10 * : the spectra of these sources are best fitted with an absorbed pl model with a photon index of @xmath18 = 1.42 and @xmath18 = 1.86 respectively . quite similar luminosity values of _
l@xmath19 _ = 2.6 @xmath310@xmath35 erg s@xmath9 and @xmath2 1 @xmath310@xmath35 erg s@xmath9 are also observed respectively .
these two sources are at different distances from the central bulge of the galaxy , however we think that xmm-6 has a noticeable contribution from the galactic bulge material , which may explain the differences in spectral shapes and luminosities .
we detected no definite variation in long - term light curve of xmm-6 , but its short - term light curve satisfies our criteria of variability .
xmm-10 , does not show any definite indication of short - term variability although its long - term light curve does show some variability ( a factor of @xmath22 ) .
we can conclude that both sources are xrbs . + * xmm-23 * : this source is located about 2@xmath38 south of the central bulge at a deprojected distance of about 2.4 kpc .
it displays a very soft spectrum that fits well with an absorbed cool bbody ( _ kt _ @xmath2 60 ev , @xmath32 @xmath2 1.3 ) model with an unabsorbed luminosity of _
l@xmath19 _ @xmath2 6@xmath310@xmath39 erg s@xmath9
. its short - term light curve does not show definite variability , but long - term light curve shows variations spanning @xmath2 2 year . with this cool temperature , similar to those found in other galaxies ( swartz et al .
2002 , di stefano et al .
2004 ) , it may be classified as an sss .
they are known as cataclysmic binaries accreting at high rate where hydrogen on the surface of the white dwarf is burned leading to an eddington luminosity ( di stefano @xmath6 kong , 2003 ) . * * we determined 21 point sources within its _
d@xmath0 _ boundaries .
these sources and relevant parameters are given in table 4 .
the detection limiting flux in the 0.2 - 12 kev energy band is @xmath2 2 @xmath310@xmath30 erg @xmath5s@xmath9 for sources inside the optical disk .
we were able to obtain x - ray spectra of 4 of them .
these sources have not been studied in the x - rays before .
the spectral parameters for these sources including unabsorbed flux and luminosities are given in tables 12 - 13 .
the fitted spectra for these sources are plotted in figure 11 . for the individual sources we note the following : + * xmm-2
* : this source is located in the disk of the galaxy . a one - component bremss model fit to data yields
a good fit with a @xmath32@xmath2 1 ; however , the fitted temperature ( _ kt _ = @xmath2 3.1 kev ) is rather high .
a two - component model pl+diskbb on the other hand gives a good fit with the same @xmath32 and lower value of _ kt _ @xmath2 0.75 kev . in the earlier work by
wmr2006 using archival data of _ xmm - newton _
, this source was not detected ; however , we were able to find the source in the next data set about one month later ( see table 3 ) .
its short - term light curve ( in fig .
5 ) shows episodes of significant variability . in the long - term
, it has brightened a factor of @xmath2 25 in the _ xmm - newton _
observations during a @xmath28 6 year period .
it has a high unabsorbed luminosity of _ l@xmath19 _
= 1.95@xmath310@xmath33 erg s@xmath9 revealing that this source can be a transient ulx . + * xmm-12 * : this source is located away from the disk of the galaxy and was detected earlier by _ rosat _ ( roberts et al .
1999 ) . in our analysis its spectrum
is adequately fitted by the one - component diskbb model ( _ t@xmath31 _ = 1.44 kev ) with an unabsorbed luminosity of _ l@xmath19 _
= 0.7@xmath310@xmath35 erg s@xmath9 .
although it does not show any short - term variability , in the long - term a variability within a factor of 3 can not be excluded .
it could be an xrb source .
+ * xmm-18 * : this source was first observed by _ rosat _ ( roberts et al .
it is located @xmath22.1@xmath38 away from the central region of galaxy .
the source spectrum is best fitted by a pl ( @xmath18 = 1.39 ) with an unabsorbed luminosity of l@xmath19 = 0.7 @xmath310@xmath35 erg s@xmath9 . while no significant variability is found in the short - term , in the long - term a variability pattern of up to
a factor of 3 is noticeable .
we suggest that this source is probably an xrb .
+ * * we detected 24 sources inside the _ d@xmath0 _ region of this galaxy ( see table 5 ) .
our detection limiting flux is @xmath2 5 @xmath310@xmath30 erg @xmath5s@xmath9 in the 0.2 - 12 kev energy band for sources inside the optical disk .
more than half of these sources have high luminosity ( _ l@xmath19 _ @xmath28 10@xmath35 ergs@xmath9 ) .
we derived one and two - component best - fitting spectral model parameters for the sources as given in tables 14 - 15 .
the spectra the best fitting models for 8 of these sources are given in figure 12 .
individual source characteristics are given in the following : + * xmm-2 * : we find that this source is a transient ulx .
earlier _ xmm - newton _ data was already interpreted by wmr2006 as a transient .
the source has been detected for the first time in 2002 ( obsid 0059140901 ) and was not detected in other available observations .
the sharp increase in count rates in 2002 data compared with previous ( 2000 , 2001 ) and later ( 2006 ) upper limit leads us to firmly conclude its transient nature .
the source also shows short - term variability .
its spectrum is best described by diskbb ( _ t@xmath31_= 1.43 kev ) with a @xmath32 value of 1.33 and a high unabsorbed luminosity , _
= 1.5 @xmath310@xmath33 erg s@xmath9 .
the spectral fit is not improved with an additional component .
our luminosity estimation for a one - component model is consistent with a ulx .
+ * xmm-3 * : the location of this source is quite far from the disk ( 2.1@xmath38 ) .
this source was first detected by _
( pietsch et al 1994 ) .
it was also identified as a ulx by _ xmm - newton _ ( wmr2006 ) and _ chandra _ ( swartz et al . 2004 ) . according to wmr2006 ,
the source is best fitted with a pl+ bb model ( _ kt _
= 0.78 kev and @xmath18 = 2.02 ) with a @xmath32 value of 1.2 for the data in 2002 .
this source was identified by _
chandra _ as ( cxou j121857.8 + 471607 ) with an absorbed pl model ( @xmath18 = 2 ) . in our
_ xmm - newton _ analysis , the source spectrum is best fitted with a one component absorbed diskbb ( t@xmath31= 1.28 kev ) with a rather high unabsorbed luminosity of _
l@xmath19 _ = 1.9 @xmath310@xmath33 erg s@xmath9 .
our analysis of the _ chandra _ data for this source also show that its spectrum is best fitted with a hot diskbb model , ( t@xmath31= 1.31 kev ) with a higher unabsorbed luminosity of _
l@xmath19 _ = 2.8 @xmath310@xmath33 erg s@xmath9 .
its short term light curve shows a significant variation , also its long - term light curve shows variation with a factor of @xmath2 4 .
the high source luminosity , leads us to conclude that it is a ulx .
we also note that the present analysis is based on a longer observation time with much better statistics than all earlier results . + * xmm-6 * : this source is located 2.4@xmath38 away from the galactic center .
its spectrum is well fitted with an absorbed pl ( @xmath18 = 1.87 ) model .
it has an unabsorbed luminosity of _ l@xmath19
_ = 1.5@xmath310@xmath33 erg s@xmath9 . a single component model fitted to the _ chandra _ data gives almost the same parameter ( @xmath18 = 1.87 ) , and a similarly high luminosity @xmath2 1@xmath310@xmath33 erg s@xmath9 helping to identify the source as an ulx ( shwartz et al . 2004 ) .
our _ chandra _ data analysis for this source also gives the same best fitted model ( pl ) and parameter with almost the same high luminosity value .
long term light curve of xmm-6 shows a factor of @xmath2 3 variability between the intervals , and there is also strong evidence for short - term variability .
this raises the possibility that it could be an accreting xrb in the galaxy .
interestingly , however , the same source also coincides with a blue stellar counterpart from dss2 image making it a likely background agn candidate .
true nature of this source requires further study .
+ * xmm-8 * : this source is best described by a two - component model .
the spectral fit is improved over that of a simple pl ( @xmath18 = 2.1 ) fit by addition of a mekal component ( @xmath18 = 1.95 and _ kt
_ = 0.57 kev ) .
it has a high unabsorbed luminosity of _
l@xmath19 _ = 1.2 @xmath310@xmath33 erg s@xmath9 . to reveal its nature
we also analyzed the _
chandra _ archival data .
two different sources can be resolved within the same position by the better angular resolution of _
chandra_. however , due to the limited amount of data , it was not possible for us to carry out a satisfactory spectral analysis for these two separate sources .
in the same neighborhood , two radio snr candidates were identified by hyman et al .
( 2001 ) ( sources 10 and 12 , table 1 ) but none coinciding with our chandra sources ( angular separations @xmath2 6@xmath26 and @xmath2 16@xmath26 respectively ) . both long - term and short - term light curves of data from the source region do show evidences of sometimes erratic variability .
the spectral shape discussed above leads us to tentatively conclude that , the source region could harbor an xrb . however ,
further long duration observations with better angular resolution are surely needed to clarify the nature of the source(s ) . + * xmm-10 * : this source is located far away ( 6.4@xmath38 ) from the central body of the galaxy .
its spectrum is best fitted with an absorbed pl model ( @xmath18 = 2.8 , @xmath32= 0.9 ) .
the unabsorbed luminosity in this observation is _
l@xmath19 _ = 9.2@xmath310@xmath35 erg s@xmath9 .
its short - term light curve shows no strong indication of variability .
however , its long - term light curve do show a variability ( an uninterrupted increase ) by a factor of at least 5 over the time interval of a decade .
this source could be an xrb .
+ * xmm-16 * : this source gives a best spectral fit with a pl model ( @xmath18 = 2.20 ) .
the unabsorbed x - ray luminosity is _
l@xmath19 _ = 7.2@xmath310@xmath35 erg s@xmath9 .
short - term light curve shows significant variability and the long - term data shows evidence of variability within a factor of @xmath2 4 after 2002 in the next @xmath2 4 years .
the spectral parameters and variation of the source suggest that this may also be a candidate xrb . + * xmm-17 * : this source is located @xmath2 2@xmath38 away from main body of the galaxy and its spectrum is described by a pl with @xmath18 @xmath2 2 with an unabsorbed luminosity of _
l@xmath19 _ = 3.3@xmath310@xmath35 erg s@xmath9 .
its short - term light curve shows significant level of variability , while its long - term data show a factor of @xmath2 3 variation between observations .
this source may also be a candidate xrb .
+ * xmm-21 * : in our analysis , this source was best fitted with a one - component absorbed pl model ( @xmath18 @xmath2 2 ) .
both its short - term and long - term data show strong evidence for variability .
the source has an unabsorbed luminosity of _
l@xmath19 _ = 2.0@xmath310@xmath35 erg s@xmath9 which suggest that this source is likely to be an xrb .
however in the list of wmr2006 it was defined as a ulx .
its spectrum was fitted with a pl and a bremss equally well at a luminosity _ l@xmath19
_ = 1.2@xmath310@xmath33 erg s@xmath9 . therefore
, further observations are needed to clarify its true nature .
in the above sections , we presented spectral and temporal analyses on the non - nuclear x - ray point sources in the three nearby galaxies , ngc 4395 , ngc 4736 , and ngc 4258 selected on the basis of their low hydrogen column densities with low activity seyfert nuclei .
our main conclusions can be noted as follows : \(1 ) total number of point sources in the _ d@xmath0 _ area of each of these galaxies are similar ( 29 for ngc 4395 , 21 for ngc 4736 , and 23 for ngc4258 ) . data at hand however , allows spectral analysis for only a smaller number ( a total of 16 out of 74 ) of these sources .
luminosity for these sources fall into the range ( 0.4 - 22.7)@xmath310@xmath35 erg s@xmath9 .
14 of these 16 sources have positions clearly indicating that they have no contamination from background or foreground objects or structures .
\(2 ) for comparison purposes , we have analyzed archival _
chandra _ data of the three galaxies .
13 of these 16 _ xmm - newton _ sources have also been detected by _
chandra_. their source positions , in general , show a clear offset @xmath2 2@xmath26 from the center of _ xmm - newton _ error circle ( 15@xmath26 )
. only two of these 13 sources have been examined spectrally due to sufficient statistics provide by _ chandra _ observations .
\(3 ) four of these 16 sources are ulx candidates owing to their spectral characteristics and high luminosities .
two of these sources are in ngc 4258 ( xmm-2 and xmm-3 ) and they are best fitted with an absorbed diskbb model with a hot inner disk temperature ( _ t@xmath31 _ @xmath2 1.43 kev and 1.28 kev respectively ) .
these values are within the range given for galactic black - hole xrbs in the high - state emission ( mcclintock & remillard 2006 ) . on the other hand ,
2 ulxs ( xmm-2 in ngc 4736 and xmm-2 in ngc 4258 ) show significant transient behavior .
the statistically significant improvement over a one - component fit is found in the pl+diskbb model fit for the ulx in ngc 4736 .
the disk temperature is also cool ( 0.75 kev ) in this case .
similar disk temperatures for ulx sources in m101 ( jenkins et al .
2004 ) and in the interacting pair galaxies ngc 4485/4490 ( gladstone & roberts , 2009 ) were already found .
based on the argument that black hole masses scale inversely with accretion disc temperatures ( _ t@xmath31 _ @xmath2 m@xmath40 ) , this was interpreted as evidence of intermediate mass black holes ( imbhs ) , with masses in the range of 10@xmath41 - 10@xmath10 m@xmath12 for some ulx sources ( colbert & mushotzky 1999 ) . on the other hand , assuming spherical accretion onto a central black - hole in a binary system at the eddington limit as suggested by makishima et al ( 2000 ) , our point source luminosities imply black hole masses in the stellar range of ( 2 - 15 ) m@xmath12 .
this is consistent with galactic stellar - mass black holes which were found to lie in the range of ( 2 - 23 ) m@xmath12 ( mcclintock @xmath6 remillard 2004 ) .
most probably , these sources belong to the extreme end of the xrb population in their galaxies .
many similar luminous point sources in nearby galaxies are also reported ( jenkins et al .
2004 ; gladstone @xmath6 roberts 2009 ) .
\(4 ) of the remaining 12 sources , 8 are best fitted with absorbed pl spectra ( @xmath18 = 1.4 - 2.8 ) .
they generally have luminosities around the eddington limit ( @xmath36 10@xmath35 ) .
photon indices of 7 of these sources are in agreement with low / hard state black - hole xrbs ( tanaka 2001 ) .
remaining 2 sources ( xmm-10 and xmm-16 in ngc 4258 ) have steeper power - law slopes ( @xmath18 @xmath28 2 ) representing strong comptonized thermal emission from a black - hole accretion disc ( jenkins et al .
2004 , mcclintock @xmath6 remillard 2004 ) . based on x - ray data these sources
are considered to be black - hole xrbs .
\(5 ) the locations of the point sources could also provide clues to the nature of the x - ray emission ; however , at times they could also be misleading . positions of our sources suggest that almost all are associated with their host galaxies except possibly one .
as explained above , this latter source ( xmm-6 in ngc 4258 ) could be a background agn due to its spectral shape ( pl with @xmath18 @xmath2 1.9 which is typical of agns ) and its coincidence with a stellar object ( in corresponding dss image ) , warranting further investigation .
\(6 ) a noticeable overall result of the spectral analysis of the sources from these three galaxies , is that x - ray sources from different classes can be distinguished .
one source ( xmm-5 of ngc 4395 ) is distinctly different from rest . a closer inspection by our optical follow up observation revealed that , it was indeed different type ; an snr with a best spectral fit to a vpshock , typical of such objects ( sonbas et al . 2010 ) .
\(7 ) we have identified only one sss with typical low luminosity values around 5@xmath310@xmath39 erg s@xmath9 in one of these galaxies ( xmm-23 in ngc 4395 ) .
this class of sources generally are known to be a white dwarf accreting at high rates resulting in nuclear burning on the surface with _ kt _
@xmath2 ( 10 - 100 ) ev ( di stefano et al .
2004 ; stiele et al . 2010 ) .
our source also confirms these limits . in more general terms
, we can conclude that x - ray astronomy as reached at a stage to differentiate several source types in the galaxies beyond our local group . with increasing statistics
, the role of such sources in the evolution of stars and such galaxies will be clearer .
aysun akyuz acknowledges support from the eu fp6 transfer of knowledge project `` astrophysics of neutron stars '' ( mktd - ct-2006 - 042722 ) .
we thank frank haberl and nazim aksaker for their very valuable help .
we also thank the tubitak national observatory ( tug ) for their support for observing times and equipment .
ngc 4395 & 12:25:48.9 + 33:32:47.8 & sa(s)m & 4.2 & 38 & 1.33 & 13.3 + ngc 4736 & 12:50:52.6 + 41:07:09.3 & ( r)sa(r)ab & 4.3 & 33 & 1.43 & 11.8 + ngc 4258 & 12:18:57.6 + 47:18:13.4 & sab(s)bc & 7.7 & 71 & 1.20 & 14.2 + ngc 4395 & xmm-2 & 2.9 & p1@xmath42 & + & xmm-5 & 2.1 & p2@xmath42 & cxou j122539.5 + 333204@xmath43 + & xmm-6 & 0.7 & p3@xmath42 & cxou j122549.063 + 333201.82@xmath44 + & xmm-10 & 2.0 & p4@xmath42 & cxou j122547.287 + 333447.75@xmath44 + & xmm-23 & 1.9 & & cxou j122545.271 + 333103.40@xmath44 + & & & & + ngc 4736 & xmm-2 & 0.9 & & cxou j125048.598 + 410742.49@xmath44 + & xmm-12 & 2.3 & x-7@xmath45 & cxou j125047.618 + 410512.26@xmath44 + & xmm-18 & 2.1 & x-13@xmath45 & cxou j125104.4 + 410725@xmath46 + & & & & + ngc 4258 & xmm-2 & 3.1 & & + & xmm-3 & 2.1 & p16@xmath47 & cxou j121857.8 + 471607@xmath43 + & xmm-6 & 2.4 & & cxou j121843.8 + 471731@xmath43 + & xmm-8 & 1.5 & p14@xmath47 & + & xmm-10 & 6.4 & p13@xmath47 & + & xmm-16 & 3.3 & & cxou j121839.317 + 471919.27@xmath44 + & xmm-17 & 2.0 & & cxou j121849.481 + 471646.42@xmath44 + & xmm-21 & 3.2 & p15@xmath47 & cxou j121856.419 + 472125.56@xmath44 + ngc 4395 & 0112521901 & 2002.05.31 & 13978 & ff / thin + & * 0142830101 * & * 2003.11.30 * & * 89000 * & * ff / medium * + + ngc 4736 & 0094360701 & 2002.06.26 & 17046 & ff / medium + & * 0404980101 * & * 2006.11.27 * & * 37245 * & *
ff / thin * + + ngc 4258 & 0110920101 & 2000.12.08 & 16548 & ff / medium + & 0059140101 & 2001.05.06 & 9488 & ff / medium + & 0059140201 & 2001.06.17 & 10058 & ff / medium + & 0059140401 & 2001.12.17 & 11798 & ff / medium + & * 0059140901 * & * 2002.05.22 * & * 13646 * & * ff / medium * + & * 0400560301 * & * 2006.11.17 * & * 62635 * & * ff / medium * + xmm-17 & 12:25:20.49 & + 33:33:00.8 & 1.22e+03 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-55 & 12:25:22.18 & + 33:33:16.9 & 1.92e+02 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-184 & 12:25:29.06 & + 33:36:11.4 & 2.05e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-107 & 12:25:29.26 & + 33:29:01.3 & 8.88e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-43 & 12:25:34.45 & + 33:32:11.8 & 2.32e+02 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-113 & 12:25:35.28 & + 33:28:31.5 & 7.00e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-147 & 12:25:39.25 & + 33:33:49.8 & 4.58e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-5 & 12:25:39.55 & + 33:32:04.1 & 5.35e+03 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-99 & 12:25:41.09 & + 33:31:09.4 & 3.62e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-109 & 12:25:43.63 & + 33:35:07.0 & 6.69e+01 & [email protected] & [email protected] & [email protected]+35 + & & & & & & + xmm-145 & 12:25:43.79 & + 33:28:54.4 & 6.23e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-16 & 12:25:43.95 & + 33:30:00.2 & 1.08e+03 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-23 & 12:25:45.11 & + 33:31:04.6 & 6.43e+02 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-41 & 12:25:47.05 & + 33:36:07.4 & 2.85e+02 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-10 & 12:25:47.20 & + 33:34:47.4 & 1.51e+03 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-199 & 12:25:48.70 & + 33:28:39.3 & 1.32e+01 & [email protected] & [email protected] & [email protected]+35 + & & & & & & + xmm-6 & 12:25:49.00 & + 33:32:03.5 & 2.30e+03 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-60 & 12:25:54.08 & + 33:29:05.9 & 1.82e+02 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-32 & 12:25:54.44 & + 33:30:47.0 & 6.10e+02 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-114 & 12:25:57.32 & + 33:29:47.1 & 5.50e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-66 & 12:25:57.35 & + 33:30:39.8 & 8.77e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-155 & 12:25:57.98 & + 33:36:09.8 & 1.34e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-100 & 12:25:58.87 & + 33:28:23.7 & 9.08e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-24 & 12:25:59.79 & + 33:33:22.0 & 5.43e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-2 & 12:26:01.42 & + 33:31:32.0 & 1.34e+05 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-86 & 12:26:04.10 & + 33:32:50.6 & 6.42e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-201 & 12:26:13.79 & + 33:33:45.9 & 1.36e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-215 & 12:26:15.83 & + 33:33:48.0 & 1.44e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-135 & 12:26:16.42 & + 33:30:18.2 & 3.74e+01 & [email protected] & [email protected] & [email protected]+36 + xmm-58 & 12:50:27.02 & + 41:05:07.5 & 7.49e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-31 & 12:50:33.10 & + 41:05:12.8 & 1.48e+02 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-73 & 12:50:34.93 & + 41:07:43.7 & 3.16e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-19 & 12:50:35.25 & + 41:10:26.4 & 2.42e+02 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-63 & 12:50:35.73 & + 41:06:43.9 & 3.54e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-76 & 12:50:37.69 & + 41:10:37.9 & 3.48e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-12 & 12:50:47.72 & + 41:05:10.8 & 5.28e+02 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-41 & 12:50:48.00 & + 41:08:35.1 & 2.25e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-2 & 12:50:48.64 & + 41:07:40.5 & 4.77e+04 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-72 & 12:50:50.14 & + 41:11:29.4 & 2.39e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-3 & 12:50:50.27 & + 41:07:10.1 & 2.63e+03 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-36 & 12:50:52.57 & + 41:02:52.7 & 1.92e+02 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-24 & 12:50:53.00 & + 41:08:43.5 & 1.06e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-56 & 12:50:53.48 & + 41:04:55.2 & 4.19e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-6 & 12:50:53.67 & + 41:06:34.2 & 3.57e+02 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-39 & 12:50:54.24 & + 41:09:06.4 & 4.94e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-4 & 12:50:56.10 & + 41:06:55.9 & 3.55e+03 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-11 & 12:50:59.87 & + 41:02:53.8 & 5.45e+02 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-44 & 12:51:02.00 & + 41:03:04.7 & 9.74e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-18 & 12:51:04.43 & + 41:07:26.0 & 2.10e+02 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-27 & 12:51:09.01 & + 41:02:57.8 & 2.37e+02 & [email protected] & [email protected] & [email protected]+36 + xmm-16 & 12:18:39.38 & + 47:19:19.1 & 2.37e+03 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-93 & 12:18:42.24 & + 47:23:46.9 & 4.19e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-6 & 12:18:43.94 & + 47:17:31.7 & 6.91e+03 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-25 & 12:18:45.44 & + 47:18:32.1 & 1.28e+03 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-10 & 12:18:45.48 & + 47:24:20.5 & 5.06e+03 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-33 & 12:18:45.78 & + 47:18:07.5 & 3.44e+01 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-27 & 12:18:46.00 & + 47:23:12.9 & 7.49e+02 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-26 & 12:18:47.29 & + 47:20:22.0 & 4.09e+02 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-2 & 12:18:47.57 & + 47:20:53.8 & 3.26e+04 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-67 & 12:18:48.46 & + 47:22:26.1 & 8.41e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-17 & 12:18:49.55 & + 47:16:46.6 & 1.41e+03 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-38 & 12:18:53.68 & + 47:17:24.5 & 8.64e+01 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-8 & 12:18:54.72 & + 47:16:49.4 & 4.93e+03 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-104 & 12:18:54.79 & + 47:13:59.7 & 2.24e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-51 & 12:18:55.14 & + 47:23:02.5 & 1.90e+02 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-82 & 12:18:56.27 & + 47:15:08.1 & 5.03e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-21 & 12:18:56.34 & + 47:21:25.6 & 1.46e+03 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-41 & 12:18:57.93 & + 47:16:56.7 & 1.70e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-3 & 12:18:57.94 & + 47:16:07.7 & 1.36e+04 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-20 & 12:18:57.96 & + 47:19:07.4 & 9.25e+01 & [email protected] & [email protected] & [email protected]+37 + & & & & & & + xmm-91 & 12:19:04.35 & + 47:14:53.3 & 5.58e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-54 & 12:19:04.90 & + 47:16:03.9 & 1.53e+02 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-102 & 12:19:13.23 & + 47:14:11.2 & 2.59e+01 & [email protected] & [email protected] & [email protected]+36 + & & & & & & + xmm-87 & 12:19:18.59 & + 47:16:14.7 & 3.12e+01 & [email protected] & [email protected] & [email protected]+36 + ngc 4395 & 882 & 2000.06.20 & 20.0 & acis - s + + ngc 4736 & 808 & 2000.05.13 & 50.0 & acis - s + & 9553 & 2008.02.16 & 25.7 & acis - i + + ngc 4258 & 350 & 2000.04.17 & 14.5 & acis - s + & 1618 & 2001.05.28 & 22.0 & acis - s + & 2340 & 2001.05.29 & 7.0 & acis - s +
l l c c c [ table:1 ] ngc 4395 & xmm-2 & 400 & 254/222 & - + & xmm-5 & 900 & 114/98 & - + & xmm-6 & 1200 & 100/74 & 97.6 per cent + & xmm-10 & 2000 & 29/44 & - + & xmm-23 & 4500 & 23/19 & - + ngc 4736 & xmm-2 & 200 & 880/194 & @xmath2899.9 per cent + & xmm-12 & 2200 & 17/17 & - + & xmm-18 & 3000 & 9/12 & - + ngc 4258 & xmm-2 & 900 & 55/27 & @xmath2899.9 per cent + & xmm-3 & 700 & 155/91 & @xmath2899.9 per cent + & xmm-6 & 750 & 223/85 & @xmath2899.9 per cent + & xmm-8 & 1100 & 94/58 & 99.8 per cent + & xmm-10 & 1500 & 47/42 & - + & xmm-16 & 1500 & 68/42 & 99.3 per cent + & xmm-17 & 2500 & 41/25 & 97.7 per cent + & xmm-21 & 2000 & 142/32 & @xmath2899.9 per cent + xmm-2 & bbody & @xmath48 & & @xmath49 & 253.60/142 & 3.43 & 6.20 + & pl & @xmath50 & @xmath51 & & 209.13/142 & 6.36 & 11.51 + & diskbb & @xmath52 & & @xmath53 & 224.29/142 & 5.23 & 9.46 + & bremss & @xmath54 & & @xmath55 & 206.86/142 & 8.37 & 15.14 + & & & & & & & + xmm-5 & bbody & 0.04 & & 0.26 & 137.22/58 & 0.39 & 0.70 + & pl & 0.31 & 3.62 & & 128.66/58 & 4.27 & 7.72 + & diskbb & 0.18 & & 0.67 & 129.77/58 & 0.93 & 0.68 + & bremss & 0.12 & & 0.35 & 127.49/58 & 0.58 & 1.05 + & @xmath56 & @xmath57 & & @xmath58 & @xmath59 & @xmath60 & @xmath61 + & & & & & & & + xmm-6 & bbody & 0.00 & & 0.59 & 188.68/56 & 0.71 & 1.28 + & @xmath62 & @xmath63 & @xmath64 & & @xmath65 & @xmath66 & @xmath67 + & diskbb & 0.03 & & @xmath68 & 74.58/56 & 1.20 & 2.17 + & bremss & @xmath69 & & @xmath70 & 62.18/56 & 1.37 & 2.47 + & & & & & & & + xmm-10 & bbody & 0.00 & & 0.47 & 59.16/29 & 0.19 & 0.34 + & @xmath62 & @xmath71 & @xmath72 & & @xmath73 & @xmath74 & @xmath75 + & diskbb & 0.02 & & @xmath76 & 33.81/29 & 0.31 & 0.56 + & bremss & @xmath77 & & @xmath78 & 31.21/29 & 0.40 & 0.72 + & & & & & & & + xmm-23 & @xmath79 & @xmath80 & & @xmath81 & @xmath82 & @xmath83 & @xmath84 + & pl & @xmath85 & @xmath86 & & 12.49/10 & 0.98 & 0.17 + & diskbb & @xmath87 & & @xmath88 & 11.57/10 & 0.62 & 0.27 + & bremss & @xmath89 & & @xmath90 & 11.67/10 & 0.61 & 0.23 + xmm-2 & pl+bbody & @xmath91 & @xmath92 & @xmath93 & 206.75/140 & 12.8 & 23.16 + & @xmath94 & @xmath95 & @xmath96 & @xmath97 & @xmath98 & @xmath99 & @xmath100 + & pl+mekal & @xmath101 & @xmath102 & @xmath103 & 168.06/140 & 23.1 & 41.81 + & & & & & & & + xmm-5 & pl+bbody & @xmath104 & @xmath105 & @xmath106 & 98.02/56 & 0.72 & 1.30 + & pl+diskbb & 0.28 & 3.44 & 0.06 & 128.78/56 & 0.08 & 0.14 + & pl+mekal & @xmath107 & @xmath108 & @xmath109 & 88.59/56 & 5.96 & 10.78 + & & & & & & & + xmm-6 & pl+bbody & @xmath110 & @xmath111 & 199.36 & 61.73/54 & 1.45 & 2.67 + & pl+diskbb & @xmath112 & @xmath113 & @xmath114 & 60.30/54 & 1.37 & 2.47 + & pl+mekal & @xmath115 & @xmath116 & @xmath117 & 56.21/54 & 2.36 & 4.27 + & & & & & & & + xmm-10 & pl+bbody & @xmath118 & @xmath119 & @xmath120 & 26.80/27 & 0.56 & 1.01 + & pl+diskbb & @xmath121 & @xmath122 & 0.005 & 26.78/27 & 0.03 & 0.04 + & pl+mekal & @xmath123 & @xmath124 & @xmath125 & 22.11/27 & 4.94 & 8.94 + & & & & & & & + xmm-23 & pl+bbody & @xmath126 & @xmath127 & @xmath128 & 9.94/8 & 18.68 & 33.81 + & pl+diskbb & 1.05 & 0.90 & 0.12 & 19.80/8 & 0.10 & 0.18 + & pl+mekal & @xmath129 & @xmath130 & @xmath131 & 10.06/8 & 6.66 & 12.04
+ h@xmath16(@xmath1326563 ) & 100 + sii(@xmath1326716 ) & 30.8 + sii(@xmath1326731 ) & 11.3 + oi(@xmath1326300 ) & 221.7 + oii(@xmath1337320,7330 ) & 413 + e@xmath37 & 0.02 + i(h@xmath16 ) & 2.3e-13 & erg @xmath5s@xmath9 + ( sii(@xmath1326716)+sii(@xmath1326731))/h@xmath16 & 0.42 + xmm-2 & pl & @xmath134 & @xmath135 & & 768.60/519 & 1.20 & 2.77 + & diskbb & @xmath136 & & @xmath137 & 695.47/519 & 6.72 & 15.52 + & bremss & @xmath138 & & @xmath139 & 534.87/519 & 8.53 & 19.70 + & & & & & & & + xmm-12 & pl & @xmath140 & @xmath141 & & 16.17/19 & 0.54 & 1.24 + & @xmath142 & @xmath143 & & @xmath144 & @xmath145 & @xmath146 & @xmath147 + & bremss & @xmath148 & & @xmath149 & 15.69/19 & 0.42 & 0.97 + & & & & & & & + xmm-18 & @xmath62 & @xmath150 & @xmath151 & & @xmath152 & @xmath153 & @xmath154 + & diskbb & @xmath155 & & @xmath156 & 19.80/20 & 0.24 & 0.50 + & bremss & @xmath157 & & @xmath158 & 17.58/20 & 0.29 & 0.66 + & & & & & & & + xmm-2 & pl+bbody & @xmath159 & @xmath160 & @xmath161 & 576.06/517 & 8.97 & 20.72 + & @xmath94 & @xmath162 & @xmath163 & @xmath164 & @xmath165 & @xmath166 & @xmath167 + & pl+mekal & @xmath168 & @xmath169 & @xmath170 & 718.94/517 & 9.92 & 24.51 + & & & & & & & + xmm-12 & pl+bbody & @xmath171 & @xmath172 & @xmath173 & 15.48/17 & 0.54 & 1.23 + & pl+diskbb & @xmath174 & @xmath175 & @xmath176 & 15.52/17 & 0.67 & 1.40 + & pl+mekal & @xmath177 & @xmath178 & 10.70 & 15.77/17 & 1.14 & 2.63 + & & & & & & & + xmm-18 & pl+bbody & @xmath179 & @xmath180 & @xmath181 & 14.62/18 & 1.24 & 2.59 + & pl+diskbb & @xmath182 & @xmath183 & @xmath184 & 14.78/18 & 3.29 & 6.87 + & pl+mekal & @xmath185 & @xmath186 & @xmath187 & 13.60/18 & 5.30 & 11.07 + & & & & & & & + xmm-2@xmath188 & pl & @xmath189 & @xmath190 & & 31.80/22 & 3.96 & 23.87 + & @xmath142 & @xmath191 & & @xmath192 & @xmath193 & @xmath194 & @xmath195 + & bremss & @xmath196 & & @xmath197 & 29.91/22 & 3.14 & 18.96 + & & & & & & & + xmm-3 & pl & @xmath198 & @xmath199 & & 297/231 & 6.60 & 39.77 + & @xmath142 & @xmath200 & & @xmath201 & @xmath202 & @xmath203 & @xmath204 + & bremss & @xmath205 & & @xmath206 & 237.76/231 & 4.23 & 25.49 + & & & & & & & + xmm-6 & @xmath62 & @xmath207 & @xmath208 & & @xmath209 & @xmath210 & @xmath211 + & diskbb & @xmath212 & & @xmath213 & 161.22/142 & 1.63 & 9.82 + & bremss & @xmath214 & & @xmath215 & 135.84/142 & 2.03 & 12.23 + & & & & & & & + xmm-8 & pl & @xmath216 & @xmath217 & & 125.29/109 & 2.04 & 12.28 + & diskbb & 0.01 & & 0.83 & 249.70/109 & 1.01 & 6.08 + & bremss & @xmath218 & & @xmath219 & 166.01/109 & 1.37 & 8.25 + & & & & & & & + xmm-10 & @xmath62 & @xmath220 & @xmath221 & & @xmath222 & @xmath223 & @xmath224 + & diskbb & @xmath225 & & @xmath226 & 87.16/96 & 1.83 & 11.02 + & bremss & @xmath227 & & @xmath228 & 86.82/96 & 2.61 & 15.73 + & & & & & & & + xmm-16 & @xmath62 & @xmath229 & @xmath230 & & @xmath231 & @xmath232 & @xmath233 + & diskbb & @xmath234 & & @xmath235 & 146.78/76 & 0.68 & 4.10 + & bremss & @xmath236 & & @xmath237 & 117.96/76 & 0.94 & 5.72 + & & & & & & & + xmm-17 & @xmath62 & @xmath238 & @xmath239 & & @xmath240 & @xmath241 & @xmath242 + & diskbb & 0.05 & & 1.10 & 169.73/80 & 0.55 & 3.31 + & bremss & @xmath243 & & @xmath244 & 131.37/80 & 0.72 & 4.34 + & & & & & & & + xmm-21 & @xmath62 & @xmath245 & @xmath246 & & @xmath247 & @xmath248 & @xmath249 + & diskbb & 0.00 & & @xmath250 & 100.56/78 & 0.20 & 1.20 + & bremss & @xmath251 & & @xmath252 & 81.61/78 & 0.29 & 1.74 + xmm-2@xmath188 & pl+bbody & @xmath253 & @xmath254 & @xmath255 & 27.76/20 & 4.05 & 24.42 + & pl+diskbb & @xmath256 & @xmath257 & @xmath258 & 27.35/20 & 2.80 & 16.88 + & pl+mekal & @xmath259 & @xmath260 & @xmath261 & 37.11/20 & 7.89 & 47.57 + & & & & & & & + xmm-3 & pl+bbody & @xmath262 & @xmath263 & @xmath264 & 244.56/229 & 4.39 & 19.76 + & pl+diskbb & @xmath265 & @xmath266 & @xmath267 & 219.18/229 & 3.28 & 19.76 + & pl+mekal & @xmath268 & @xmath269 & @xmath270 & 278.16/229 & 4.82 & 29.05 + & & & & & & & + xmm-6 & pl+bbody & @xmath271 & @xmath272 & @xmath273 & 133.26/140 & 2.30 & 13.85 + & pl+diskbb & @xmath274 & @xmath275 & @xmath276 & 133.54/140 & 2.42 & 14.58 + & pl+mekal & @xmath277 & @xmath278 & 4.12 & 135.64/140 & 2.22 & 13.37 + & & & & & & & + xmm-8 & pl+bbody & @xmath279 & @xmath280 & @xmath281 & 94.53/107 & 7.84 & 47.24 + & pl+diskbb & @xmath216 & @xmath217 & 0.00 & 124.87/107 & 2.86 & 17.23 + & @xmath282 & @xmath283 & @xmath284 & @xmath285 & @xmath286 & @xmath287 & @xmath288 + & & & & & & & + xmm-10 & pl+bbody & @xmath289 & @xmath290 & 199.36 & 81.24/92 & 5.19 & 31.27 + & pl+diskbb & @xmath291 & @xmath292 & @xmath293 & 77.65/92 & 5.13 & 30.91 + & pl+mekal & @xmath294 & @xmath295 & @xmath296 & 79.81/92 & 1.85 & 11.14 + & & & & & & & + xmm-16 & pl+bbody & @xmath297 & @xmath298 & 199.36 & 99.31/74 & 2.16 & 13.02 + & pl+diskbb & @xmath299 & @xmath300 & @xmath301 & 101.47/74 & 1.6 & 9.64 + & pl+mekal & @xmath302 & @xmath303 & @xmath304 & 76.69/74 & 0.43 & 2.59 + & & & & & & & + xmm-17 & pl+bb & @xmath305 & @xmath306 & @xmath307 & 101.54/78 & 1.12 & 6.75 + & pl+diskbb & @xmath308 & @xmath309 & @xmath310 & 101.55/78 & 1.83 & 11.03 + & pl+mekal & @xmath311 & @xmath312 & @xmath313 & 76.36/78 & 0.42 & 2.53 + & & & & & & & + xmm-21 & pl+bb & @xmath314 & @xmath315 & @xmath316 & 84.56/76 & 0.34 & 2.05 + & pl+diskbb & @xmath317 & @xmath318 & @xmath319 & 80.06/76 & 0.32 & 1.92 + & pl+mekal & @xmath320 & @xmath321 & @xmath322 & 94.63/76 & 0.33 & 1.98 + | we present results from a study of the non - nuclear discrete sources in a sample of three nearby spiral galaxies ( ngc 4395 , ngc 4736 , and ngc 4258 ) based on _ xmm - newton _ archival data supplemented with _ chandra _ data for spectral and timing analyses .
a total of 75 x - ray sources has been detected within the _ d@xmath0 _ regions of the target galaxies .
the large collecting area of _ xmm - newton _ makes the statistics sufficient to obtain spectral fitting for 16 ( about 20@xmath1 ) of these sources . compiling the extensive archival exposures available , we were able to obtain the detailed spectral shapes of diverse classes of point sources .
we have also studied temporal properties of these luminous sources .
11 of them are found to show short - term ( less than 80 ks ) variation while 8 of them show long - term variation within factors of @xmath2 2 to 5 during a time interval of @xmath2 2 to 12 years .
timing analysis provides strong evidence that most of these sources are accreting x - ray binary ( xrb ) systems .
one source that has properties different than others was suspected to be a supernova remnant ( snr ) , and our follow - up optical observation confirmed it .
our results indicate that sources within the three nearby galaxies are showing a variety of source populations , including several ultraluminous x - ray sources ( ulxs ) , x - ray binaries ( xrbs ) , transients together with a super soft source ( sss ) and a background active galactic nucleus ( agn ) candidate . |
we used the global matrix method @xcite to simulate the skull behavior to first approximation .
we first divide the space into three flat isotropic solid layers of thickness @xmath62 sandwiched by two fluid semi - infinite spaces .
then , as a solution to the helmholtz equation , we assume plane ( inhomogeneous ) waves in each layer with @xmath63 time dependence . for @xmath64 corresponding to longitudinal and transverse waves , respectively , the complex bulk wavenumbers @xmath65 determine the wavevectors @xmath66 , with the polar angle @xmath67 . for simplicity ,
the fluids 0 and 4 are considered the same and the subindex @xmath68 is omitted as only longitudinal waves can propagate on it .
the plane ( inhomogeneous ) wave potentials @xmath69 represent the wave propagation in the fluid , whereas in the solid layers @xmath70 , @xmath71 describes both longitudinal and transverse plane ( inhomogeneous ) wave potentials . setting @xmath72 at the first fluid - solid interface with @xmath53 pointing upwards ( fig .
[ fig : diagram ] ) we can explicitly write @xmath73 where @xmath74 correspond to the incident , reflected , and transmitted complex wave amplitudes , respectively . in the solid , volumetric @xmath15 and shear @xmath16 viscosities are included in the lam constants as @xmath75 and @xmath76 to account for absorption @xcite , which is reflected in the imaginary part of the bulk wavenumbers . applying continuity of the displacement and the stresses at each boundary we obtain a system of 14 equations were the unknowns are @xmath77and @xmath78 .
the calculation starts by solving the transmission problem for @xmath79 and real @xmath80 in a given @xmath81 range and step size @xmath82 .
once the transmission maxima are identified with the different modes @xmath37 , the @xmath83 values serve as an input to a golden section search for the @xmath84 pair ( now with complex @xmath80 ) that produces a singular matrix under the condition @xmath85 .
a thoroughly explanation on a similar method can be found in @xcite .
once the mode s dispersion is known , the pressure in the fluid on top of the skull can be expanded as @xmath86 .
following @xcite , we can calculate the fate of an initial pressure field @xmath87 and its derivative @xmath88 as @xmath39 where @xmath41 depends on the initial conditions , @xmath89 and @xmath45 . as the solid
is assumed to be isotropic , one can take advantage of the symmetry around the azimuthal angle @xmath90 and obtain @xmath91 which corresponds to a zero - order hankel transform over the wavenumber in the radial direction @xmath92 . using the shorthand notation from @xcite
it becomes @xmath93 thus , we can write @xmath94 the numerical evaluation of eq . [ eq : timehankel ] is performed using bessel series expansion @xcite with 1500 points in total , distributed in 15 mm to avoid unwanted reflections that arise when this method is used .
@xmath95 was extracted from the filtered experimental data from fig .
[ fig : gresults](c ) .
the results of the simulations using the different viscoelastic constants for the negative and positive propagation direction are shown in fig .
[ fig : simulatedfield ] .
calculation of the rms pressure across time for fig .
[ fig : simulatedfield](a ) and ( b ) yields the theoretical curves from fig .
[ fig : xdecay](a ) .
spatio - temporal skull - guided - wave propagation from simulations ( ( a ) , ( b ) ) and experiments ( c ) . the color scale represents the normalized pressure and the dashed black line the speed of sound in the fluid . ]
we can calculate the effect of the detector s finite size and geometry on near - field experiments , particularly when measuring evanescent waves . for a circular detector of radius @xmath96 detecting plane waves of amplitude @xmath97 and wavevector @xmath54 ,
the sensitivity @xmath98 is given by the well known result for normal incidence ( @xmath100 ) with the detector s sensitivity being proportional to its area is recovered .
now , for an inhomogeneous plane wave propagating on top of a flat surface placed at a distance @xmath101 and rotated an angle @xmath102 relative to the detector s normal vector where @xmath104 , @xmath105 , and @xmath106 . taking @xmath107 with @xmath108 900 m / s , @xmath109 0.5 mm and normalizing by @xmath110 we obtain the relative spatial sensitivity for different cases shown in fig .
[ fig : sensitivity ] .
the detection of inhomogeneous plane waves seems more efficient than for plane waves in the low frequency regime .
however , there is a strong cut - off below 1.5 mhz for inhomogeneous plane waves at different @xmath101 and @xmath102 .
this simple calculation is valid only far away from the source and serves to illustrate how the detector affects the measurements in the skull s near - field .
close to the source , the exact shape of the field has to be taken in account .
42ifxundefined [ 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/10.1121/1.381852 [ * * , ( ) ] link:\doibase 10.1117/1.nph.1.1.011003 [ * * , ( ) ] http://stacks.iop.org/1741-2552/13/i=3/a=031003 [ * * , ( ) ] \doibase doi:10.1186/2050 - 5736 - 2 - 17 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.addr.2014.01.008 [ * * , ( ) ] , http://stacks.iop.org/0031-9155/47/i=8/a=301 [ * * , ( ) ]
\doibase http://dx.doi.org/10.1016/s0301-5629(02)00503-3 [ * * , ( ) ] \doibase http://dx.doi.org/10.1121/1.1645610 [ * * , ( ) ] http://stacks.iop.org/0031-9155/54/i=9/a=001 [ * * , ( ) ]
[ * * , ( ) ] link:\doibase 10.1118/1.3668316 [ * * , ( ) ] link:\doibase 10.1002/jbio.201400152 [ * * , ( ) ] http://stacks.iop.org/0031-9155/61/i=5/a=1932 [ * * , ( ) ] link:\doibase 10.1126/science.262.5138.1422 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.45.554 [ * * , ( ) ] link:\doibase 10.1103/physrevlett.102.144301 [ * * , ( ) ] link:\doibase 10.1103/physrevb.85.174301 [ * * , ( ) ] link:\doibase 10.1007/978 - 1 - 4612 - 5856 - 8 [ _ _ ] ( , ) @noop _ _ , ed .
( , , ) @noop * * , ( ) http://www.jstor.org/stable/93792 [ * * , ( ) ] http://www.jstor.org/stable/94262 [ * * , ( ) ] @noop * * , ( ) \doibase http://dx.doi.org/10.1016/0041-624x(91)90168-8 [ * * , ( ) ] http://stacks.iop.org/0034-4885/73/i=1/a=016102 [ * * , ( ) ] link:\doibase 10.1109/tuffc.2008.790 [ * * , ( ) ] `` , '' in link:\doibase 10.1007/978 - 94 - 007 - 0017 - 8_7 [ _ _ ] , ( , , ) pp .
link:\doibase 10.1109/tuffc.2016.2587079 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.ultrasmedbio.2013.10.018 [ * * , ( ) ] @noop link:\doibase 10.1109/58.393096 [ * * , ( ) ] \doibase http://dx.doi.org/10.1016/j.ultrasmedbio.2010.02.012 [ * * , ( ) ] @noop _ _ ( , ) @noop _ _ ( , ) @noop _ _ , ed . , , vol .
( , ) _ _ , @noop ph.d .
thesis , ( ) @noop _ _ ( , ) @noop _ _ ( , ) link:\doibase 10.1364/josaa.21.000053 [ * * , ( ) ] \doibase http://dx.doi.org/10.1121/1.413048 [ * * , ( ) ] | ultrasound waves propagating in water or soft biological tissue are strongly reflected when encountering the skull , which limits the use of ultrasound - based techniques in transcranial imaging and therapeutic applications .
current knowledge on the acoustic properties of the cranial bone is restricted to far - field observations , leaving its near - field properties unexplored .
we report on the existence of skull - guided acoustic waves , which was herein confirmed by near - field measurements of optoacoustically - induced responses in ex - vivo murine skulls immersed in water .
dispersion of the guided waves was found to reasonably agree with the prediction of a multilayered flat plate model .
it is generally anticipated that our findings may facilitate and broaden the application of ultrasound - mediated techniques in brain diagnostics and therapy .
the skull comprises a solid multilayered bony structure which plays a crucial role in protecting the brain from injuries , setting the intracranial pressure balance and determining many other anatomical and functional properties of a living organism . due to its mechanical properties , ultrasound waves propagating through water or soft biological tissue
are strongly reflected when encountering the skull @xcite . yet , several applications , such as optoacoustic neuroimaging @xcite , ultrasound neuromodulation @xcite , focused ultrasound surgery @xcite , and blood brain barrier opening for drug delivery @xcite , are able to make efficient use of the ultrasound waves transcranially , mainly because they only require a one - way transmission of the ultrasound wave through the skull . nonetheless , the spatial resolution and penetration of these transcranial techniques is severely limited due to the skull s presence . to this end ,
characterization of the acoustic properties of the cranial bone has been performed either with the aim of focusing ultrasound deep inside the human brain or was otherwise aimed at visualizing the mouse brain vasculature @xcite .
both cases consider the ultrasound source and/or the target region to be located far away from the skull , i.e. in its far - field . on the contrary
, the near - field of an object contains information that is irremediably lost if the object is scrutinized using only the far - field observations . for exam6ple , in near - field scanning optical microscopy @xcite , a spatial resolution higher than the far - field diffraction limit is achieved when scanning in the immediate vicinity of the object . also holographic techniques
@xcite can take advantage of the near - field information . in the context of acoustic waves in a solid plate , for which no direct electromagnetic wave analogy exists @xcite ,
the near - field properties may considerably change if the object effectively comprises a waveguide .
the related physical phenomenon , whose manifestation spans from rayleigh waves triggered by earthquakes @xcite down to microchips @xcite , was first studied by rayleigh @xcite and lamb @xcite whereas stoneley @xcite and scholte @xcite have further included the effect of solid / solid and solid / fluid interfaces .
currently , guided acoustic waves ( gaws ) are employed in non - destructive testing of plates and pipes @xcite or material characterization at micro- and nano - scopic scales @xcite .
gaws were also characterized in long cylindrical bones aiming at diagnosing osteoporosis @xcite , e.g. by determining the thickness of the human radius with multimode gaws in the axial transmission configuration @xcite .
the gaws were also excited optoacoustically in human radius phantoms @xcite .
however , near - field properties of the skull have not yet been explored . here
we prove the existence of guided acoustic waves in the murine skull bone by means of water - borne laser ultrasound experiments .
the experimental findings are supported by numerical simulations performed using a fluid - loaded flat multilayered plate model .
excitation of a fluid - immersed plate - like structure with a short laser pulse is expected to reveal a number of different wave propagation mechanisms , as depicted in fig .
[ fig : diagram ] . in general
, three main mechanisms may exist for ultrasonic wave propagation in the fluid - immersed skull : ( a ) leaky lamb - like waves , ( b ) bulk radiation to the fluid , and ( c ) guided waves .
the first are waves which are radiated into the fluid at an angle given by @xmath0 while their propagation along the skull occurs at supersonic speed @xmath1 .
schematic of the skull geometry depicting the acoustic wave propagation mechanisms and the experimental setup .
the right inset details the structure of the skull bone .
the @xmath2 coordinates follow the skull curvature , while the @xmath3 do not .
the wave propagation in the lower part of the skull has been omitted for clarity . ]
the second mechanism corresponds to direct radiation from the skull into the surrounding fluid , hence propagating at the speed of sound in the fluid @xmath4 .
our underlying hypothesis is that , if the skull further supports a mode having subsonic speed @xmath5 , the corresponding wave will be bound to the skull and will not emit radiation into the far - field . as a result
, one may only detect its existence in the near - field .
due to obvious experimental challenges involved in direct excitation of ultrasound waves in the skull s near - field , we used a pulsed laser radiation to induce an optoacoustic response in a thin layer of black burnish attached to the interior surface of the skull .
the layer effectively acts as a point broadband ultrasound source due to the thermoelastic effect ( see fig .
[ fig : diagram ] ) . a q - switched diode laser of 532 nm in wavelength ( edgewave gmbh , wrselen , germany ) was used to generate 3 @xmath6j pulses of 10 ns duration , which were focused onto the absorbing film attaining a beam diameter of approximately 100 @xmath6 m at the surface .
accurate mapping of the near - field acoustic field is achieved by scanning a 0.5 mm diameter polivynil difluoride ( pvdf ) needle hydrophone ( precision acoustics , uk ) in close proximity ( @xmath7100 @xmath6 m ) to the skull surface . due to the curvature of the mouse skull
, the exact three - dimensional scanning pattern was constructed using high resolution pulse - echo images of the skull surface acquired with a 30 mhz spherically focused pvdf ultrasound transducer ( precision acoustics , uk ) , as shown in the inset of fig . [
fig : timespace ] . during the experiments , the skull bone of a 6 weeks old mouse
was immersed in phosphate buffered saline ( pbs ) solution ( life technologies corp . , uk ) .
the detected ultrasonic signals covering the left frontal , parietal , and occipital bones are depicted in fig .
[ fig : timespace ] for the different time instants following the laser pulse .
the supersonic waves can only be observed shortly after the main bang ( 36 ns , blue rectangles ) . measured signal amplitude across the scanning path ( right inset ) for three different time points .
leaky waves are enclosed by blue squares .
an animated video with the wave propagation sequence is available @xcite . ]
the main features of both wavefronts traveling away from the source follow the speed of sound in the fluid @xmath8 [email protected] ) and the salinity ( 10 g@xmath9l ) of pbs according to @xcite ] note that the traces of the waves propagating at a subsonic speed are only evident after the spatio - temporal data @xmath10 is filtered in reciprocal space @xmath11 .
near- and far - field are then separated by the dashed line representing @xmath4 in fig .
[ fig : gresults](a ) after applying a two - dimensional fourier transform .
the existence of subsonic modes is evident as well the asymmetry regarding the propagation direction , the latter expected due to the inhomogeneous conformation of the skull . as a first approximation , we calculated the modal dispersion of a flat viscoelastic plate consisting of three isotropic layers immersed in water . assuming plane wave propagation and applying appropriate boundary conditions results in a linear system of equations yielding the mode dispersion as a function of @xmath11 @xcite ( for more details see appendix a ) .
overlaid in fig .
[ fig : gresults](a ) is the mode dispersion obtained using the density @xmath12 and longitudinal speed of sound @xmath13 of the cortical and trabecular bone layers @xcite , while the transverse speed of sound @xmath14 and the volumetric @xmath15 and shear @xmath16 viscoelastic losses were adjusted to fit the experimental results .
1969 kg@xmath9m@xmath17 , @xmath18 3476 m@xmath9s , @xmath19 1@xmath2010@xmath21 pa@xmath22s , @xmath23 1760 m@xmath9s , @xmath24 1@xmath2010@xmath25 pa@xmath22s .
trabecular bone : @xmath26 1055 kg@xmath9m@xmath17 , @xmath18 1886 m@xmath9s , @xmath19 1.5 pa@xmath22s , @xmath27 650 m@xmath9s , @xmath28 800 m@xmath9s , @xmath24 3 pa@xmath22s ] ( a ) measured and calculated ( overlay ) dispersion .
raw ( b ) and filtered ( c ) wave propagation as a function time and distance .
the dashed lines depict the speed of sound in water . ] the relative thickness of the individual layers ( see inset in fig . [
fig : diagram ] ) was assumed to follow the proportion of 0.26 , 0.5 , and 0.24 @xcite for the total skull thickness of @xmath29 320 @xmath6 m and @xmath30 300 @xmath6 m in the negative and positive propagation directions , respectively .
two subsonic lamb - rayleigh - like modes nearly overlap for frequencies above 2 mhz and match the experimental results with @xmath31 m@xmath9s . at lower frequencies the modes branch , thus resembling the behavior of cut - off - free symmetric s@xmath32 and antisymmetric a@xmath32 lamb modes . by defining a window in the subsonic region as @xmath33 , using a threshold to reject noise , and applying an inverse two - dimensional fourier transform , one obtains the filtered spatio - temporal propagation of the subsonic mode ( figs .
[ fig : gresults](c ) and [ fig : timespace ] ) .
this type of processing makes the subsonic wave clearly distinguishable , in contrast to the unfiltered data ( fig .
[ fig : gresults](b ) ) . the decay of the guided - wave in the propagation direction @xmath34 ( fig .
[ fig : xdecay](a ) ) is subsequently calculated by taking the root - mean - square ( rms ) of the filtered wave using an appropriate spatio - temporal window to isolate the propagating wave component from noise and any scattering events . if the effects of the skull inhomogeneities were negligible , one could expect a decay @xmath35 @xcite far away from the source with @xmath36 being an attenuation constant .
in addition , due to the generally broad - band nature of the optoacoustic source , the dispersion ( see fig .
[ fig : gresults](a ) ) needs to be considered for each individual mode @xmath37 . the pressure in the fluid on top of the skull ( @xmath38 ) can be calculated as @xmath39 where @xmath40 , @xmath41 depends on the initial pressure and its time derivative , @xmath42 with @xmath43 s@xmath44/m @xcite , and @xmath45 ( see appendix a ) . ( a ) guided wave decay along the propagation direction @xmath46 extracted from fig . [
fig : gresults](c ) and compared against the calculations .
( b ) angle of the skull surface relative to the scanning direction @xmath47 .
( c ) ultrasound pulse - echo image of the mouse skull in the scanned region . ]
the symmetry of the experimental curve is broken by a local maximum around @xmath48 mm away from the optoacoustic source .
the prediction of the flat - plate model is not able to explain the experimentally observed slow decay in the vicinity of the source .
the deviation may be attributed to the non - uniform spatial sensitivity of the 0.5 mm diameter detector used for the measurements as well as the acoustic scattering effects at the suture boundaries and other inhomogeneities of the skull .
indeed , the guided wave propagating parallel to the skull surface will reach the detector at different angles due to the skull curvature .
the sutures are located at @xmath49 mm and @xmath50 mm . while the suture to the right is involved in the local maximum of the experimental decay curve , the amplitude of the wave propagating to the left approaches noise levels just before reaching the left suture .
however , one may also observe in fig .
[ fig : gresults](c ) that the latter wave component continues propagating also for @xmath51 mm after abruptly changing its amplitude .
furthermore , close inspection of the imaginary part of the wavenumber @xmath52 reveals that the losses due to viscosity play a relatively minor role in the propagation of the skull - guided - waves within the studied frequency range ( see fig .
[ fig : attenuation](a ) ) since the effective losses are orders of magnitude weaker as compared to the corresponding losses in water for bulk wave propagation .
this made it difficult the selection of appropriate viscosity constants as both negligible and high volumetric @xmath15 and shear viscosities @xmath16 effectively gave rise to the same propagation scenario .
the attenuation in the @xmath53 direction remains also nearly unaffected by changes in the viscoelastic losses , as it is mostly determined by the geometrical relation between @xmath54 and @xmath55 .
figure [ fig : attenuation ] shows the different behavior for s@xmath32 and a@xmath32-like modes , particularly at the low frequency range . as a reference , 10 mm@xmath25 results in a characteristic half - decay distance of about 70 @xmath6 m .
imaginary part of the wavenumber in the ( a ) @xmath56 and the ( b ) @xmath53 directions as a function of the frequency .
the attenuation in water is plotted in ( a ) as reference .
the inset in ( b ) shows the ratio between the subsonic wavelength averaged for the @xmath57 and @xmath58 direction @xmath59 and the wavelength in the fluid @xmath60 as a function of the averaged skull thickness @xmath61 normalized as well by @xmath60 . ]
although our simplified model generally matches the experimental dispersion , limitations in the experimental spatio - temporal resolution along with additional effects induced by skull s curvature and inhomogeneities have made some of the measured modes difficult to discern .
particularly challenging is the accurate distinction between the mode located close to the sound line and the bulk wave mode propagating at a grazing angle ( see fig .
[ fig : gresults](a ) ) .
in general , the guided wave has been observed here to propagate for distances beyond 3 mm across the mouse skull .
although the experimental measurements are affected by scattering and artifacts due to the finite size of the detector and the skull curvature , the flat - plate model has been found to agree within an order of magnitude with the experimental results .
in addition , the theoretical model clearly points towards curvature and scattering as the main sources of attenuation for the skull - guided wave propagation . from a practical stand - point ,
the reported skull - guided waves could be potentially used as sole or complementary carriers of acoustic information across the skull in imaging or therapeutic applications . due to similitudes of the cranial bone structure in small mammals and humans
@xcite , it is generally expected that similar phenomena also exists in human skulls , although must be scaled accordingly in frequency and space .
acoustic scattering and exponential decay along the depth direction ( @xmath53 ) may restrict the practical use of the guided - wave phenomenon to low frequencies .
however , the achievable imaging resolution would not be compromised due to the shorter wavelength of the subsonic waves in comparison to bulk longitudinal waves in water for the same frequency ( see inset in fig . [
fig : attenuation](b ) ) .
although the dispersive nature of the skull - guided waves has to be further considered , the propagation problem is effectively 2.5-dimensional .
naturally , penetration depth is limited by the evanescent decay in the @xmath53 direction , yet could be sufficient to target cortical brain structures per - cranially
. parameters of the immersion fluid are crucial in determining whether the particular mode remains guided or turns leaky . in this
regard , our choice for water as a coupling medium is not random as its acoustic impedance is very close to most soft tissues @xcite including the brain , while water - based gels are further used as coupling medium in many biological applications .
however , if the skull is surrounded on its either side by a different medium , e.g. air , the wave phenomena may dramatically differ . in summary
, we report for the first time , to the best of our knowledge , on the existence of guided acoustic waves in skull bone surrounded by water - like media .
solid experimental evidence was further obtained on the dispersion of the skull - guided waves by means of near - field mapping of optoacoustically - generated broadband waves .
reasonable agreement was found with the predictions of a multilayered flat plate model .
characteristic decays in both the lateral and depth directions were characterized both experimentally and theoretically .
due to the general scalability of our model and the experimental observations , the reported guided waves are also expected to exist in human skulls .
thus , our findings can be used to facilitate and broaden the application of ultrasound - mediated techniques in brain diagnostics and therapy . |
in order to introduce the problem we analyze in the paper , let us start with some motivating examples . in high frequency trading
, an automatic agent decides the next action to be performed as sending or canceling a buy / sell order , on the basis of some market variables as well as private variables ( e.g. , stock price , traded volume , volatility , order books distributions as well as complex relations among these variables ) .
for instance in @xcite the trading strategy is learned in the form of a discrete function , described as a table , that has to be evaluated whenever a new scenario is faced and an action ( sell / buy ) has to be taken .
the rows of the table represent the possible scenarios of the market and the columns represent the variables taken into account by the agent to distinguish among the different scenarios . for each scenario
, there is an associated action .
every time an action need to be taken , the agent can identify the scenario by computing the value of each single variable and proceed with the associated action .
however , recomputing all the variable every time might be very expensive . by taking into account the structure of the function / table together with information on the probability distribution on the scenarios of the market and also the fact that some variables are more expensive ( or time consuming ) to calculate than others , the algorithm could limit itself to recalculate only some variables whose values determine the action to be taken .
such an approach can significantly speed up the evaluation of the function .
since market conditions change on a millisecond basis , being able to react very quickly to a new scenario is the key to a profitable strategy . in a classical bayesian active learning problem ,
the task is to select the right hypothesis from a possibly very large set @xmath1 each @xmath2 is a mapping from a set @xmath3 called the query / test space to the set ( of labels ) @xmath4 it is assumed that the functions in @xmath5 are unique , i.e. , for each pair of them there is at least one point in @xmath3 where they differ .
there is one function @xmath6 which provides the correct labeling of the space @xmath3 and the task is to identify it through queries / tests .
a query / test coincides with an element @xmath7 and the result is the value @xmath8 each test @xmath9 has an associated cost @xmath10 that must be paid in order to acquire the response @xmath11 since the process of labeling an example may be expensive either in terms of time or money ( e.g. annotating a document ) . the goal is to identify the correct hypothesis spending as little as possible .
for instance , in automatic diagnosis , @xmath12 represents the set of possible diagnoses and @xmath13 the set of symptoms or medical tests , with @xmath14 being the exact diagnosis that has to be achieved by reducing the cost of the examinations . in @xcite , a more general variant of the problem
was considered where rather than the diagnosis it is important to identify the therapy ( e.g. , for cases of poisoning it is important to quickly understand which antidote to administer rather than identifying the exact poisoning ) .
this problem can be modeled by defining a partition @xmath15 on @xmath12 with each class of @xmath15 representing the subset of diagnoses which requires the same therapy .
the problem is then how to identify the class of the exact @xmath14 rather than @xmath14 itself .
this model has also been studied by golovin et al .
@xcite to tackle the problem of erroneous tests responses in bayesian active learning .
the above examples can all be cast into the following general problem . + * the discrete function evaluation problem * ( dfep ) .
an instance of the problem is defined by a quintuple @xmath16 where @xmath17 is a set of objects , @xmath18 is a partition of @xmath19 into @xmath20 classes , @xmath21 is a set of tests , @xmath22 is a probability distribution on @xmath23 and @xmath24 is a cost function assigning to each test @xmath25 a cost @xmath26 a test @xmath27 , when applied to an object @xmath28 , incurs a cost @xmath29 and outputs a number @xmath30 in the set @xmath31 .
it is assumed that the set of tests is complete , in the sense that for any distinct @xmath32 there exists a test @xmath25 such that @xmath33 the goal is to define a testing procedure which uses tests from @xmath21 and minimizes the testing cost ( in expectation and/or in the worst case ) for identifying the class of an unknown object @xmath34 chosen according to the distribution @xmath35 the dfep can be rephrased in terms of minimizing the cost of evaluating a discrete function that maps points ( corresponding to objects ) from some finite subset of @xmath36 into values ( corresponding to classes ) , where an object @xmath37 corresponds to the point @xmath38 obtained by applying each test of @xmath21 to @xmath39 .
this perspective motivates the name we chose for the problem .
however , for the sake of uniformity with more recent work @xcite we employ the definition of the problem in terms of objects / tests / classes . +
* decision tree optimization . * any testing procedure can be represented by a _ decision tree _ , which is a tree where every internal node is associated with a test and every leaf is associated with a set of objects that belong to the same class .
more formally , a decision tree @xmath40 for @xmath41 is a leaf associated with class @xmath42 if every object of @xmath19 belongs to the same class @xmath42 .
otherwise , the root @xmath43 of @xmath40 is associated with some test @xmath44 and the children of @xmath43 are decision trees for the sets @xmath45 , where @xmath46 , for @xmath47 , is the subset of @xmath19 that outputs @xmath42 for test @xmath25 . given a decision tree @xmath40 , rooted at @xmath43 , we can identify the class of an unknown object @xmath34 by following a path from @xmath43 to a leaf as follows : first , we ask for the result of the test associated with @xmath43 when performed on @xmath34 ; then , we follow the branch of @xmath43 associated with the result of the test to reach a child @xmath48 of @xmath43 ; next , we apply the same steps recursively for the decision tree rooted at @xmath48 .
the procedure ends when a leaf is reached , which determines the class of @xmath34 .
we define @xmath49 as the sum of the tests cost on the root - to - leaf path from the root of @xmath40 to the leaf associated with object @xmath39 .
then , the _ worst testing cost _ and the _ expected testing cost _ of @xmath40 are , respectively , defined as @xmath50 @xmath51 figure [ fig : decisiontree0 ] shows an instance of the dfep and a decision tree for it .
the tree has worst testing cost @xmath52 and expected testing cost @xmath53 . ,
@xmath54 and @xmath55 .
letters and numbers in the leaves indicate , respectively , classes and objects . , title="fig:",width=415 ] [ fig : decisiontree0 ] -0.3 in * our results . *
our main result is an algorithm that builds a decision tree whose expected testing cost and worst testing cost are at most @xmath56 times the minimum possible expected testing cost and the minimum possible worst testing cost , respectively .
in other words , the decision tree built by our algorithm achieves simultaneously the best possible approximation achievable with respect to both the expected testing cost and the worst testing cost .
in fact , for the special case where each object defines a distinct class known as the _ identification problem_ both the minimization of the expected testing cost and the minimization of the worst testing cost do not admit a sub - logarithmic approximation unless @xmath57 as shown in @xcite and in @xcite , respectively .
in addition , in section [ sec : inapprox ] , we show that the same inapproximability results holds in general for the case of exactly @xmath20 classes for any @xmath58 it should be noted that in general there are instances for which the decision tree that minimizes the expected testing cost has worst testing cost much larger than that achieved by the decision tree with minimum worst testing cost . also there are instances where the converse happens .
therefore , it is reasonable to ask whether it is possible to construct decision trees that are efficient with respect to both performance criteria .
this might be important in practical applications where only an estimate of the probability distribution is available which is not very accurate .
also , in medical applications like the one depicted in @xcite , very high cost ( or equivalently significantly time consuming therapy identification ) might have disastrous / deadly consequences . in such cases , besides being able to minimize the expected testing cost , it is important to guarantee that the worst testing cost also is not large ( compared with the optimal worst testing cost ) . with respect to the minimization of the expected testing cost , our result improves upon the previous @xmath59 approximation shown in @xcite and @xcite , where @xmath60 is the minimum positive probability among the objects in @xmath19 . from the result in these papers
an @xmath56 approximation could be attained only for the particular case of uniform costs via a technique used in @xcite . from a high - level perspective
, our method closely follows the one used by gupta _
@xcite for obtaining the @xmath56 approximation for the expected testing cost in the identification problem .
both constructions of the decision tree consist of building a path ( backbone ) that splits the input instance into smaller ones , for which decision trees are recursively constructed and attached as children of the nodes in the path . a closer look , however , reveals that our algorithm is much simpler than the one presented in @xcite .
first , it is more transparently linked to the structure of the problem , which remained somehow hidden in @xcite where the result was obtained via an involved mapping from adaptive tsp .
second , our algorithm avoids expensive computational steps as the sviridenko procedure @xcite and some non - intuitive / redundant steps that are used to select the tests for the backbone of the tree .
in fact , we believe that providing an algorithm that is much simpler to implement and an alternative proof of the result in @xcite is an additional contribution of this paper .
* state of the art . *
the dfep has been recently studied under the names of class equivalence problem @xcite and group identification problem @xcite and long before it had been described in the excellent survey by moret @xcite .
both @xcite and @xcite give @xmath61 approximation algorithms for the version of the dfep where the expected testing cost has to be minimized and both the probabilities and the testing costs are non - uniform .
in addition , when the testing costs are uniform both algorithms can be converted into a @xmath62 approximation algorithm via kosaraju approach @xcite .
the algorithm in @xcite is more general because it addresses multiway tests rather than binary ones .
for the minimization of the worst testing cost , moshkov has studied the problem in the general case of multiway tests and non - uniform costs and provided an @xmath56-approximation in @xcite . in the same paper
it is also proved that no @xmath63-approximation algorithm is possible under standard the complexity assumption @xmath64 the minimization of the worst testing cost is also investigated in @xcite under the framework of covering and learning . the particular case of the dfep where each object belongs to a different class known as the _ identification problem_has been more extensively investigated @xcite . both the minimization of the worst and the expected testing cost do not admit a sublogarithmic approximation unless @xmath65 as proved by @xcite and @xcite . for the expected testing cost , in the variant with multiway tests , non uniform probabilities and non uniform testing costs , an @xmath66 approximation is given by guillory and blimes in @xcite .
@xcite improved this result to @xmath56 employing new techniques not relying on the generalized binary search ( gbs)the basis of all the previous strategies .
an @xmath56 approximation algorithm for the minimization of the worst testing cost for the identification problem has been given by arkin et .
@xcite for binary tests and uniform cost and by hanneke @xcite for case with mutiway tests and non - uniform testing costs . in the case of boolean functions
, the dfep is also known as stochastic boolean function evaluation ( sbfe ) , where the distribution over the possible assignments is a product distribution defined by assuming that variable @xmath67 has a given probability @xmath68 of being one independently of the value of the other variables .
another difference with respect to the dfep as it is presented here , is that in stochastic boolean function evaluation the common assumption is that the complete set of associations between the assignments of the variables and the value of the function is provided , directly or via a representation of the function , e.g. , in terms of its dnf or cnf .
the present definition of dfep considers the more general problem where only a sample of the boolean function is given and from this we want to construct a decision tree with minimum expected costs and that exactly fits the sample . results on the exact solution of the sbfe for different classes of boolean functions can be found in the survey paper @xcite . in a recent paper deshpande et al .
@xcite , provide a @xmath69-approximation algorithm for evaluating boolean linear threshold formulas and an @xmath70 approximation algorithm for the evaluation of cdnf formulas , where @xmath71 and @xmath72 is the number of clauses of the input cnf and @xmath72 is the number of terms of the input dnf .
the same result had been previously obtained by kaplan et al .
@xcite for the case of monotone formulas and uniform distribution ( in a slightly different setting ) .
both algorithms of @xcite are based on reducing the problem to stochastic submodular set cover introduced by golovin and krause @xcite and providing a new algorithm for this latter problem .
other special cases of the dfep like the evaluation of and/or trees ( a.k.a .
read - once formulas ) and the evaluation of game trees ( a central task in the design of game procedures ) are discussed in @xcite . in @xcite ,
charikar _ et al .
_ considered discrete function evaluation from the perspective of competitive analysis ; results in this alternative setting are also given in @xcite .
given an instance @xmath73 of the dfep , we will denote by @xmath74 ( @xmath75 ) the expected testing cost ( worst testing cost ) of a decision tree with minimum possible expected testing cost ( worst testing cost ) over the instance @xmath76 when the instance @xmath77 is clear from the context , we will also use the notation @xmath78 ( @xmath79 ) for the above quantity , referring only to the set of objects involved .
we use @xmath60 to denote the smallest non - zero probability among the objects in @xmath19 .
let @xmath80 be an instance of dfep and let @xmath81 be a subset of @xmath19 .
in addition , let @xmath82 and @xmath83 be , respectively , the restrictions of @xmath84 and @xmath24 to the set @xmath81 .
our first observation is that every decision tree @xmath40 for @xmath41 is also a decision tree for the instance @xmath85 .
the following proposition immediately follows .
[ prop : subadditivity ] let @xmath86 be an instance of the dfep and let @xmath81 be a subset of @xmath19 . then , @xmath87 and @xmath88 where @xmath89 is the restriction of @xmath77 to @xmath81 .
one of the measures of progress of our strategy is expressed in terms of the number of pairs of objects belonging to different classes which are present in the set of objects satisfying the tests already performed .
the following definition formalizes this concept of pairs for a given set of objects .
let @xmath90 be an instance of the dfep and @xmath91 we say that two objects @xmath92 constitute a pair of @xmath93 if they both belong to @xmath93 but come from different classes .
we denote by @xmath94 the number of pairs of @xmath95 in formulae , we have @xmath96 where for @xmath97 and @xmath98 @xmath99 denotes the number of objects in @xmath100 belonging to class @xmath101 as an example , for the set of objects @xmath19 in figure [ fig : decisiontree0 ] we have @xmath102 and the following set of pairs @xmath103 we will use @xmath34 to denote the initially unknown object whose class we want to identify .
let @xmath104 be a sequence of tests applied to identify the class of @xmath34 ( it corresponds to a path in the decision tree ) and let @xmath93 be the set of objects that agree with the outcomes of all tests in @xmath104 .
if @xmath105 , then all objects in @xmath93 belong to the same class , which must coincide with the class of the selected object @xmath34 .
hence , @xmath105 indicates the identification of the class of the object @xmath106 notice that @xmath34 might still be unknown when the condition @xmath105 is reached . for each test @xmath27 and for each @xmath107
, let @xmath108 be the set of objects for which the outcome of test @xmath25 is @xmath109 for a test @xmath110 the outcome resulting in the largest number of pairs is of special interest for our strategy .
we denote with @xmath111 the set among @xmath112 such that @xmath113 ( ties are broken arbitrarily ) .
we denote with @xmath114 the set of objects not included in @xmath115 i.e. , we define @xmath116 . whenever @xmath19 is clear from the context we use @xmath117 instead of @xmath114 .
given a set of objects @xmath19 , each test produces a tripartition of the pairs in @xmath19 : the ones with both objects in @xmath118 those with both objects in @xmath111 and those with one object in @xmath117 and one object in @xmath119 we say that the pairs in @xmath117 are _ kept _ by @xmath25 and the pairs with one object from @xmath117 and one object from @xmath111 are _ separated _ by @xmath120 we also say that a pair is _ covered _ by the test @xmath25 if it is either kept or separated by @xmath120 analogously , we say that a test @xmath25 covers an object @xmath39 if @xmath121 . for any set of objects @xmath122
the probability of @xmath123 is @xmath124
in this section , we describe our algorithm dectree and analyze its performance
. the concept of the separation cost of a sequence of tests will turn useful for defining and analyzing our algorithm . *
the separation cost of a sequence of tests . * given an instance @xmath73 of the dfep , for a sequence of tests @xmath125 we define the separation cost of @xmath104 in the instance @xmath126 denoted by @xmath127 as follows : fix an object @xmath128 if there exists @xmath129 such that @xmath130 then we set @xmath131 if @xmath132 for each @xmath133
then we set @xmath134 let @xmath135 denote the _ cost of separating @xmath9 in the instance @xmath77 by means of the sequence _
@xmath136 then , the _ separation cost of @xmath104 _ ( in the instance @xmath77 ) is defined by @xmath137 in addition , we define @xmath138 as the total cost of the sequence @xmath104 , i.e. , @xmath139 * lower bounds on the cost of an optimal decision tree for the dfep . *
we denote by @xmath140 the minimum separation cost in @xmath77 attainable by a sequence of tests in @xmath21 which covers all the pairs in @xmath19 and @xmath141 as the minimum total cost attainable by a sequence of tests in @xmath21 which covers all the pairs in @xmath142 the following theorem shows lower bounds on both the expected testing cost and the worst case testing cost of any instance @xmath73 of the dfep .
[ theo : lowerbound ] for any instance @xmath73 of the defp , it holds that @xmath143 and @xmath144 let @xmath40 be a decision tree for the instance @xmath77 .
let @xmath145 be the nodes in the root - to - leaf path in @xmath40 such that for each @xmath146 the node @xmath147 is on the branch stemming from @xmath148 which is associated with @xmath149 , and the leaf node @xmath150 is the child of @xmath151 associated with the objects in @xmath152 let @xmath153 .
abusing notation let us now denote with @xmath147 the test associated with the node @xmath147 so that @xmath154 is a sequence of tests .
in particular , @xmath104 is the sequence of tests performed according to the strategy defined by @xmath40 when the object @xmath34 whose class we want to identify , is such that @xmath155 holds for each test @xmath25 performed in the sequence .
notice that , by construction , @xmath104 is a sequence of tests covering all pairs of @xmath19 . _ claim .
_ for each object @xmath39 it holds that @xmath156 if for each @xmath157 we have that @xmath158 then it holds that @xmath159 conversely , let @xmath147 be the first test in @xmath104 for which @xmath160 therefore , we have that @xmath161 is a prefix of the root to leaf path followed when @xmath39 is the object chosen .
it follows that @xmath162 the claim is proved . in order to prove the first statement of the theorem , we let @xmath40 be a decision tree which achieves the minimum possible expected cost , i.e. , @xmath163 then
, we have @xmath164 in order to prove the second statement of the theorem , we let @xmath40 be a decision tree which achieves the minimum possible worst testing cost , i.e. , @xmath165 let @xmath28 be such that , for each @xmath166 it holds that @xmath167 then , by the above claim it follows that @xmath168 using ( [ eq : totcost - worstcase ] ) , we have @xmath169 the proof is complete
. the following subadditivity property will be useful .
[ prop : subadditivity ] let @xmath170 be a partition of the object set @xmath142 we have @xmath171 and @xmath172 , where @xmath173 and @xmath174 are , respectively , the minimum expected testing cost and the worst case testing cost when the set of objects is @xmath175 * the optimization of submodular functions of sets of tests . *
[ sec : submodularotimization ] let @xmath176 be an instance of the dfep .
a set function @xmath177 is submodular non - decreasing if for every @xmath178 and every @xmath179 , it holds that @xmath180 ( submodularity ) and @xmath181 ( non - decreasing ) .
it is easy to verify that the functions @xmath182 @xmath183 are non - negative non - decreasing submodular set functions . in words
, @xmath184 is the function mapping a set of tests @xmath185 into the number of pairs covered by the tests in @xmath185 .
the function @xmath186 , instead , maps a set of tests @xmath185 into the probability of the set of objects covered by the tests in @xmath185 .
let @xmath187 be a positive integer .
consider the following optimization problem defined over a non - negative , non - decreasing , sub modular function @xmath188 : @xmath189 in @xcite , wolsey studied the solution to the problem @xmath190 provided by algorithm [ greedy - wolsey ] below , called the adapted greedy heuristic .
( @xmath191 ) [ line : one ] [ line : two ] [ line : three ] [ line : four ] [ line : five ] [ line : six ] the following theorem summarizes results from @xcite[theorems 2 and 3 ] .
@xcite [ theo : wolsey ] let @xmath192 be the solution of the problem @xmath190 and @xmath193 be the set returned by algorithm [ greedy - wolsey ] .
moreover , let @xmath194 be the base of the natural logarithm and @xmath195 be the solution of @xmath196 then we have that @xmath197 .
moreover , if there exists @xmath198 such @xmath199 for each @xmath27 and @xmath198 divides @xmath200 then we have @xmath201 [ cor : wolsey ] let @xmath202 be the sequence of all the tests selected by adapted - greedy , i.e. , the concatenation of the two possible outputs in line 7 .
then , we have that the total cost of the tests in @xmath104 is at most @xmath203 and @xmath204 our algorithm for building a decision tree will employ this greedy heuristic for finding approximate solutions to the optimization problem @xmath190 over the submodular set functions @xmath184 and @xmath186 defined in ( [ eq : problemp ] ) .
we will show that algorithm [ algo : main ] attains a logarithmic approximation for dfep .
the algorithm consists of 4 blocks .
the first block ( lines 1 - 2 ) is the basis of the recursion , which returns a leaf if all objects belong to the same class @xmath205 . if @xmath206 , we have that @xmath207 and the algorithm returns a tree that consists of a root and two leaves , one for each object , where the root is associated with the cheapest test that separates these two objects .
clearly , this tree is optimal for both the expected testing cost and the worst testing cost . the second block ( line 3 ) calls procedure findbudget to define the budget @xmath187 allowed for the tests selected in the third and fourth blocks .
findbudget finds the smallest @xmath187 such that adapted - greedy(@xmath208 ) returns a set of tests @xmath185 covering at least @xmath209 pairs .
( @xmath210 ) [ b1start ] [ b1end ] [ line : budget ] [ b2start ] [ b3-start ] [ line : main - opt1 - 1 ] [ line : main - ui1 ] [ line : main - call-1 ] [ b3-end ] [ b4-start ] [ line : main - greedy2 ] [ line : main - ui2 ] [ line : main - call-2 ] [ b4-end ] [ line : main - call-3 ] ( @xmath211 ) the third ( lines 4 - 10 ) and the fourth ( lines 11 - 17 ) blocks are responsible for the construction of the backbone of the decision tree ( see fig .
2 ) as well as to call dectree recursively to construct the decision trees that are children of the nodes in the backbone .
the third block ( the * while * loop in lines 4 - 10 ) constructs the first part of the backbone ( sequence @xmath212 in fig .
2 ) by iteratively selecting the test that covers the maximum uncovered mass probability per unit of testing cost ( line 5 ) .
the selected test @xmath213 induces a partition @xmath214 on the set of objects @xmath215 , which contains the objects that have not been covered yet . in lines 7 and 8 ,
the procedure is recursively called for each set of this partition but for the one that is contained in the subset @xmath216 . with reference to figure 2
, these calls will build the subtrees rooted at nodes not in @xmath212 which are children of some node in @xmath212 .
similarly , the fourth block ( the * repeat - until * loop ) constructs the second part of the backbone ( sequence @xmath217 in fig .
2 ) by iteratively selecting the test that covers the maximum number of uncovered pairs per unit of testing cost ( line 12 ) . the line [ line : main - call-3 ] is responsible for building a decision tree for the objects that are not covered by the tests in the backbone .
we shall note that both the third and the fourth block of the algorithm are based on the adapted greedy heuristic of algorithm [ greedy - wolsey ] .
in fact , @xmath218 in line [ line : main - opt1 - 1 ] ( third block ) corresponds to @xmath219 in algorithm [ greedy - wolsey ] because , right before the selection of the @xmath71-th test , @xmath100 is the set of tests @xmath220 and @xmath221 .
thus , @xmath222 and @xmath223 so that @xmath224 a similar argument shows that @xmath225 in line [ line : main - greedy2 ] ( fourth block ) corresponds to @xmath226 in algorithm [ greedy - wolsey ] .
these connections will allow us to apply both theorem [ theo : wolsey ] and corollary [ cor : wolsey ] to analyze the cost and the coverage of these sequences .
while in the lowest - right gray subtree it is at most @xmath227 ( see the proof of theorem [ theo : main]).,title="fig:",width=302 ] [ fig : decisiontree ] -0.1 in let @xmath228 denote the sequence of tests obtained by concatenating the tests selected in the * while * loop and in the * repeat - until * loop of the execution of dectree over instance @xmath76 we delay to the next section the proof of the following key result .
[ theo : key ] let @xmath195 be the solution of @xmath229 and @xmath230 there exists a constant @xmath231 such that for any instance @xmath73 of the dfep , the sequence @xmath228 covers at least @xmath232 pairs , and it holds that @xmath233 and @xmath234 applying theorem [ theo : key ] to each recursive call of dectree we can prove the following theorem about the approximation guaranteed by our algorithm both in terms of worst testing cost and expected testing cost .
[ theo : main ] for any instance @xmath235 of the dfep , the algorithm dectree outputs a decision tree with expected testing cost at most @xmath236 and with worst testing cost at most @xmath237 .
for any instance @xmath126 let @xmath238 be the decision tree produced by the algorithm dectree .
first , we prove an approximation for the expected testing cost .
let @xmath239 be such that @xmath240 , where @xmath241 is the constant given in the statement of theorem [ theo : key ] .
let us assume by induction that the algorithm guarantees approximation @xmath242 , for the expected testing cost , for every instance @xmath243 on a set of objects @xmath93 with @xmath244 let @xmath245 be the set of instances on which the algorithm @xmath246 is recursively called in lines 8,15 and 18 .
we have that @xmath247 the first equality follows by the recursive way the algorithm dectree builds the decision tree .
inequality ( [ 3 ] ) follows from ( [ 1 ] ) by the subadditivity property ( proposition [ prop : subadditivity ] ) and simple algebraic manipulations .
the inequality in ( [ 3 - 1 ] ) follows by theorem [ theo : key ] together with theorem [ theo : lowerbound ] yielding @xmath248 the inequality ( [ 4 ] ) follows by induction ( we are using @xmath249 to denote the number of pairs of instance @xmath243 ) . to prove that the inequality in ( [ 6 ] ) holds we have to argue that every instance @xmath250 has at most @xmath251 pairs .
let @xmath252 as in the lines 8 and 15 .
first we show that the number of pairs of @xmath253 is at most @xmath254 .
we have @xmath255 and @xmath216 is the set with the maximum number of pairs in the partition @xmath256 , induced by @xmath213 on the set @xmath19 .
it follows that @xmath257 now it remains to show that the instance @xmath243 , recursively called , in line 18 has at most @xmath227 pairs .
this is true because the number of pairs of @xmath243 is equal to the number of pairs not covered by @xmath228 which is bounded by @xmath258 by theorem [ theo : key ] .
now , we prove an approximation for the worst testing cost of the tree @xmath238 .
let @xmath259 be such that @xmath260 .
let us assume by induction that the worst testing cost of @xmath261 is at most @xmath262 for every instance @xmath243 on a set of objects @xmath93 with @xmath263 we have that @xmath264 inequality ( [ 8 ] ) follows from the subadditivity property ( proposition [ prop : subadditivity ] ) for the worst testing cost .
the inequality ( [ 9 ] ) follows by theorem [ theo : lowerbound ] .
the inequality ( [ 10 ] ) follows from theorem [ theo : key ] , the induction hypothesis ( we are using @xmath249 to denote the number of pairs of instance @xmath243 ) and from the fact mentioned above that every instance in @xmath245 has at most @xmath265 pairs . since @xmath266 it follows that the algorithm provides an @xmath56 approximation for both the expected testing cost and the worst testing cost .
the previous theorem shows that algorithm dectree provides simultaneously logarithmic approximation for the minimization of expected testing cost and worst testing cost .
we would like to remark that this is an interesting feature of our algorithm . in this respect ,
let us consider the following instance of the dfep : let @xmath267 ; @xmath268 , for @xmath269 and @xmath270 ; the set of tests is in one to one correspondence with the set of all binary strings of length @xmath271 so that the test corresponding to a binary string @xmath272 outputs @xmath273 for object @xmath274 if and only if the @xmath42th bit of @xmath272 is 0(1 ) .
moreover , all tests have unitary costs .
this instance is also an instance of the problem of constructing an optimal prefix coding binary tree , which can be solved by the huffman s
algorithm @xcite .
let @xmath275 and @xmath276 be , respectively , the decision trees with minimum expected cost and minimum worst testing cost for this example . using huffman s algorithm , it is not difficult to verify that @xmath277 and @xmath278 .
in addition , we have that @xmath279 .
this example shows that the minimization of the expected testing cost may result in high worst testing cost and vice versa the minimization of the worst testing cost may result in high expected testing cost .
clearly , in real situations presenting such a dichotomy , the ability of our algorithm to optimize simultaneously both measures of cost might provide a significant gain over strategies only guaranteeing competitiveness with respect to one measure .
we now return to the proof of theorem [ theo : key ] for which will go through three lemmas .
[ lemma : length ] for any instance @xmath73 of the dfep , the value @xmath187 returned by the procedure findbudget@xmath280 satisfies @xmath281 .
let us consider the problem @xmath190 in equation ( [ eq : problemp ] ) with the function @xmath184 that measures the number of pairs covered by a set of tests .
let @xmath282 be the number of pairs covered by the solution constructed with adapted - greedy when the budget the righthand side of equation ( [ eq : problemp])is @xmath9 . by construction
, findbudget finds the smallest @xmath187 such that @xmath283 .
let @xmath284 be a sequence that covers all pairs in @xmath19 and that satisfies @xmath285 .
arguing by contradiction we can show that @xmath286 suppose that this was not the case , then @xmath284 would be the sequence which covers @xmath287 pairs using a sequence of tests of total cost not larger than some @xmath288 by theorem 2 , the procedure adapted - greedy provides an @xmath289-approximation of the maximum number of pairs covered with a given budget .
therefore , when run with budget @xmath290 adapted - greedy is guaranteed to produce a sequence of total cost @xmath291 which covers at least @xmath209 pairs . however , by the minimality of @xmath187 it follows that such a sequence does not exist .
since this contradiction follows by the hypothesis @xmath292 it must hold that @xmath293 as desired .
given an instance @xmath126 for a sequence of tests @xmath294 and a real @xmath295 , let @xmath296 be the separation cost of @xmath104 when every non - covered object is charged @xmath297 , that is , @xmath298 the proofs of the following technical lemma is deferred to the appendix .
[ lemma : key2 ] let @xmath212 be the sequence obtained by concatenating the tests selected in the * while * loop of algorithm [ algo : main ] . then
, @xmath299 and @xmath300 where @xmath301 is a positive constant and @xmath187 is the budget calculated at line 3 .
[ lemma : key3 ] the sequence @xmath228 covers at least @xmath302 pairs and it holds that @xmath303 the sequence @xmath228 can be decomposed into the sequences @xmath212 and @xmath217 , that are constructed , respectively , in the * while * and * repeat - until * loop of the algorithm dectree ( see also fig .
it follows from the definition of @xmath187 that there is a sequence of tests , say @xmath104 , of total cost not larger than @xmath187 that covers at least @xmath209 pairs for instance @xmath77 .
let @xmath304 be the number of pairs of instance @xmath77 covered by the sequence @xmath212 .
thus , the tests in @xmath104 , that do not belong to @xmath212 , cover at least @xmath305 pairs in the set @xmath306 of objects not covered by @xmath212 .
the sequence @xmath217 coincides with the concatenation of the two possible outputs of the procedure adapted - greedy(@xmath307 ( algorithm 1 ) , when it is executed on the instance defined by : the objects in @xmath215 ( those not covered by @xmath212 ) ; the tests that are not in @xmath308 the submodular set function @xmath309 and bound @xmath310 by corollary [ cor : wolsey ] , we have that @xmath311 and @xmath217 covers at least @xmath312 uncovered pairs . therefore , since @xmath313 altogether
, we have that @xmath228 covers at least @xmath314 pairs and @xmath303 the proof of theorem [ theo : key ] will now follow by combining the previous three lemmas .
* proof of theorem [ theo : key ] .
* first , it follows from lemma [ lemma : key3 ] that @xmath228 covers at least @xmath302 pairs . to prove that @xmath315
, we decompose @xmath228 into @xmath316 and @xmath317 , the sequences of tests selected in the * while * and in the * repeat - until * loop of algorithm [ algo : main ] , respectively . for @xmath318 ,
let @xmath319 .
in addition , let @xmath320 be the set of objects which are not covered by the tests in @xmath321 thus , @xmath322 where the last inequality follows from lemma [ lemma : key2 ] .
it remains to show that @xmath323 .
this inequality holds because lemma [ lemma : key3 ] assures that @xmath324 and lemma [ lemma : length ] assures that @xmath325 .
the proof is complete .
let @xmath326 be a set of @xmath271 elements and @xmath327 be a family of subsets of @xmath215 .
the minimum set cover problem asks for a family @xmath328 of minimum cardinality such that @xmath329 .
it is known that no sub logarithmic approximation is achievable for the minimum set cover problem under the standard assumption that @xmath330 more precisely , by the result of raz and safra @xcite it follows that there exists a constant @xmath331 such that no @xmath332-approximation algorithm for the minimum set cover problem exists unless @xmath333 @xcite .
we will show that an @xmath334 approximation algorithm for the minimization of the expected testing cost dfep with exact @xmath335 classes for @xmath336 , implies that the same approximation can be achieved for the minimum set cover problem .
this implies that one can not expect to obtain a sublogarithmic approximation for the dfep unless @xmath65 .
the reduction we present can also be used to show the same inapproximability result for the minimization of the worst testing cost version of the dfep .
given an instance @xmath337 for the minimum set cover problem as defined above , we construct an instance @xmath338 for the dfep as follows : the set of objects is @xmath339 .
the family of classes @xmath340 is defined as follows : all the objects of @xmath215 belong to class @xmath341 while the object @xmath342 , for @xmath343 , belongs to class @xmath344 . notice that @xmath345 and the objects of @xmath215 belong to the same class . in order to define the set of tests @xmath346
we proceed as follows : for each set @xmath347 we create a test @xmath348 such that @xmath348 has value @xmath349 for the objects in @xmath350 and value 1 for the remaining objects .
in addition , we create a test @xmath351 which has value @xmath349 for objects in @xmath215 and value @xmath352 for object @xmath342 ( @xmath353 ) . for our later purposes ,
we notice here that the test @xmath351 can not distinguish between @xmath345 and the elements in @xmath215 .
each test has cost 1 , i.e. , the cost assignment * c * is given by @xmath354 for each @xmath355 finally , we set the probability of @xmath356 to be equal to @xmath357 and the probability of the other objects equal to @xmath358 , for some fixed @xmath359 let @xmath360 be the decision tree with minimum expected testing cost for @xmath361 and let @xmath362 be a minimum set cover for instance @xmath363 , where @xmath364 .
we first argue that @xmath365 .
in fact , we can construct a decision tree @xmath40 by putting the test @xmath366 associated with @xmath367 in the root of the tree , then the test @xmath368 associated with @xmath369 as the child of @xmath366 and so on . notice that , for @xmath370 we have that @xmath371 has two children , one is @xmath372 and the other is a leaf mapping to the class @xmath373 as for @xmath374 one of its children in again a leaf mapping to @xmath341 , the other child is set to the test @xmath351 , whose children are all leaves .
the expected testing cost of @xmath40 can be upper bounded by @xmath375 since we have @xmath376 for any @xmath377 and @xmath378 for any @xmath379 . on the other hand ,
let @xmath40 be a decision tree for @xmath361 and let @xmath380 be the path from the root of @xmath40 to the leaf where the object @xmath356 lies .
it is easy to realize that the subsets associated with the tests on this path cover all the elements in @xmath215in fact these tests separate @xmath356 from all the other objects from @xmath381 let @xmath382 be the solution to the set cover problem provided by the sets associated with the tests on the path @xmath383 we have that @xmath384 now assume that there is an algorithm that for any instance @xmath73 of the dfep can guarantee a solution with approximation @xmath385 for some @xmath386 therefore , given an instance @xmath337 for set cover we can use this algorithm on the transformed instance @xmath361 defined above , where @xmath387 we obtain a decision tree @xmath40 for @xmath361 such that @xmath388 where we upper bound @xmath389 and @xmath390 from @xmath40 , as seen above we can construct a solution @xmath382 for the set cover problem such that @xmath391 hence , it would follow that @xmath382 is an approximate solution for the set cover instance satisfying : @xmath392 which by the result of @xcite is not possible unless @xmath393 the same construction can be used for analyzing the case of the worst testing cost , in which case we have that ( [ inapprox:1 ] ) becomes @xmath394 and ( [ inapprox:2 ] ) becomes @xmath395 leading to the inapproximability of the dfep w.r.t .
the worst testing cost within a factor of @xmath396 for any @xmath397 notice that an analogous result regarding the worst testing cost had been previously shown by moshkov @xcite based on the result of feige @xcite .
thus , we have the following theorem [ theo : inappr ] both the minimization of the worst case and the expected case of the dfep do nt admit an @xmath63 approximation unless @xmath398
we presented a new algorithm for the discrete function evaluation problem , a generalization of the classical bayesian active learning also studied under the names of equivalence class determination problem @xcite and group based active query selection problem of @xcite .
our algorithm builds a decision tree which asymptotically matches simultaneously for the expected and the worst testing cost the best possible approximation achievable under standard complexity assumptions @xmath330 this way , we close the gap left open by the previous @xmath59 approximation for the expected cost shown in@xcite and @xcite , where @xmath60 is the minimum positive probability among the objects in @xmath19 and in addition we show that this can be done with an algorithm that guarantees the best possible approximation also with respect to the worst testing cost . with regards to the broader context of learning , given a set of samples labeled according to an unknown function , a standard task in machine learning is to find a good approximation of the labeling function ( hypothesis ) . in order to guarantee that the hypothesis chosen has some generalization power w.r.t . to the set of samples
, we should avoid overfitting .
when the learning is performed via decision tree induction this implies that we shall not have leaves associated with a small number of samples so that we end up with a decision tree that have leaves associated with more than one label .
there are many strategies available in the literature to induce such a tree . in the problem considered in this paper
our aim is to over - fit the data because the function is known a priori and we are interested in obtaining a decision tree that allows us to identify the label of a new sample with the minimum possible cost ( time / money ) .
the theoretical results we obtain for the `` fitting '' problem should be generalizable to the problem of approximating the function . to this aim
we could employ the framework of covering and learning from @xcite along the following lines : we would interrupt the recursive process in algorithm 2 through which we construct the tree as soon as we reach a certain level of learning ( fitting ) w.r.t .
the set of labeled samples .
then , it remains to show that our decision tree is at logarithmic factor of the optimal one for that level of learning .
this is an interesting direction for future research .
30 adler , m. and heeringa , b. approximating optimal binary decision trees .
approx / random 08 , pp . 19 , 2008 .
allen , s.r . , hellerstein , l. , kletenik , d , and nlyurt , t. evaluation of dnf formulas .
in _ proc . of isaim 2014_. arkin , e. m. , meijer , h. , mitchell , j.s.b . , rappaport , d. and skiena , s.s .
decision trees for geometric models . in _ proc . of scg 93
_ , pp . 369378 , 1993 .
bellala , g. , bhavnani , s. k. , and scott , c. group - based active query selection for rapid diagnosis in time - critical situations .
_ ieee trs .
theor . _ , 580 ( 1):0 459478 , 2012 .
chakaravarthy , v. t. , pandit , v. , roy , s. , awasthi , p. , and mohania , m. decision trees for entity identification : approximation algorithms and hardness results . in _ proc .
pods 07 _ , pp . 5362 , 2007 .
chakaravarthy , v. t. , pandit , v. , roy , s. , and sabharwal , y. approximating decision trees with multiway branches . in _ proc .
icalp 09 _ , pp . 210221 , 2009 .
charikar , m. , fagin , r. , guruswami , v. , kleinberg , j. m. , raghavan , p. , and sahai , a. query strategies for priced information . _ journal of computer and system sciences _ , 640 ( 4):0 785819 , 2002 .
cicalese , f. and laber , e. s. on the competitive ratio of evaluating priced functions .
_ j. acm _ , 580 ( 3):0 9 , 2011 . cicalese , f. , jacobs , t. , laber , e. , and molinaro , m. on greedy algorithms for decision trees . in _ proc . of isaac10 _ , 2010 .
cormen , t. h. , leiserson , c. e. , rivest , r. l. , and stein , c. _ introduction to algorithms_. mit press , cambridge , ma , 2001 .
dasgupta , s. analysis of a greedy active learning strategy . in _ nips04 _ , 2004 .
deshpande , a. , hellerstein , l. , and kletenik , d .. approximation algorithms for stochastic boolean function evaluation and stochastic submodular set cover . in _ proc . of soda
2014 _ , pp . 14531467 , 2014 . a threshold of @xmath399 for approximating set cover
_ journal of acm _ 45 ( 1998 ) 634652 .
garey , m. optimal binary identification procedures .
_ siam journal on applied mathematics _
, 23(2):173186 , 1972 .
golovin , d. , krause , a. , and ray , d. near - optimal bayesian active learning with noisy observations . in _ proc . of nips10
_ , pp . 766774 , 2010 .
golovin , d. , krause , a .. adaptive submodularity : theory and applications in active learning and stochastic optimization
. _ journal of artificial intelligence research _ , vol .
42 , pages 427 - 486 , 2011 .
greiner , r. , howard , r. , jankowska , m. , and malloy , m. finding optimal satisfiscing strategies for and - or trees .
_ artificial intelligence _ , 1700 ( 1):0 1958 , 2005 .
guillory , a. and bilmes , j. average - case active learning with costs . in _ proc . of alt09
_ , pp .
141155 , 2009 . guillory , a. and bilmes , j. interactive submodular set cover . in _ proc .
icml10 _ , pp . 415422 .
guillory , a. and bilmes , j. simultaneous learning and covering with adversarial noise .
_ icml11 _ , pp .
gupta , a. , nagarajan , v. , and ravi , r. approximation algorithms for optimal decision trees and adaptive tsp problems . in _ proc .
icalp10 _ , pp . 690701 , 2010 .
hanneke , s. the cost complexity of interactive learning .
_ unpublished _ , 2006 .
hyafil , l. and rivest , r. l. constructing optimal binary decision trees is np - complete . _
inf . process .
_ , 50 ( 1):0 1517 , 1976 .
kaplan , h. , kushilevitz , e. , and mansour , y .. learning with attribute costs . in _ proc .
of stoc 2005 _ , pp . 356365 , 2005 .
kosaraju , s. rao , przytycka , teresa m. , and borgstrom , ryan s. on an optimal split tree problem . in _ proc . of wads
99 _ , pp . 157168 , 1999 .
laber , e. s. and nogueira , l. t. on the hardness of the minimum height decision tree problem . _
discrete appl .
_ , 144:0 209212 , 2004 .
larmore , l. l. and hirschberg , d. s. a fast algorithm for optimal length - limited huffman codes .
_ jacm : journal of the acm _ , 37 , 1990 .
ecision trees and diagrams .
_ acm computing surveys _ , pp .
593623 , 1982 .
moshkov , j.m .
approximate algorithm for minimization of decision tree depth . in _ proc . of 9th intl .
conf . on rough sets , fuzzy sets , data mining , and granular computing _ , pp . 611 - 614 , 2003 .
moshkov , j.m .
greedy algorithm with weights for decision tree construction .
, 104 : 285292 , 2010 .
nemhauser , g. , wolsey , l. , and fisher , m. l. an analysis of approximations for maximizing submodular set functions - i .
_ math . programming _ , 14:0 265294 , 1978 .
nevmyvaka , y. , feng , y. , and kearns , m .. reinforcement learning for optimized trade execution . in _ proc . of icml06
_ , pp .
673680 , 2006 . and
safra , s .. a sub - constant error - probability low - degree test , and sub - constant error - probability pcp characterization of np . ,
acm , 1997 , pp .
475484 .
saks , m. e. and wigderson , a. probabilistic boolean decision trees and the complexity of evaluating game trees . in _ proc . of focs86
_ , pp . 2938 , 1986 .
sviridenko , m. a note on maximizing a submodular set function subject to a knapsack constraint .
_ operations research letters _ , 320 ( 1):0 41 43 , 2004 .
tarsi , m. optimal search on some game trees .
_ journal of the acm _ , 300 ( 3):0 389396 , july 1983 .
nlyurt , t. sequential testing of complex systems : a review .
, 142(1 - 3):189205 , 2004 .
wolsey , l. maximising real - valued submodular functions .
_ math . of operation research _ , 7(3):410425 , 1982 .
* lemma [ lemma : key2 ] . * _ let @xmath212 be the sequence obtained by concatenating the tests selected in the * while * loop of algorithm 2 .
then , @xmath299 and @xmath300 where @xmath301 is a positive constant and @xmath187 is the budget calculated at line [ line : budget ] . _
clearly , the algorithm 2 in the * while * loop constructs a sequence @xmath212 such that @xmath400 in order to prove the second inequality in the statement of the lemma , it will be convenient to perform the analysis in terms of a variant of our problem which is explicitly defined with respect to the separation cost of a sequence of tests .
we call this new problem the _ pair separation problem _ ( psp ) : the input to the psp , as in the dfep , is a 5-uple @xmath401 , where @xmath17 is a set of objects , @xmath18 is a partition of @xmath19 into @xmath20 classes , @xmath402 is a family of subsets of @xmath19 , @xmath22 is a probability distribution on @xmath23 and @xmath24 is a cost function assigning to each @xmath403 a cost @xmath404 the only difference between the input of these problems is that the set of tests @xmath21 in the input of dfep is replaced with a family @xmath3 of subsets of @xmath19 .
we say that @xmath403 covers an object @xmath39 iff @xmath405 .
moreover , we say that @xmath403 covers a pair of objects @xmath406 if at least one of the conditions hold : ( i ) @xmath405 or ( ii ) @xmath407 .
we say that a pair @xmath406 is covered by a sequence of tests if some test in the sequence covers @xmath406 .
the separation cost of a sequence @xmath408 in the instance @xmath409 of psp is given by : @xmath410 the _ pair separation problem _ consists of finding a sequence of subsets of @xmath3 with minimum separation cost , @xmath411 , among those sequences that cover all pairs in @xmath19 .
an instance @xmath235 of the dfep induces an instance @xmath412 of the psp where @xmath413 and for every test @xmath27 we have a corresponding subset @xmath414 such that @xmath415 .
thus , in our discussion we will use the term test @xmath416 to refer to a subset @xmath417 . in the body of this paper
we implicitly work with the instance of the psp induced by the input instance of the dfep .
it is easy to realize that @xmath418 .
in addition , @xmath419 where @xmath420 is the sequence obtained from @xmath212 when every @xmath421 is replaced with @xmath422 .
thus , in order to establish the lemma it suffices to prove that @xmath423 it is useful to observe that @xmath420 is equal to the sequence @xmath424 returned by procedure greedypsp in algorithm [ alg : appendix ] when it is executed on the instance @xmath425 .
this algorithm corresponds to lines 4,5,9 and 10 of the while loop of algorithm [ algo : main ] . in algorithm
[ alg : appendix ] , the greedy criteria consists of choosing the test @xmath416 that maximizes the ratio @xmath426 this is equivalent to the maximization of @xmath427 defining the greedy choice in algorithm [ algo : main ] .
( @xmath428 : instance of psp , @xmath187:budget ) ) ( * ) [ alg : appendix ] the proof consists of the following steps : 1 .
we construct an instance @xmath429 of the psp from @xmath409 2 .
we prove that the optimal separation cost for @xmath243 is no larger than the optimal one for @xmath409 , that is , @xmath430 .
we prove that separation cost @xmath431 of any sequence of tests @xmath432 returned by the above pseudo - code on the instance @xmath433 is at a constant factor of @xmath434 , that is , @xmath431 is @xmath435 .
we prove that there exists a sequence of tests @xmath436 possibly returned by greedypsp when executed on the instance @xmath437 such that @xmath438 . by chaining these inequalities ,
we conclude that @xmath439 is @xmath440 the steps ( ii ) , ( iii ) and ( iv ) are proved in claims 1,2 and 3 , respectively .
we start with the construction of instance @xmath243 .
_ construction of instance @xmath243 .
_ for every test @xmath403 , we define @xmath441 . the instance @xmath442 is constructed from @xmath443 as follows .
let @xmath444 . for each @xmath445
we add @xmath446 objects to @xmath81 , each of them with probability @xmath447 and with class equal to that of @xmath39 .
if an object @xmath448 is added to set @xmath81 due to @xmath39 , we say that @xmath448 is generated from @xmath39 . for every test @xmath403
we add @xmath449 tests to the set @xmath450 , each of them with cost @xmath451 .
if a test @xmath452 is added to set @xmath450 due to @xmath416 , we say that @xmath452 is generated from @xmath416 .
it remains to define to which subset of @xmath81 each test @xmath453 corresponds to . if @xmath454 then @xmath455 for every @xmath448 generated from @xmath39 and every @xmath452 generated from @xmath416 .
let @xmath456 be the set of tests that contains the object @xmath28 .
note that the number of tuples @xmath457 , where @xmath458 is a test generated from @xmath459 is @xmath460 .
thus , we create a one to one correspondence between these tuples and the numbers in the set @xmath461 . for a test @xmath462 , generated from @xmath463 , let @xmath464 be the set of numbers that correspond to the tuples that includes @xmath465 .
note that @xmath466 in addition , we associate each object @xmath467 , generated from @xmath39 , with a number @xmath468 in a balanced way so that each number in @xmath469 is associated with @xmath470 objects .
thus , a test @xmath462 , generated from @xmath463 , covers an object @xmath448 generated from @xmath39 if and only if @xmath471 .
for the instance @xmath243 we have the following useful properties : * if @xmath472 covers object @xmath28 then each test @xmath473 , generated from @xmath416 , covers exactly @xmath474 objects generated from @xmath39 .
moreover , each object generated from @xmath39 is covered by exactly one test generated from @xmath416 .
* if a set of tests @xmath475 covers all pairs of @xmath243 then the set @xmath476 all tests generated from @xmath416 belong to @xmath477 covers all pairs of @xmath409 .
property ( a ) holds because a test @xmath465 generated by @xmath416 is associated with @xmath478 numbers in @xmath469 and to each number in @xmath469 we have @xmath479 objects associated with . to see that property ( b )
holds , let us assume that @xmath480 covers all pairs of the instance @xmath243 and @xmath93 does not cover a pair @xmath481 .
let @xmath482 and @xmath483 be the set of tests that covers @xmath484 and @xmath485 , respectively .
the fact that @xmath93 does not cover @xmath481 implies that @xmath486 so that for each @xmath487 , there is a test @xmath488 , generated from @xmath489 , that does not belong to @xmath480 .
similarly , for each @xmath490 , there is a test @xmath491 , generated from @xmath492 , that does not belong to @xmath480 .
let @xmath493 be an object , generated from @xmath484 , that is mapped , via function @xmath494 , into the number in @xmath495 that corresponds to the tuple @xmath496 .
moreover , let @xmath497 be an object , generated from @xmath485 , that is mapped , via function @xmath494 , into the number in @xmath498 that corresponds to the tuple @xmath499 .
the pair @xmath500 is not covered by @xmath480 , which is a contradiction . * claim 1 . * the optimal separation cost for @xmath243 is no larger than the optimal separation cost for @xmath501 i.e. , @xmath502 given a sequence @xmath503 for @xmath409 that covers all @xmath287 pairs we can obtain a sequence @xmath504 for @xmath243 by replacing each test @xmath505 with the @xmath506 tests in @xmath450 that were generated from @xmath507 , each of which has cost @xmath508 it is easy to see that @xmath504 covers all the pairs in @xmath243 and the separation cost of @xmath504 is not larger than that of @xmath509 .
this establishes our claim .
now let @xmath432 be a sequence of tests returned by procedure greedypsp in algorithm [ alg : appendix ] when it is executed on the instance @xmath433 .
* claim 2 . *
the separation cost of the sequence @xmath432 is at most a constant factor of that of @xmath510 , which is the sequence of tests with minimum separation cost among all sequences of tests covering all the pairs , for the instance @xmath511 i.e. , @xmath512 for some constant @xmath513 let @xmath514 ( resp .
@xmath515 ) be the sum of the probabilities of the objects covered by the first @xmath516 tests in @xmath432 ( resp .
@xmath517 ) .
in particular , we have @xmath518 in addition , let @xmath123 be the sum of the probabilities of all objects in @xmath81 . notice that , with the above notation , we can rewrite the separation cost of the sequence @xmath519 as @xmath520 let @xmath521 be such that @xmath522 , where @xmath187 is the budget in the statement of the lemma . for @xmath523 , let @xmath524 and @xmath525 .
furthermore , let @xmath526}$ ] be the sum of the probabilities of the objects covered by the first @xmath527 tests of @xmath432 .
in formulae , @xmath526 } = p \left ( \bigcup_{k=1}^{i_j } x'_k \right).$ ] analogously , let @xmath526}_*$ ] be the sum of the probabilities of the objects covered by the first @xmath528 tests in @xmath529 in formulae , @xmath526 } _ * = p\left(\bigcup_{k=1}^{i^*_j } x^*_k \right).$ ] for the sake of definiteness , we set @xmath530 and @xmath531 } = p^{[-1]}_*=0 $ ] then , we have @xmath532 } ) \leq \sum_{j=0}^{\ell-1 } 2^{j } ( q - p^{[j-1 ] } ) \\ & \leq & q+ \sum_{j=1}^{\ell-1 } 2^{j } ( q - p^{[j-1 ] } ) \leq q+ \sum_{j=0}^{\ell-2 } 2^{j+1 } ( q - p^{[j]}),\end{aligned}\ ] ] where the first inequality holds because @xmath533 and the second one holds because @xmath534}= p_{i_{j-1 } } $ ] for @xmath535 .
we now devise a lower bound on the separation cost of @xmath529 for this , we first note that the length @xmath536 of @xmath517 is at least @xmath537 , for otherwise the property ( b ) of instance @xmath243 would guarantee the existence of a sequence of tests of total cost smaller than @xmath187 that covers all pairs for instance @xmath409 ( and for the instance @xmath77 of the dfep as well ) , which contradicts lemma [ lemma : length ] .
therefore , we can lower bound the the separation cost of the sequence @xmath517 as follows : @xmath538 } ) \geq \frac{3q}{4 } + \frac{1}{2 } \sum_{j=1}^{\ell-1 } 2^{j } ( q - p_*^{[j ] } ) \label{eq:27may2}\end{aligned}\ ] ] the inequality in ( [ eq:27may1 ] ) follows from ( [ eq:27may0 ] ) by considering in the summation on the right hand side of ( [ eq:27may1 ] ) only the first @xmath539 tests . the term @xmath540 in the first inequality ( [ eq:27may2 ] ) is the contribution of the first two tests of the sequence @xmath517 to the separation cost . to prove that @xmath541 in the last inequality , we note that that @xmath542 because the probability covered by the first test @xmath543 of sequence @xmath544 is @xmath545 , where @xmath416 is the test that generates @xmath543 . in the last inequality we used the fact that @xmath546 for all @xmath403
let @xmath547 be the set of objects covered by the sequence of tests @xmath548 , which is the prefix of length @xmath71 of the sequence of tests @xmath432 .
we shall note that for @xmath549 , the subsequence @xmath550 of @xmath432 coincides with the sequence of tests constructed through the execution of adapted - greedy over the instance ( @xmath551 ) , where * @xmath552 is a set of tests , all of them with cost @xmath553 * the function @xmath186 maps a set of tests into the probability of the objects in @xmath554 that are covered by the tests in the set ; * @xmath555 since the set @xmath556 is a feasible solution for this instance , it follows from theorem 2 that @xmath557 , where @xmath558 by setting @xmath559 and @xmath560 we get that @xmath561 } - p^{[j-1 ] } \geq \hat{\alpha } ( p_*^{[j-1 ] } - p^{[j-1]}).\ ] ] it follows that @xmath562 } \leq \hat{\alpha}(q - p_*^{[j-1 ] } ) + ( 1-\hat{\alpha } ) ( q - p^{[j-1]}).\ ] ] thus , setting @xmath563}),\ ] ] which is the upper bound we derived on the separation cost of the sequence @xmath432 , we have @xmath564})\\ & \leq & q + \hat{\alpha } \sum_{j=0}^{\ell-2 } 2^{j+1 } ( q - p_*^{[j-1 ] } ) + ( 1-\hat{\alpha } ) \sum_{j=0}^{\ell-2 } 2^{j+1 } ( q - p^{[j-1]})\\ & = & q + 2\hat{\alpha}q + \hat{\alpha}\sum_{j=1}^{\ell-2 } 2^{j+1 } ( q - p_*^{[j-1 ] } ) + 2(1-\hat{\alpha})q + ( 1-\hat{\alpha } ) \sum_{j=1}^{\ell-2 } 2^{j+1 } ( q - p^{[j-1]})\\ & = & q + 2 q + 2\hat{\alpha } \sum_{j=0}^{\ell-3 } 2^{j+1 } ( q - p_*^{[j ] } ) + 2 ( 1-\hat{\alpha } ) \sum_{j=0}^{\ell-3 } 2^{j+1 } ( q - p^{[j]})\\ & \leq & q + 2 q + 4 \hat{\alpha } q + 2\hat{\alpha } \sum_{j=1}^{\ell-3 } 2^{j+1 } ( q - p_*^{[j ] } ) + 2 ( 1-\hat{\alpha } ) \sum_{j=0}^{\ell-3 } 2^{j+1 } ( q - p^{[j]})\\ & = & ( 1 - 2(1-\hat{\alpha})+2 + 4\hat{\alpha } ) q + 4\hat{\alpha } \sum_{j=1}^{\ell-3 } 2^{j } ( q - p_*^{[j ] } ) + 2(1-\hat{\alpha})\left(q + \sum_{j=0}^{\ell-3 } 2^{j+1 } ( q - p^{[j]})\right)\\ & \leq & ( 1 + 6 \hat{\alpha } ) q + 4\hat{\alpha } \sum_{j=1}^{\ell-1 } 2^{j } ( q - p_*^{[j ] } ) + 2(1-\hat{\alpha})u\\ & \leq & ( 8\hat{\alpha}+4/3 ) \ , sepcost(i ' , { \bf x}^ * ) + 2(1-\hat{\alpha } ) u,\end{aligned}\ ] ] where the last inequality follows from equation [ eq:27may2 ] .
thus , we obtain @xmath565 for the last claim let @xmath420 be the sequence obtained by greedypsp ( algorithm [ alg : appendix ] ) when it is executed on instance @xmath425 .
* claim 3 .
* there exists an execution of procedure greedypsp ( algorithm [ alg : appendix ] ) on instance @xmath433 which returns a sequence @xmath436 satisfying @xmath438 .
let @xmath566 be the @xmath42-th test of sequence @xmath420 and let @xmath567 be the first test of @xmath568 that is not the test which maximizes @xmath569 among all the tests in @xmath3 in line ( * ) of algorithm [ alg : appendix ] .
note that @xmath567 is chosen by algorithm [ alg : appendix ] rather than @xmath570 , the test which maximizes @xmath571 , because @xmath570 has cost larger than remaining budget @xmath572 .
the case where @xmath567 does not exist is easier to handle and will be discussed at the end of the proof . because @xmath573 is a prefix of @xmath420
we have @xmath574 thus , to establish the claim it suffices to show that @xmath575 where @xmath436 is a possible output of greedypsp ( algorithm [ alg : appendix ] ) on instance @xmath576 for @xmath577 , let @xmath578 } = \langle z^{[j]}_1 , \dots , z^{[j]}_{n(x_j^a)}\rangle$ ] be a sequence of tests defined by some permutation of the @xmath579 tests in @xmath450 , generated from @xmath580 let @xmath572 and @xmath581 } = \langle z^{[r+1]}_1 , \dots , z^{[r+1]}_{2z}\rangle$ ] be a sequence of @xmath582 of the @xmath583 tests in @xmath450 , generated from @xmath584 the proof of the following proposition is deferred to section [ app : prop ] . [ prop : claim3-apendix ] let @xmath585 } \ , { \bf z}^{[2]}\ , \dots \ , { \bf z}^{[r ] } \ , { \bf z}^{[r+1]}$ ] be the sequence obtained by the juxtaposition of the sequences @xmath586 } , \dots , { \bf z}^{[r+1]}.$ ] then , for @xmath436 the following conditions hold : * for each @xmath587 and @xmath588 , it holds that @xmath589}_{\kappa } \cap z^{[j]}_{\kappa ' } = \emptyset\ ] ] * for each @xmath587 and each test @xmath123 in @xmath590 with @xmath416 being the test in @xmath3 from which @xmath123 is generated , it holds that @xmath591}_{\kappa ' } ) } { c(q ) } = \frac{p(x - \bigcup_{i=1}^{j-1 } x^a_{i})}{c(x)}\ ] ] * @xmath436 is a feasible output for greedypsp ( algorithm [ alg : appendix ] ) on instance @xmath576 first note that the sequence @xmath436 has length @xmath203 and total cost @xmath187 .
this is easily verified by recalling that : ( i ) each test in the sequence @xmath436 has cost @xmath451 ; ( ii ) for each @xmath592 the subsequence @xmath578}$ ] has length @xmath593 hence @xmath594 } ) = n(x^a_j)/2 = c(x^a_j)$ ] ; ( iii ) the subsequence @xmath581}$ ] has length @xmath595}|)$ ] hence @xmath596 } ) = z = b - \sum_{j=1}^r totcost(i ' , { \bf z}^{[j]})$ ] .
let @xmath597 , and for @xmath592 let @xmath598 . by the observations in the previous paragraph
, we also have @xmath599}\,\dots \ , { \bf z}^{[j ] } > ) = \sum_{i=1}^j \sum_{\kappa = 1}^{n(x^a_i ) } c(z_{\kappa}^{[i]}).$ ] by grouping objects which incur the same cost in @xmath600 we can write @xmath601 as follows @xmath602 now , we notice that for each @xmath587 and @xmath608 the set of objects covered by @xmath609}_{\kappa}$ ] and not covered by the previous tests are @xmath610 } - \left(\bigcup_{i=1}^{j-1 } \bigcup_{\kappa'=1}^{n(x^a_i ) } z^{[i]}_{\kappa ' } \right ) - \left ( \bigcup_{\kappa'=1}^{\kappa-1 } z^{[j]}_{\kappa ' } \right ) = z_{\kappa}^{[j ] } - \left(\bigcup_{i=1}^{j-1 } \bigcup_{\kappa'=1}^{n(x^a_i ) } z^{[i]}_{\kappa ' } \right)\ ] ] where the equality follows because by proposition [ prop : claim3-apendix ] ( i ) we have that @xmath605 } \cap \left ( \bigcup_{\kappa'=1}^{\kappa-1 } z^{[j]}_{\kappa ' } \right ) = \emptyset.$ ] moreover , by proposition [ prop : claim3-apendix ] ( ii ) we have that @xmath611 } - \left(\bigcup_{i=1}^{j-1 } \bigcup_{\kappa'=1}^{n(x^a_i ) } z^{[i]}_{\kappa ' } \right)\right ) = \frac{p\left(x^a_j - \bigcup_{i=1}^{j-1 } x^a_i \right)}{n(x^a_j)}.\ ] ] finally , the set @xmath612}_{\kappa ' } \right ) - \left ( \bigcup_{\kappa=1}^{2z } z^{[r+1]}_{\kappa } \right)\ ] ] that appears in the third term of the righthand side of ( [ eq:24 ] ) can be spilt into the objects covered by the tests generated from @xmath570 and the remaining ones . by using arguments similar to those employed above one can realize that the objects covered by the tests generated from @xmath570 are exactly those generated by the objects in @xmath613 that are not covered by the tests in @xmath614}_{\kappa } \right)$ ] .
thus , their contribution to @xmath615 is @xmath616 . on the other hand ,
the contribution to @xmath615 of the remaining objects is @xmath617 therefore , we can rewrite ( [ eq:24 ] ) as follows @xmath618 via simple algebraic manipulation on the first term in the right hand side of ( [ eq:25 ] ) we have that @xmath619 and , analogously , for the second term in the right hand side of ( [ eq:25 ] ) we have @xmath620 hence , we have @xmath621 finally , we observe that @xmath622 and @xmath623 then , the sum of the second and third term in the right hand side of ( [ eq:26 ] ) can be lower bounded with @xmath624 and we get @xmath625 putting together ( [ eq:27 ] ) and ( [ eq : sep_b - x ] ) we have the desired result @xmath626 it remains to argue about the case where @xmath567 does not exist , which means that all tests that maximize the greedy criteria in algorithm [ alg : appendix ] have cost smaller than the current budget @xmath187 . in this case ,
the analysis becomes simpler and be easily handled in the same way as above .
in fact , the only difference is that the last term in ( [ eq : sep_b - x ] ) disappears , as well as all the terms referring to @xmath570 and @xmath581}.$ ] the lemma follows from the correctness of the three claims .
* proposition [ prop : claim3-apendix ] . * _ let @xmath585 } \ , { \bf z}^{[2]}\ , \dots \ , { \bf z}^{[r ] } \ , { \bf z}^{[r+1]}$ ] be the sequence obtained by the juxtaposition of the sequences @xmath586 } , \dots , { \bf z}^{[r+1]}.$ ] then , for @xmath436 the following conditions hold : _ * for each @xmath587 and @xmath588
, it holds that @xmath589}_{\kappa } \cap z^{[j]}_{\kappa ' } = \emptyset\ ] ] * for each @xmath587 and each test @xmath123 in @xmath590 with @xmath416 being the test in @xmath3 from which @xmath123 is generated , it holds that @xmath591}_{\kappa ' } ) } { c(q ) } = \frac{p(x - \bigcup_{i=1}^{j-1 } x^a_{i})}{c(x)}\ ] ] * @xmath436 is a feasible output for greedypsp ( algorithm [ alg : appendix ] ) on instance @xmath576 in order to prove ( ii ) , we observe that , from the definition of the sequences @xmath627}$ ] ( @xmath628 ) , it follows that the elements of @xmath629}_{\kappa'}$ ] are all the elements in @xmath81 which are generated from @xmath630 therefore , the elements of @xmath631 are precisely the elements of @xmath123 which are generated from @xmath632 for each @xmath633 there are precisely @xmath634 elements in @xmath123 that are generated from @xmath39 , and each one of them has probability @xmath635 hence we have @xmath636}_{\kappa ' } \right ) } { c(q ) } = \frac{1}{c(q)}\sum_{s \in x - \bigcup_{i=1}^{j-1 } x^a_i } \frac{n}{n(x)}\frac{p(s)}{n } = \frac{1}{c(q ) n(x ) } \sum_{s \in x - \bigcup_{i=1}^{j-1 } x^a_i } p(s),\ ] ] from which we have ( ii ) , since @xmath637 _ claim . _ for each @xmath638 and @xmath639 we have that @xmath640 } - \left(\bigcup_{i=1}^{j-1 } \bigcup_{\kappa'=1}^{n(x^a_i ) } z^{[i]}_{\kappa ' } \right ) - \left ( \bigcup_{\kappa'=1}^{\kappa-1 } z^{[j]}_{\kappa ' } \right)\right)}{c(z_{\kappa}^{[j ] } ) } \geq \frac{p\left(q - \left(\bigcup_{i=1}^{j-1 } \bigcup_{\kappa'=1}^{n(x^a_i ) } z^{[i]}_{\kappa ' } \right ) - \left ( \bigcup_{\kappa'=1}^{\kappa-1 } z^{[j]}_{\kappa ' } \right)\right)}{c(q)}\ ] ] for any @xmath641 this claim says that , for each @xmath587 and @xmath642 if @xmath436 has been constructed up to the test preceding @xmath605}$ ] then with respect to the tests already chosen , the test @xmath605}$ ] satisfies the greedy criterium of procedure greedypsp .
this implies that @xmath436 is a feasible output for greedypsp , as desired .
_ proof of the claim .
_ let @xmath185 be the quantity on the right hand side of ( [ eq : z - greedy ] ) , and @xmath416 be the test in @xmath3 from which @xmath123 is generated .
then we have @xmath643}_{\kappa ' } \right)\right ) } { c(q ) } \label{ineq : greedy - z-1}\\ & = & \frac{p(x - \bigcup_{i=1}^{j-1 } x^a_{i})}{c(x ) } \label{ineq : greedy - z-2 } \\ &
\leq & \frac{p\left(z_{\kappa}^{[j ] } - \left(\bigcup_{i=1}^{j-1 } \bigcup_{\kappa'=1}^{n(x^a_i ) } z^{[i]}_{\kappa ' } \right ) \right)}{c(z_{\kappa}^{[j ] } ) } \label{ineq : greedy - z-3}\\ & = & \frac{p\left(z_{\kappa}^{[j ] } - \left(\bigcup_{i=1}^{j-1 } \bigcup_{\kappa'=1}^{n(x^a_i ) } z^{[i]}_{\kappa ' } \right ) - \left ( \bigcup_{\kappa'=1}^{\kappa-1 } z^{[j]}_{\kappa ' } \right)\right)}{c(z_{\kappa}^{[j ] } ) } \label{ineq : greedy - z-4}\end{aligned}\ ] ] inequality ( [ ineq : greedy - z-1 ] ) holds since the set whose probability is considered at the numerator of the right hand side of ( [ ineq : greedy - z-1 ] ) is a superset of the set whose probability is considered at the numerator of the right hand side of ( [ eq : z - greedy ] ) .
@xmath647 } - \left(\bigcup_{i=1}^{j-1 } \bigcup_{\kappa'=1}^{n(x^a_i ) } z^{[i]}_{\kappa ' } \right ) \right)}{c(z_{\kappa}^{[j]})}\ ] ] and the last equality follows from property ( ii ) of the proposition under analysis . if @xmath644 we have that , by definition is the test in @xmath3 which maximizes the greedy criterium , but is not chosen because it exceeds the available budget ] of @xmath570 and the sequence @xmath581},$ ] it holds that @xmath648 } - \left(\bigcup_{i=1}^{j-1 } \bigcup_{\kappa'=1}^{n(x^a_i ) } z^{[i]}_{\kappa ' } \right ) \right)}{c(z_{\kappa}^{[j]})}\ ] ] where the last equality follows from property ( ii ) above .
finally , ( [ ineq : greedy - z-4 ] ) follows from ( [ ineq : greedy - z-3 ] ) because of property ( i ) above , from which we have that @xmath605 } \cap \left ( \bigcup_{\kappa'=1}^{\kappa-1 } z^{[j]}_{\kappa ' } \right ) = \emptyset$ ] hence , + @xmath649 } - \left(\bigcup_{i=1}^{j-1 } \bigcup_{\kappa'=1}^{n(x^a_i ) } z^{[i]}_{\kappa ' } \right ) - \left ( \bigcup_{\kappa'=1}^{\kappa-1 } z^{[j]}_{\kappa ' } \right)\right ) = p\left(z_{\kappa}^{[j ] } - \left(\bigcup_{i=1}^{j-1 } \bigcup_{\kappa'=1}^{n(x^a_i ) } z^{[i]}_{\kappa ' } \right)\right).$ ] | in several applications of automatic diagnosis and active learning a central problem is the evaluation of a discrete function by adaptively querying the values of its variables until the values read uniquely determine the value of the function .
in general , the process of reading the value of a variable might involve some cost , computational or even a fee to be paid for the experiment required for obtaining the value .
this cost should be taken into account when deciding the next variable to read .
the goal is to design a strategy for evaluating the function incurring little cost ( in the worst case or in expectation according to a prior distribution on the possible variables assignments ) .
our algorithm builds a strategy ( decision tree ) which attains a logarithmic approximation simultaneously for the expected and worst cost spent .
this is best possible under the assumption that @xmath0 . 0.3 in |
in the nearly 100 years since galaxy morphology became a topic of research , much has been learned about galactic structure and dynamics . known only as nebulae " at that time , galaxies were found to have a wide range of largely inexplicable forms whose relations to one another were a mystery . as data accumulated , it became clear that galaxies are fundamental units of matter in space , and an understanding of how they formed and evolved became one of the major goals of extragalactic studies . even in the era of space observations , galaxy morphology continues to be the backbone of extragalactic research as modern instruments provide information on galactic structure across a wide range of distances and lookback times . in spite of the advances in instrumentation and the explosion of data , classical galaxy morphology ( i. e. ,
the visual morphological classification in the style of hubble and others ) has not lost its relevance .
the reasons for this are as follows : \1 .
morphology is still a logical starting point for understanding galaxies .
sorting galaxies into their morphological categories is similar to sorting stars into spectral types , and can lead to important astrophysical insights .
any theory of galaxy formation and evolution will have to , at some point , account for the bewildering array of galactic forms .
galaxy morphology is strongly correlated with galactic star formation history .
galaxies where star formation ceased gigayears ago tend to look very different from those where star formation continues at the present time .
classical morphology recognizes these differences in an ordered way .
information on galaxy morphology , in the form of new types of galaxies , multi - wavelength views of previously known galaxy types , and higher resolution views of all or part of some galaxies , has exploded as modern instrumentation has superceded the old photographic plates that were once used exclusively for galaxy classification .
galaxy classification has gone beyond the realm of a few thousand galaxies to that of a _ million _ galaxies through the galaxy zoo project .
not only this , but galaxyzoo has taken morphology from the exclusive practice of a few experts to the public at large , thus facilitating citizen science at its best .
galaxy zoo images are also in _ color _ , thus allowing the recognition of special galaxy types and features based on stellar populations or gaseous emission .
finally , deep surveys with the hubble space telescope have extended morphological studies well beyond the realm of the nearby galaxies that dominated early catalogues , allowing detailed morphology to be distinguished at unprecedented redshifts . now , more than ever , galaxy morphology is a vibrant subject that continues to provide surprises as more galaxies are studied for their morphological characteristics across the electromagnetic spectrum .
it is clear that a variety of effects are behind observed morphologies , including environmental density and merger / interaction history , internal perturbations , gas accretions , nuclear activity , secular evolution , as well as the diversity in star formation histories , and that a global perspective based on large numbers of galaxies will improve theoretical models and give a more reliable picture of galactic evolution .
the goal of this article is to present the phenomenology of galaxy morphology in an organized way , and highlight recent advances in understanding what factors influence morphology and how various galaxy types are interpreted .
the article is a natural follow - up to the excellent review of galaxy morphology and classification by sandage ( 1975 ) in volume ix of the classic _ stars and stellar systems _ series .
it also complements the recently published _
de vaucouleurs atlas of galaxies _ ( buta , corwin , & odewahn 2007 , hereafter the dva ) , which provided a detailed review of the state and technique of galaxy classification up to about the year 2005 .
illustrations are very important in a review of this nature , and the article draws on a large number of sources of images . for this purpose , the sloan digital sky survey ( sdss ) , the nasa / ipac extragalactic database ( ned ) , and the dva have been most useful .
as extended objects rather than point sources , galaxies show a wide variety of forms , some due to intrinsic structures , others due to the way the galaxy is oriented to the line of sight .
the random orientations , and the wide spread of distances , are the principal factors that can complicate interpretations of galaxy morphology .
if we could view every galaxy along its principal axis of rotation , and from the same distance , then fairer comparisons would be possible .
nevertheless , morphologies seen in face - on galaxies can also often be recognized in more inclined galaxies ( figure [ difftilts ] ) .
it is only for the highest inclinations that morphology switches from face - on radial structure to vertical structure . in general
we either know the planar structure in a galaxy , or we know its vertical structure , but we usually can not know both well from analysis of images alone .
galaxy morphology began to get interesting when the leviathan of parsonstown " , the 72-inch meridian - based telescope built in the 1840s by william parsons , third earl of rosse , on the grounds of birr castle in ireland , revealed spiral patterns in many of the brighter herschel and messier nebulae . " the nature of these nebulae as galaxies was nt fully known at the time , but the general suspicion was that they were star systems ( island universes " ) like the milky way , only too distant to be easily resolved into their individual stars .
in fact , one of parsons motivations for building the leviathan " was to try and resolve the nebulae to prove this idea .
the telescope did not convincingly do this , but the discovery of spiral structure itself was very important because such structure added to the mystique of the nebulae .
the spiral form was not a random pattern and had to be significant in what it meant .
the telescope was not capable of photography , and observers were only able to render what they saw with it in the form of sketches .
the most famous sketch , that of m51 and its companion ngc 5195 , has been widely reproduced in introductory astronomy textbooks .
while visual observations could reveal some important aspects of galaxy morphology , early galaxy classification was based on photographic plates taken in the blue region of the spectrum .
silver bromide dry emulsion plates were the staple of astronomy beginning in the 1870s and were relatively more sensitive to blue light than to red light .
later , photographs taken with kodak 103a - o and iia - o plates became the standard for galaxy classification . in this part of the spectrum ,
massive star clusters , dominated by spectral class o and b stars , are prominent and often seen to line the spiral arms of galaxies .
these clusters , together with extinction due to interstellar dust , can give blue light images a great deal of detailed structure for classification .
it is these types of photographs which led to the galaxy classification systems in use today . in such photographs , we see many galaxies as a mix of structures .
inclined galaxies reveal the ubiquitous _ disk _ shape , the most highly flattened subcomponent of any galaxy .
studies of doppler wavelength shifts in the spectra of disk objects ( like hii regions and integrated star light ) reveal that disks rotate differentially .
if a galaxy is spiral , the disk is usually where the arms are found , and also where the bulk of interstellar matter is found .
the radial luminosity profile of a disk is usually _ exponential _ , with departures from an exponential being due to the presence of other structures . in the central area of a disk - shaped galaxy
, there is also often a bright and sometimes less flattened mass concentration in the form of a _
bulge_. the nature of bulges and how they form has been a topic of much recent research , and is discussed further in section 9 .
disk galaxies range from virtually bulge - less to bulge - dominated . in the center there
may also be a conspicuous _ nucleus _ , a bright central concentration that was usually lost to overexposure in photographs .
nuclei may be dominated by ordinary star light , or may be _ active _ , meaning their spectra show evidence of violent gas motions .
_ bars _ are the most important internal perturbations seen in disk - shaped galaxies .
a bar is an elongated mass often made of old stars crossing the center .
if spiral structure is present , the arms usually begin near the ends of the bar .
although most easily recognized in the face - on view , bars have generated great interest recently in the unique ways they can also be detected in the edge - on view .
not all bars are made exclusively of old stars . in some bulge - less galaxies ,
the bar has considerable gas and recent star formation .
related to bars are elongated disk features known as _
ovals_. ovals usually differ from bars in lacking higher order fourier components ( i.e. , have azimuthal intensity distributions that vary mainly as 2@xmath0 ) , but nevertheless can be major perturbations in a galactic disk .
the entire disk of a galaxy may be oval , or a part of it may be oval .
oval disks are most easily detected if there is considerable light or structure at larger radii .
_ rings _ are prominent features in some galaxies .
often defined by recent star formation , rings may be completely closed features or may be partial or open , the latter called pseudorings . "
rings can be narrow and sharp or broad and diffuse .
it is particularly interesting that several kinds of rings are seen , and that some galaxies can have as many as four recognizeable ring features .
_ nuclear rings _ are the smallest rings and are typically seen in the centers of barred galaxies . _
inner rings _ are intermediate - scale features that often envelop the bar in a barred galaxy .
outer rings are large , low surface brightness features that typically lie at about twice the radius of a bar .
other kinds of rings , called accretion rings , polar rings , and collisional rings , are also known but are much rarer than the inner , outer , and nuclear rings of barred galaxies .
the latter kinds of rings are also not exclusive to barred galaxies , but may be found also in nonbarred galaxies .
_ lenses _ are features , made usually of old stars , that have a shallow brightness gradient interior to a sharp edge .
they are commonly seen in hubble s disk - shaped s0 class ( section 5.2 ) .
if a bar is present , the bar may fill a lens in one dimension .
lenses may be round or slightly elliptical in shape .
if elliptical in shape they would also be considered ovals .
_ nuclear bars _
are the small bars occasionally seen in the centers of barred galaxies , often lying within a nuclear ring .
when present in a barred galaxy , the main bar is called the primary bar " and the nuclear bar is called the secondary bar . "
it is possible for a nuclear bar to exist in the absence of a primary bar . _
dust lanes _ are often seen in optical images of spiral galaxies , and may appear extremely regular and organized .
they are most readily evident in edge - on or highly inclined disk galaxies , but are still detectable in the face - on view , often on the leading edges of bars or the concave sides of strong inner spiral arms .
spiral arms may also show considerable morphological variation .
spirals may be regular 1 , 2 , 3 , or 4-armed patterns , and may also be higher order multi - armed patterns .
spirals may be tightly wrapped ( low pitch angle ) or very open ( high pitch angle . )
a _ grand - design _ spiral is a well - defined global pattern , often detectable as smooth variations in the stellar density of old disk stars .
a _ flocculent _
spiral is made of small pieces of spiral structure that appear sheared by differential rotation .
their appearance can be strongly affected by dust , such that at longer wavelengths a flocculent spiral may appear more grand - design .
pseudorings can be thought of as variable pitch angle spirals which close on themselves , as opposed to continuously opening , constant pitch angle , logarithmic spirals .
there are also numerous structures outside the scope of traditional galaxy classification , often connected with strong interactions between galaxies . plus , the above described features are not necessarily applicable or relevant to what we see in very distant galaxies .
accounting for all of the observed features of nearby galaxies , and attempting to connect what we see nearby to what is seen at high redshift , is a major goal of morphological studies .
as noted by sandage ( 1975 ) , the first step in studying any class of objects is a classification of those objects .
classification built around small numbers of shared characteristics can be used for sorting galaxies into fundamental categories , which can then be the basis for further research . from such research ,
physical relationships between identified classes may emerge , and these relationships may foster a theoretical interpretation that places the whole class of objects into a global context . there is no doubt that such an approach greatly contributed to the development of the science of biology , and this is no less true for galaxies .
the genesis of galaxy classification is to take the complex combinations of structures described in the previous section and summarize them with a few type symbols .
sandage ( 1975 ) describes the earlier classification systems of wolf , reynolds , lundmark , and shapley that fell into dis - use more than 50 years ago .
the morgan ( 1958 ) spectral type / concentration classification system , which was based on a connection between morphology ( specifically central concentration ) and the stellar content of the central regions , was used recently by bershady , jangren , & conselice ( 2000 ) in a mostly quantitative manner ( see also abraham et al . 2003 ) . thus , morgan s system has in a way survived into the modern era but not in the purely visual form that he proposed .
only one morgan galaxy type , the supergiant cd type , is still used extensively ( section 10.6 ) .
van den bergh s luminosity / arm morphology classification system is described by van den bergh ( 1998 ; see section 6.5 ) .
the big survivor of the early visual classification systems was that of hubble ( 1926 , 1936 ) , as later revised and expanded upon by sandage ( 1961 ) and de vaucouleurs ( 1959 ) .
sandage ( 1975 ) has argued that one reason hubble s view prevailed is that he did not try and account for every superficial detail , but kept his classes broad enough that the vast majority of galaxies could be sorted into one of his proposed bins .
these bins were schematically illustrated in hubble s famous tuning fork " ( hubble 1936 ; reproduced in figure [ tuning_fork_schematic ] ) recognizing a sequence of progressive flattenings from ellipticals to spirals .
ellipticals had only two classification details : the smoothly declining brightness distribution with no inflections , and no evidence for a disk ; and the ellipticity of the isophotes , indicated by a number after the e " symbol .
( for example , e3 means the ellipticity is 0.3 . )
spirals were systems more flattened than an e7 galaxy that could be subdivided according to the degree of central concentration , the degree of openness of the arms , the degree of resolution of the arms into complexes of star formation ( all three criteria determining position along the fork ) , and on the presence or absence of a bar ( determining the appropriate prong of the fork ) .
the s0 class at the juncture of the prongs of the fork was still hypothetical in 1936 . as armless disk galaxies , " s0s were mysterious because all examples known in 1936 were barred .
these were classified as sba , but this was a troubling inconsistency because nonbarred sa galaxies had full spiral patterns .
hubble predicted the existence of nonbarred s0s to fill the gap between type e7 and sa and cure what he felt was a cataclysmic " transition .
it was not long before hubble himself realized that the tuning fork could not adequately represent the full diversity of galaxy morphologies , and after 1936 he worked on a revision that included real examples of the sought group : nonbarred s0 galaxies .
based on fragmentary notes he left behind , sandage ( 1961 ) prepared the _
hubble atlas of galaxies _ to illustrate hubble s revision , and also added a third dimension : the presence or absence of a ring .
this was the first major galaxy atlas illustrating a classification system in a detailed , sophisticated way with beautifully produced photographs .
hubble s revision , with van den bergh luminosity classes ( sandage & tammann 1981 ) , was updated and extended to types later than sc by sandage & bedke ( 1994 ) .
because sandage ( 1961 ) and sandage & bedke ( 1994 ) describe the hubble - sandage revision so thoroughly , the details will not be repeated here . instead
, the focus of the next section will be on the de vaucouleurs revision as outlined in the dva .
the reasons for this are : ( 1 ) the de vaucouleurs classification provides the most familar galaxy types to extragalactic researchers , mostly because of extensive continuing use of the third reference catalogue of bright galaxies ( rc3 , de vaucouleurs et al .
1991 ) ; and ( 2 ) the de vaucouleurs classification is still evolving to cover more details of galaxy morphology considered significant at this time .
it should be noted that both the de vaucouleurs and hubble - sandage revisions are strictly applicable only to @xmath1 @xmath2 0 galaxies and that it is often difficult to fit objects having @xmath1 @xmath3 0.5 neatly into the categories defining these classification systems .
high redshift galaxy morphology is described in section 13 .
the hubble tuning fork is useful because it provides a visual representation of information hubble ( 1926 ) had only stated in words .
the fork contains an implication of continuity .
for example , it does not rule out that there might be galaxies intermediate in characteristics between an `` sa '' or `` sb '' spiral , or between a normal `` s '' spiral and a barred `` sb '' spiral .
continuity along the elliptical galaxy sequence was always implied as a smooth variation from round ellipticals ( e0 ) to the most flattened ellipticals ( e7 ) .
sandage ( 1961 ) describes the modifications that made the hubble system more three - dimensional : the introduction of the ( r ) ( inner ring ) and ( s ) ( pure spiral ) subtypes .
continuity even with this characteristic was possible , using the combined subtype ( rs ) .
thus , already by 1961 , the hubble classification system had become much more complicated than it was in 1926 or 1936 .
the addition of the s0 class was one reason for this , but the ( r ) and ( s ) subtypes were another . in the hubble - sandage classification , it became common to denote galaxies on the left part of the hubble sequence as early - type " galaxies and those on the right part as late - type " galaxies . by the same token
, sa and sba spirals became early - type spirals " while sc and sbc spirals became late - type spirals .
" sb and sbb types became known as intermediate - type spirals . " the reason for these terminologies was convenience and borrows terminology often used for stars .
young , massive stars of spectral classes o and b were known as early - type stars " while older stars of cooler spectral types were known as late - type stars . "
hubble stated that his use of these temporal descriptions for galaxies had no evolutionary implications .
an irony in this is that it eventually became clear that early - type galaxies are dominated by late - type stars , while late - type galaxies often have significant numbers of early - type stars .
de vaucouleurs ( 1959 ) took the idea of continuity of galaxy morphology a step further by developing what he referred to as the _ classification volume _
( figure [ classification_volume ] ) . in this revision of the hubble - sandage ( 1961 ) classification ,
galaxy morphology represents a continuous sequence of forms in a three - dimensional volume with a long axis and circular cross - sections of varying size .
the long axis of the volume is the _ stage _ , or type , and it represents the long axis of the original hubble tuning fork .
the short axes are the family and the variety , which refer to apparent bar strength and the presence or absence of an inner ring , respectively .
in addition to hubble s original stages e , sa , sb , and sc , the classification volume includes new stages : late ellipticals : e@xmath4 , very late " spirals : sd , magellanic spirals " : sm , and magellanic irregulars " : i m .
the s0 class is included in the same position along the sequence , between e s and spirals , but is subdivided into three stages .
the characteristics defining individual stages are described further in section 5.2 .
the stage is considered the most fundamental dimension of the classification volume because measured physical parameters , such as integrated color indices , mean surface brightnesses , and neutral hydrogen content correlate well with position along the sequence ( e. g. , buta et al .
early - type galaxies tend to have redder colors , higher average surface brightnesses , and lower neutral hydrogen content than late - type galaxies .
the family and variety axes of the classification volume indicate the considerable variations in morphology at a given stage . a famous sketch of families and varieties near stage sb , drawn by de vaucouleurs himself during a cloudy night at mcdonald observatory circa 1962 , is shown in sandage ( 1975 ) and in figure 1.13 of the dva .
the classification volume is broader in the middle compared to the ends because the diversity of galaxy morphology is largest at stages like s0/a and sa .
bars and rings are often most distinct and most recognizable at these stages .
such features are not characteristic of e galaxies , so the volume must be narrow at that end . along the s0 sequence , bars and rings are barely developed among the earliest s0s ( s0@xmath5 ) and well - developed among the late s0s ( s0@xmath4 ) , thus the volume begins to broaden .
among very late - type galaxies , sd , sdm , sm , and i m , bars are actually very frequent , but closed rings ( r ) are not .
thus , the volume narrows at that end as well . for the purposes of illustrating morphology , blue light digital images converted to units of magnitudes per square arcsecond are used when available .
this approach is described in the dva , and requires calibration of the images , usually based on published photoelectric multi - aperture photometry .
in addition , some of the illustrations used ( especially in section 15 ) are from the sloan digital sky survey or from other sources .
these are not in the same units but still provide excellent illustrations of morphology .
sdss color images differ from dva images mainly in the central regions , where sdss images sometimes lose detail .
unlike the dva , the scope of this article extends beyond the traditional @xmath6 wavebands .
it is only during the past 20 years that significant morphological information has been obtained for galaxies outside these bands , mostly at mid- and far - ultraviolet and mid - ir wavelengths from space - based observatories capable of imaging in these wavelengths to unprecedented depths , providing a new view of galaxy morphology that is only beginning to be explored .
a useful review of many issues in morphology is provided by van den bergh ( 1998 ) .
elliptical galaxies are smooth , amorphous systems with a continuously declining brightness distribution and no breaks , inflections , zones , or structures , as well as no sign of a disk .
figure [ egals ] shows some good examples . because ellipticals are dominated by old stars and are relatively dust - free , they look much the same at different wavelengths .
hubble s subclassification of ellipticals according to apparent ellipticity ( en , where n=10(1-b / a ) , b / a being the apparent flattening ) was useful but virtually no physical characteristics of ellipticals correlate with this parameter ( kormendy & djorgovski 1989 ) . the n value in the en classification is simply the projected ellipticity and not easily interpreted in terms of a true flattening without direct knowledge of the orientation of the symmetry planes .
luminous ellipticals are thought to be triaxial in structure with an anisotropic velocity dispersion tensor , while lower luminosity ellipticals are more isotropic oblate rotators ( davies et al .
1983 ) . studies of rotational to random kinetic energy ( @xmath7/@xmath8 ) versus apparent flattening ( @xmath9 ) show that massive ellipticals are slow anisotropic rotators .
ellipticals follow a fundamental plane relationship between the effective radius @xmath10 of the light distribution , the central velocity dispersion @xmath11 , and mean effective surface brightness @xmath12 ( see review by kormendy & djorgovski 1989 ) .
dwarf elliptical galaxies may not follow the same relation as massive ellipticals ; this is discussed by ferguson & binggeli ( 1994 ) . the lack of fundamental significance of hubble s en classification led some authors to seek an alternative , more physically useful approach .
kormendy & bender ( 1996 ; see also the review by schweizer 1998 ) proposed a revision to hubble s tuning fork handle that orders ellipticals according to their velocity anisotropy , since this is a significant determinant of e galaxy intrinsic shapes .
velocity anisotropy correlates with the _ deviations _ of e galaxy isophotes from pure elliptical shapes , measured by the parameter @xmath13 , the relative amplitude of the @xmath14 fourier term of these deviations .
if this relative amplitude is positive , then the isophotes are pointy , disky ovals , while if negative , the isophotes are boxy ovals . a boxy elliptical
is classified as e(b ) , while a disky elliptical is classified as e(d ) .
the correlation with anisotropy is such that e(b ) galaxies have less rotation on average and more velocity anistropy than e(d ) galaxies .
the kormendy & bender proposed classification is shown in figure [ boxydisky ] , together with two exceptional examples that show the characteristic isophote shapes .
the leftmost of these two , ngc 7029 , is an unusually obvious boxy elliptical at larger radii .
this boxiness is the basis for the classification e(b)5 .
however , ngc 7029 is not boxy throughout : it shows evidence of a small inner disk and hence is disky at small radii .
this is not necessarily taken into account in the classification .
the other image shown in figure [ boxydisky ] is ngc 4697 , a galaxy whose isophotes are visually disky .
the idea with the kormendy & bender classification is that it is the disky ellipticals which connect to s0 galaxies , and not the boxy ellipticals .
however , ngc 7029 demonstrates that diskiness and boxiness can be a function of radius , thus perhaps a smooth connection between e(b ) and e(d ) types [ i.e. , type e(b , d ) ] is possible and the order shown in the kormendy & bender revision to the hubble tuning fork , with boxy es blending into the disky es , could be reasonable . note
that while the classical en classifications of ellipticals were designed by hubble to be estimated by eye , this is not easily done for the e(b ) and e(d ) classifications , which are most favored to be seen only when the disk is nearly edge - on . face - on es with imbedded disks will not show disky isophotes .
for example , ngc 7029 and 4697 are extreme cases where the isophotal deviations are obvious by eye .
but for most e galaxies , the e(b)n and e(d)n classifications can only be judged reliably with measurements of the @xmath13 parameter .
the de vaucouleurs classification of ellipticals includes a slightly more advanced type called e@xmath4 , or late " ellipticals .
it was originally intended to describe the first stage of the transition to the s0 class " ( de vaucouleurs 1959 ) .
five examples of e@xmath4 galaxies are shown in the second row of figure [ egals ] .
galaxies classified as e@xmath4 can be the most subtle s0s , but many of the e@xmath4 cases listed in rc3 are the brightest members of clusters that have shallow enough brightness profiles to appear to have an extended envelope ( see section 10.6 ) . of the five e@xmath4 galaxies shown ,
only ngc 4623 ( rare type e@xmath47 ) seems consistent with de vaucouleurs s original view .
it is also a much lower luminosity system than the other four cases shown .
while s0@xmath5 is the type most often confused with ellipticals in visual classification , the bin has a wide spread from the most obvious to the least obvious cases .
thus , a type like e@xmath4 is still useful for distinguishing transitions from e to s0 galaxies .
the photometric properties of ellipticals depend on luminosity . in terms of sersic @xmath15 profile
fits , large , luminous ellipticals tend to have profiles described better by @xmath16@xmath174 , while smaller , lower luminosity ellipticals tend to have @xmath16@xmath184 , with values as low as 1 ( e. g. , caon , capaccioli , & donofrio 1993 ) .
graham & guzmn ( 2003 ) discuss the implications of this correlation on proposed dichotomies of elliptical galaxies ( e.g. , kormendy 1985 ; see also ferrarese et al .
these studies received a major impetus from the massive photometric analysis of virgo cluster elliptical galaxies by kormendy et al .
two issues were considered by these authors ( figure [ cores ] ) .
the first was whether galaxies classified as dwarf ellipticals ( de `` ; see section 15.2 ) in the virgo cluster really are the low luminosity extension of more massive , conventional ellipticals , or something different altogether . based on parameter correlations , such as @xmath19 , the major axis radius of the isophote containing 10% of the total visual luminosity , versus @xmath20 , the surface brightness of this isophote , kormendy et al . show that even the most elliptical - like and luminous de galaxies lie on a distinct sequence from normal elliptical galaxies , which tend to lie on a higher density sequence . in a graph of @xmath21-band central surface brightness versus absolute @xmath21-band magnitude , the de galaxies lie in a region occupied by magellanic irregular galaxies , suggesting a link between the groups and consistent with the earlier conclusions of kormendy ( 1985 ) .
kormendy et al .
( 2009 ) suggest reclassifying binggeli , sandage , & tammann s ( 1985 ) ' ' de `` galaxies and related objects ( like dwarf s0 , or ds0 galaxies ) as ' ' spheroidal `` ( sph ) galaxies , including the type ' ' sph , n `` , meaning ' ' nucleated spheroidal " galaxy ( section 15.2 ) .
figure [ cores ] shows several examples of sph , n galaxies as compared with several genuine elliptical galaxies .
the morphological appearance alone does not necessarily distinguish the two classes .
the classification is physical , being based mainly on parameter correlations .
kormendy et al .
suggest that sph galaxies are formed from late - type systems by environmental effects and supernovae .
the second issue considered was the physical distinction between core " elliptical galaxies , those where the surface brightness profile approaches either a constant level or a slightly sloped level with radius approaching zero , and coreless " ellipticals ( also known as power law " es ) where the inner profile steepens with decreasing radius ( kormendy 1999 ) .
kormendy et al .
( 2009 ) illustrated both types relative to a sersic @xmath22 fit to the outer regions of the luminosity profiles . in this representation ,
core es are missing light " relative to the fit while coreless es have extra light . "
the top row of figure [ cores ] shows one core e ( ngc 4472 ) and two coreless es [ ngc 4458 and ic 798 ( vcc 1440 ) , the latter a low - luminosity dwarf ] .
the subtle distinctions are evident in these images , with ngc 4472 showing a soft center and ngc 4458 showing a strong center .
the terminology for both types is mostly historical ( kormendy 1999 ) and somewhat counter to the visual impression ( i.e. , ngc 4472 lacks a bright core while ngc 4458 has one , yet the latter is technically coreless ) .
kormendy et al .
show that core and coreless e galaxies have different sersic indices , velocity dispersion anisotropy , isophote shapes , and rotational character , with the core es being of the boxy type and the coreless es of the disky type in figure [ boxydisky ] .
the distinction may be tied to the number of mergers that formed the system .
the full classification of spiral and s0 galaxies involves the recognition of the stage , family , and variety . in de
vaucouleurs s classification approach , the implication for bars , inner rings , and stages is a continuum of forms ( de vaucouleurs 1959 ) , so that there are no sharp edges to any category or cell " apart from the obvious ones ( for example , there are no galaxies less barred " than a nonbarred galaxy , nor are there galaxies more ringed than those with a perfectly closed ring ) . the classification of s0 galaxies depends on recognizing the presence of a disk and a bulge at minimum , and usually a lens as well , and no spiral arms .
examples are shown in figure [ s0gals ] .
the display of galaxy images in units of mag arcsec@xmath23 makes lenses especially easy to detect , as noted in the dva .
even if a lens is nt obvious , a galaxy could still be an s0 if it shows evidence of an exponential disk .
( lenses are also not exclusive to s0s . )
the no spiral arms " characteristic is much stricter in the hubble - sandage classification than in the de vaucouleurs interpretation , because varieties ( r , rs , and s ) are carried into the de vaucouleurs classifications of s0s .
this allows the possibility of a classification like sa(s)0@xmath5 , which would be very difficult to recognize .
bars enter in the classification of s0s in a similar manner as for spirals .
figure [ s0gals ] shows mainly stage differences among nonbarred and barred s0s .
the stage for s0s ranges from early ( s0@xmath5 ) , to intermediate ( s0@xmath24 ) , and finally to late ( s0@xmath4 ) , in a succession of increasing detail .
the earliest nonbarred s0s may be mistaken for elliptical galaxies on photographic images , and indeed sandage & bedke ( 1994 ) note cases where they believe an s0 galaxy has been misclassified as an elliptical by de vaucouleurs in his reference catalogues ( see also _ the revised shapley - ames catalogue _ , rsa , sandage & tammann 1981 ) .
this kind of misinterpretation is less likely for types s0@xmath24 and s0@xmath4 , because these will tend to show more obvious structure .
the morphological distinction between e and s0 galaxies has been considered from a quantitative kinematic point of view by emsellem et al .
these authors argue that the division of early - type galaxies into e and s0 types is contrived " , and that it is more meaningful to divide them according to a quantitative kinematic parameter called @xmath25 , the specific angular momentum of the stellar component , which is derived from a two - dimensional velocity field obtained with the sauron integral field spectrograph ( bacon et al . 2001 ) . based on this parameter ,
early - type galaxies are divided into slow and fast rotators , i. e. , whether they are characterized by large - scale rotation or not . in a sample of 48 early - types ,
most were found to be fast rotators classified as a mix of e and s0 types , while the remainder were found to be slow rotators classified as es .
this kind of approach , which provides a more physical distinction among early - types , does not negate completely the value of the e and s0 subdivisions , but highlights again the persistent difficulty of distinguishing the earliest s0s from es by morphology alone .
the transition type s0/a shows the beginnings of spiral structure .
two examples are included in figure [ s0gals ] , one nonbarred and the other barred .
type s0/a is a well - defined stage characterized in the de vaucouleurs 3d classification volume as having a high diversity in family and variety characteristics .
the type received a negative characterization as the garbage bin " of the hubble sequence at one time because troublesome dusty irregulars , those originally classified as irr ii " by holmberg ( 1950 ) and later as i0 " by de vaucouleurs , seemed to fit better in that part of the sequence .
[ in fact , de vaucouleurs , de vaucouleurs , & corwin ( 1976 ) assigned the numerical stage index @xmath26=0 to both s0/a and i0 galaxies .
] however , this problem is only a problem at optical wavelengths . at longer wavelengths ( e.g. , 3.6 microns ) ,
types such as irr ii or i0 are less needed as they are defined mainly by dust ( buta et al .
2010a ) .
in general , the stage for spirals is based on the appearance of the spiral arms ( degree of openness and resolution ) and also on the relative prominence of the bulge or central concentration .
these are the usual criteria originally applied by hubble ( 1926 , 1936 ) .
figure [ tuning_fork ] shows the stage sequence for spirals divided according to bar classification ( sa , sab , sb ) , and as modified and extended by de vaucouleurs ( 1959 ) to include sd and sm types . intermediate stages , such as sab , sbc , scd , and sdm , are shown in figure [ intermediate_types ] .
as noted by de vaucouleurs ( 1963 ) , these latter stages are almost as common as the basic ones .
the three hubble criteria are basically seen in the illustrated galaxies .
sa galaxies tend to have significant bulges , and tightly - wrapped and relatively smooth spiral arms .
sab galaxies are similar to sa galaxies , but show more obvious resolution of the arms .
sb galaxies have more resolution and more open arms , and generally smaller bulges than sab galaxies .
sbc galaxies have considerable resolution and openness of the arms , and also usually significant bulges . in sc galaxies
, the bulge tends to be very small and the arms patchy and open .
scd galaxies tend to be relatively bulgeless , patchy armed sc galaxies .
stage sd is distinctive mainly as almost completely bulgeless late - type spirals with often ill - defined spiral structure .
stages sdm and sm are the most characteristically asymmetric stages , the latest spiral types along the de vaucouleurs revised hubble sequence .
they are described in detail by de vaucouleurs & freeman ( 1972 ) and by odewahn ( 1991 ) .
sm is generally characterized by virtually no bulge and a single principle spiral arm .
if a bar is present , it is usually not at the center of the disk isophotes , unlike what is normally seen in earlier type barred spirals .
this leads to the concept of an _ offset barred galaxy_. the single spiral arm emanates from one end of the bar . as noted by freeman ( 1975 ) , this is a basic and characteristic asymmetry of the mass distribution of magellanic barred spirals .
sdm galaxies are similar , but may show a weaker or shorter second arm . in figure [ tuning_fork ] ,
ngc 4618 is an especially good example of an sbm type ( odewahn 1991 ) , while in figure [ intermediate_types ] , ngc 4027 is illustrated as type sbdm .
an important issue regarding these galaxies is whether the optically offset bar is also offset from the dynamical rotation center of the disk . in a detailed hi study of the interacting galaxy pair ngc 4618 and 4625 , bush & wilcots ( 2004 ) found very regular velocity fields and extended hi disks , but no strong offset of the rotation center from the center of the bar
this is similar to what pence et al .
( 1988 ) found for the offset barred galaxy ngc 4027 , based on optical fabry - perot interferometry .
in contrast , both magellanic clouds , which are also offset barred galaxies , were found to have hi rotation centers significantly offset from the center of the bar ( kerr & de vaucouleurs 1955 ) .
in general , the application of hubble s three spiral criteria allows consistent classification of spiral types .
nevertheless , sometimes the criteria are inconsistent .
for example , small bulge sa galaxies are described by sandage ( 1961 ) and sandage & bedke ( 1994 ) . barred galaxies with nuclear rings can have spiral arms like those of an earlier hubble type and very small bulges .
in such conflicting cases , the emphasis is usually placed on the appearance of the arms . also , while late - type sdm and sm galaxies are characteristically asymmetric , other types may be asymmetric as well . on average , the bulge - to - total luminosity ratio is related to hubble type , but the result is sensitive to how galaxies are decomposed ( e. g. , laurikainen et al .
asymmetry has been quantified by conselice ( 1997 ) .
the family classifications sa , sab , and sb are purely visual estimates of bar strength , for both spirals and s0s . they are highlighted already in figures [ s0gals] [ intermediate_types ] , but the continuity of this characteristic is better illustrated in figure [ continuity ] , where de vaucouleurs ( 1963 ) underline classifications ( s@xmath27b and sa@xmath28 ) are also shown .
an sa galaxy has no evident bar in general , although high inclination can cause a mistaken sa classification if a bar is highly foreshortened .
also , internal dust may obscure a bar ( see , e. g. , eskridge et al .
an sb galaxy should have a clear , well - defined bar .
the intermediate bar classification sab is one of the hallmarks of the de vaucouleurs system , and is used to recognize galaxies having characteristics intermediate between barred and nonbarred galaxies .
it is used for well - defined ovals or simply weaker - looking normal bars .
the weakest primary bars are denoted s@xmath27b while the classification sa@xmath28 is usually assigned to more classical bars that appear only somewhat weaker than conventional bars .
most of the time , galaxies which should be classified as sa@xmath28 are simply classified as sb .
variety is also treated as a continuous classification parameter ( figure [ continuity ] , second row ) .
a spiral galaxy having a completely closed or very nearly completely closed inner ring is denoted ( r ) .
the spiral arms usually break from the ring .
if the spiral arms break directly from the central region or the ends of a bar , forming a continuously winding , open pattern , the variety is ( s ) .
the intermediate variety ( rs ) is also well - defined .
inner rings which appear broken or partial are in this category .
the dash - dot - dash in brackets " morphology : ( -o- ) , where a bar with a bulge is bracketted by spiral arcs overshooting the bar axis , is very typical of variety ( rs ) .
the example of this shown in figure [ continuity ] is ngc 4548 .
we use the notation @xmath29s to denote an inner ring made up of tightly wrapped spiral arms that do not quite close , while the notation r@xmath30 is used for very open , barely recognizable , inner pseudorings .
a good example of the former is ngc 3450 , while an example of the latter is ngc 5371 . a _
spindle _ is a highly inclined disk galaxy . for blue - light images , usually an sp "
after the classification automatically implies considerable uncertainty in the interpretation , because family and variety are not easily distinguished when the inclination is high .
figure [ edge - ons ] shows , however , that stages can be judged reasonably reliably for edge - on galaxies .
one important development in the classification of edge - on galaxies has been the ability to recognize edge - on bars through boxy / peanut and x "- shapes .
boxy / peanut bulges in edge - on galaxies were proven to be bars seen edge - on from kinematic considerations ( e. g. , kuijken & merrifield 1995 ) .
this shape is evident in ngc 4425 ( row 1 , column 4 of figure [ edge - ons ] ; see also section 9 ) . for spiral and s0 galaxies that are not too highly inclined ( i.e. , not spindles ) , once the stage , family , and variety are determined these are combined in the order family , variety , stage for a final full type .
for example , ngc 1300 is of the family sb , variety ( s ) , and stage b , thus its full type is sb(s)b .
the s0@xmath4 galaxy ngc 4340 has both a bar and inner ring and its full type is sb(r)0@xmath4 .
the classification is flexible enough that if , for example , the family and variety of a galaxy can not be reliably determined owing to high inclination , while the stage can still be assessed , then the symbols can be dropped and a type such as sb " or s0 " can still be noted .
magellanic irregular galaxies represent the last normal stage of the de vaucouleurs revised hubble sequence .
several examples are shown in figure [ irregulars ] .
the objects illustrated in the top row are all examples of ( s)-variety irregulars with bars or some trace of a bar .
nevertheless , not all magellanic irregulars have bars . irregulars of the lowest luminosities are usually classified simply as i m since the sophistication of structure needed to distinguish something like family " may not exist for such galaxies .
irregular galaxies are important for their star formation characteristics . as noted by hunter ( 1997 ) , irregulars are similar to spirals in having both old and young stars , as well as dust , atomic , molecular , and ionized gas , but lack the spiral structure that might trigger star formation .
thus , they are useful laboratories for examining how star formation occurs in the absence of spiral arms .
although irregulars are largely defined by a lack of well - organized structure like spiral arms , the two lower right galaxies in figure [ irregulars ] are not so disorganized looking and seem different from the other cases shown .
ngc 5253 looks almost like a tilted s0 galaxy , yet it has no bulge at its center nor any obvious lenses .
instead , the central area is an irregular zone of active star formation .
the central zone was interpreted by van den bergh ( 1980a ) as fossil evidence " for a burst of star formation , possibly triggered by an interaction with neighboring m83 .
this is a case where the de vaucouleurs classification of i0 seems reasonable : ngc 5253 is an early - type galaxy with a central starburst , probably the youngest and closest example known ( vanzi & sauvage 2004 ) .
it is a magellanic irregular galaxy imbedded in a smooth s0-like background known to have an early - type star spectrum .
ngc 1705 , also shown in figure [ irregulars ] , is similar but has a super star cluster near the center and obvious peculiar filaments .
it is classified as a blue compact dwarf by gil de paz et al .
both galaxies are classified as amorphous by sandage & bedke ( 1994 ) .
the de vaucouleurs classification volume recognizes three principal aspects of galaxy morphology , but clearly there are many more dimensions than three .
stage , family , and variety are the dimensions most clearly highlighted in blue light images and have a wide scope .
other dimensions may be considered and for some there is explicit notation in use . published de vaucouleurs types include an extra dimension known as the outer ring / pseudoring classification .
several examples of outer rings and pseudorings are shown in figure [ outer_rings ] .
an outer ring is a large , often diffuse structure , typically seen in barred early - type galaxies ( stages s0@xmath4 to sa ) at a radius approximately twice that of the bar .
closed outer rings are recognized with the type symbol ( r ) preceding the main part of the classification .
for example , an sb(r)0@xmath4 galaxy having an outer ring has a full classification of ( r)sb(r)0@xmath4 .
interestingly , rare cases of double outer ring galaxies , type ( rr ) , are known , where two detached outer rings are seen ; an example is ngc 2273 shown in the upper left frame of figure [ outer_rings ] . in later - type galaxies , a large outer ring - like feature
is often seen made of outer spiral arms whose variable pitch angle causes them to close together .
these features are classified as outer pseudorings , symbolized by ( r@xmath31 ) preceding the main type symbols [ e.g. , as in ( r@xmath31)sb(r)ab ] .
outer pseudorings are mainly observed in sa to sbc galaxies , and are only rarely seen in the very late stages sc - sm . among bright nearby galaxies , outer rings and pseudorings
are found at about the 10% level ( buta & combes 1996 ) .
typically , outer rings are fainter than 24 mag arcsec@xmath23 in blue light . with such low surface brightnesses ,
the rings can be easily lost to galactic extinction .
the division between outer rings and pseudorings is also not sharp .
some outer pseudorings are only barely distinguishable from outer rings .
continuity applies to these features as it does for inner rings although there is no symbol other than s " for an outer spiral pattern which does not close into an outer pseudoring .
although closed outer rings ( r ) are equally well - recognized in the rsa and the carnegie and hubble atlases , outer pseudorings are a unique feature of the de vaucouleurs revision .
the value of recognizing these features is that many show morphologies consistent with the theoretical expectations of the outer lindblad resonance ( olr , schwarz 1981 ) , one of the major low - order resonances that can play a role in disk evolution .
resonance rings are discussed further in section 10.1 , but figure [ olr_schematic ] shows schematics of the morphologies generally linked to this resonance .
the schematics are designed to highlight the subtle but well - defined aspects of these features , while figure [ olr_subclasses ] shows images of several examples of each morphology , including the models " used for the schematic .
outer rings of type r@xmath32 are closed rings that are slightly dimpled towards the bar axis , a shape which connects directly to one of the main periodic orbit families near the olr as shown in schwarz ( 1981 ) and in the dva .
outer pseudorings of type r@xmath33 are similar to type r@xmath32 but are made of two spiral arms that wind approximately 180@xmath24 with respect to the ends of the bar .
even these will usually show a dimpled shape .
outer pseudorings of type r@xmath34 are different from this in that two spiral arms wind 270@xmath24 with respect to the ends of the bar , such that the arms are doubled in the quadrants immediately trailing the bar .
the shapes r@xmath32 , r@xmath33 , and r@xmath34 were predicted by schwarz ( 1981 ) based on sticky - particle " numerical simulations . not predicted by those simulations ( but later shown in extensions of those simulations by byrd et al .
1994 and rautiainen & salo 2000 ) is an interesting combined ring morphology called r@xmath32r@xmath34 , which consists of a closed r@xmath32 ring and an r@xmath34 pseudoring .
this combination is especially important because it demonstrates not only a continuity of morphologies among outer rings and pseudorings different from the continuity between outer rings and pseudorings in general , but also it is a morphology that can be linked directly to the dynamics of barred galaxies .
note that the classification shown in figures [ olr_schematic ] and [ olr_subclasses ] does not depend on whether the rings are in fact linked to the olr .
the illustrated morphologies are abundant and easily recognized regardless of how they are interpreted .
( section 10.1 discusses other interpretations that have been proposed . )
although the schwarz models guided the search for these morphologies , rautiainen , salo , & buta ( 2004 ) and treuthardt et al .
( 2008 ) showed that some outer pseudorings classified as r@xmath33 are more likely related to the outer 4:1 resonance and not the olr .
these cases are generally recognizable by the presence of secondary spiral arcs in a four - armed pattern in the area of the bar ( ngc 1433 in figure [ difftilts ] and eso 566@xmath3524 in figure [ spiraltypes ] are examples ) .
the olr subclassifications are used in the same manner as the plain outer ring and pseudoring classifications .
for example , ngc 3081 has the full type ( r@xmath32r@xmath34)sab(r)0/a .
the value of recognizing lenses as significant morphological components was first emphasized by kormendy ( 1979 ) , who suggested a dynamical link between inner lenses , which are often filled by a bar in one dimension , and dissolved or dissolving bars .
kormendy noted that lenses can be of the inner or outer type , in a manner analogous to inner and outer rings .
he suggested the notation ( l ) for inner lenses and ( l ) for outer lenses to be used in the same position of the classification as inner and outer rings . for example , the galaxy ngc 1543 is type ( r)sb(l)0/a while galaxy ngc 2983 is type ( l)sb(s)0@xmath4 .
figure [ lenses ] demonstrates the continuity between rings and lenses , which is evident not only among barred galaxies but among nonbarred ones as well .
this continuity is recognized by the type symbol ( rl ) , also used by kormendy ( 1979 ) .
this type refers to a low contrast inner ring at the edge of a clear lens .
even underline classifications ( @xmath29l ) and ( r@xmath36 ) may be recognized .
a rare classification , ( r@xmath31l ) , refers to an inner pseudoring / lens , a type of feature that is seen in ngc 4314 and recognized as such in the dva .
similarly , figure [ outer_lenses ] shows a continuity between outer rings and outer lenses through the type classification ( rl ) , referring to an outer lens with a weak ring - like enhancement .
underline types @xmath37l and r@xmath38 may also be recognized .
the origin of outer lenses could be in highly evolved outer rings .
the central regions of barred galaxies often include distinct morphological features in the form of small rings and secondary bars .
the rings , known as nuclear rings because of their proximity to the nucleus well inside the ends of the primary bar , are sites of some of the most spectacular starbursts known in normal galaxies .
the rings are typically @xmath21.5 kpc in linear diameter and intrinsically circular in shape .
figure [ nrnb ] ( top row ) shows three examples : ngc 1097 , 3351 , and 4314 .
these images highlight the small bulges that seem characteristic of nuclear - ringed barred galaxies .
the three galaxies illustrated have types ranging from sa to sb , but based on the bulge size the types would be considerably later .
for example , ngc 3351 has the bulge of an sd galaxy .
comern et al .
( 2010 ) carried out an extensive statistical study of nuclear ring radii , and identified a subclass known as ultra - compact " nuclear rings ( ucnrs ) .
such rings were recognized mainly in hubble space telescope images and are defined to be less than 200pc in diameter .
( see figure [ bulges ] for an example in ngc 3177 . )
comern et al .
showed that ucnrs are the low size tail of the global nuclear ring population .
this study also showed that bar strength impacts the sizes of nuclear rings , with stronger bars generally hosting smaller nuclear rings than weaker bars .
comern et al .
( 2010 ) were also able to derive a reliable estimate of the relative frequency of nuclear rings as 20%@xmath392% over the type range s0@xmath5 to sd , confirming with smaller error bars the previous result of knapen ( 2005 ) .
assuming that nuclear rings are a normal part of galaxy evolution , these authors argue that the rings may survive for 2 - 3 gyr .
interestingly , it was also found that 19%@xmath394% of nuclear rings occur in nonbarred galaxies , implying either that the rings may have formed when a bar was stronger ( evidence of bar evolution ) or that ovals or other mechanisms can lead to their formation .
mazzuca et al .
( 2009 ; see also knapen 2010 ) connect some of the properties of nuclear rings to the rate at which the rotation curve rises in the inner regions .
the most extreme nuclear ring known is found in the sba galaxy eso 565@xmath3511 ( see also section 7 ) . at 3.5kpc in diameter , not only is it one of the largest known nuclear rings , but also the ring has an extreme elongated shape compared to more typical nuclear rings .
nuclear bars lie in the same radial zone as nuclear rings and sometimes lie inside a nuclear ring .
three examples are shown in the second row of figure [ nrnb ] . these average about one - tenth the size of a primary bar .
there is no preferred angle between the axis of the nuclear bar and the primary bar , suggesting that the two features have different pattern speeds ( buta & combes 1996 ; dva ) .
neither nuclear rings nor nuclear bars were recognized in the original hubble - sandage - de vaucouleurs classifications , presumably in part because the use of small - scale photographic plates for extensive galaxy classification limited the detectability of the features in the ( typically ) overexposed centers .
modern multi - band digital imaging greatly facilitates the detection of the small rings and bars , allowing their inclusion in the classification .
buta & combes ( 1996 ) and buta et al . (
2010a ) suggested the notation nr for nuclear rings and nb for nuclear bars , respectively , to be used as part of the variety classification as in , for example , sb(r , nr)b for ngc 3351 , or sab(l , nb)0/a for ngc 1291 .
continuity may exist for these features like other rings and primary bars .
[ for example , nuclear lenses ( nl ) may also be recognized . ] in blue light images , the appearance of the central region of a barred galaxy can be strongly affected by dust . for example , ngc 1365 shows a nuclear spiral in blue light , while in the infrared , the morphology is that of a nuclear ring ( buta et al .
the morphologies of some galaxies have a full complement of classifiable features .
for example , accounting for all the rings and bars seen in ngc 3081 , the classification is ( r@xmath32r@xmath34)sab(r , nr , nb)0/a . lisker et al . ( 2006 ) use the terminology s2b " for double - barred galaxies , a reasonable alternative approach to classifying these objects .
lisker et al .
successfully identified nuclear bars in galaxies at redshifts @xmath1=0.10 - 0.15 ( from hst acs observations ) , the most distant ones recognized thus far .
a classification such as sb " tells one that a galaxy is a spiral of moderate pitch angle and degree of resolution of the arms , and that a significant bulge may be present .
the type does not directly tell : ( 1 ) the multiplicity of the spiral pattern ; ( 2 ) the character of the arms ( massive , filamentary , grand design , or flocculent ) ; or ( 3 ) the sense of winding of the arms ( leading or trailing the direction of rotation ) .
these are nevertheless additional dimensions to galaxy morphology .
the multiplicity of the spiral pattern refers to the actual number of spiral arms , usually denoted by the integer @xmath40 .
examples of spirals having @xmath40=1 to 5 are illustrated in figure [ spiraltypes ] .
the multiplicity is not necessarily straightforward to determine and may be a function of radius .
for example , a spiral may be two - armed in the inner regions and multi - armed in the outer regions .
spirals of low @xmath40 are usually _ grand design _ , a term referring to a well - defined global ( meaning galaxy - wide ) pattern of strong arms .
the typical grand design spiral has two main arms , as in ngc 5364 ( lower left frame of figure [ spiraltypes ] ) .
in contrast , a flocculent spiral has piecewise continuous arms but no coherent global pattern ( elmegreen 1981 ) .
ngc 5055 is an example shown in the middle left frame of figure [ spiraltypes ] .
this category is relevant mainly to optical wavebands . in the infrared , an optically flocculent spiral like
ngc 5055 reveals a more coherent global grand design spiral ( thornley 1996 ; see also figure [ nir03 ] ) , indicating that dust is partly responsible for the flocculent appearance .
the terms massive " and filamentary " arms are due to reynolds ( 1927 ) and are discussed by sandage ( 1961 , 1975 ) .
massive arms are broad , diffuse , and of relatively low contrast , as in m33 , while filamentary arms are relatively thin in comparison , and lined by knots or filaments , as in ngc 5457 ( m101 ) .
de vaucouleurs ( 1956 ) originally used these distinctions as part of his classification , but later dropped the references to spiral arm character probably because of the complexity it added to his types .
elmegreen & elmegreen ( 1987 ) used a different approach to spiral arm character by recognizing a series of spiral _ arm classes _ based on arm continuity and length ( but not necessarily contrast ) .
ten classes ranging from flocculent ( acs 1 - 4 ) to grand design ( 5 - 12 ; numbers 10 and 11 were later dropped ) .
examples of each are illustrated in figure [ arm_classes ] ( see elmegreen & elmegreen 1987 for a description of each class ) .
thus , spiral character is a well - developed additional dimension to galaxy classification .
a simpler approach advocated by elmegreen & elmegreen is g " for grand - design , f " for flocculent , and m " for multiple - armed .
the arms of grand design spirals are in general thought to be density waves and may in fact represent quasi - steady wave modes ( e.g. , bertin et al . 1989 ; zhang 1996 , 1998 , 1999 ) , although there is also some evidence that spirals may be transient ( see review by sellwood 2010 ) .
flocculent spirals may be sheared self - propogating star formation regions ( seiden & gerola 1982 ) .
figure [ spiraltypes ] also shows two examples of a new class of spirals , called _
counter - winding _ spirals . in these cases ,
an inner spiral pattern winds outward in the opposite sense to an outer spiral pattern . in the case of ngc 4622 ( row 2 , middle ) , the inner pattern has only a single arm and the outer pattern has two arms , while in ngc 3124 ( row 2 , middle right ) , the inner pattern is two - armed while the outer pattern is at least four - armed .
the two cases are very different because ngc 4622 is essentially nonbarred while in ngc 3124 , the inner spiral is classified as a bar .
the presence of oppositely winding spiral patterns in the same galaxy means that one set of arms is trailing ( opening opposite the direction of rotation ) while the other set is leading ( opening into the direction of rotation ) . in general , studies of the dust distribution as well as the rotation of spirals has shown that trailing arms are the rule ( de vaucouleurs 1958 ) .
surprisingly , straightforward analysis of a velocity field and the dust pattern in ngc 4622 led buta , byrd , & freeman ( 2003 ) to conclude that the strong outer two - armed pattern in this galaxy is leading , while the inner single arm is trailing .
this led to the characterization of ngc 4622 as a backwards spiral galaxy , " apparently rotating the wrong way .
an additional nonbarred counter - winding spiral has been identified in eso 297@xmath3527 by grouchy et al .
( 2008 ) . in this case
, the same kind of analysis showed that an inner single arm leads while a 3-armed outer pattern trails .
no comparable analysis has yet been made for ngc 3124 .
vaisanen et al .
( 2008 ) have shown that a two - armed ( but not counter - winding ) spiral in the strongly interacting galaxy iras 18293 - 3413 is leading .
even with this , the number of known leading spirals is very small ( dva ) .
leading spirals are not expected to be as long - lived as trailing spirals since they do not transfer angular momentum outwards and this is needed for the long - term maintenance of a spiral wave ( lynden - bell & kalnajs 1972 ) .
an interesting example of leading armlets " was described by knapen et al .
( 1995a ) , who used @xmath41-band imaging of the center of the grand design spiral m100 to reveal details of its nuclear bar and well - known nuclear ring / spiral .
the nuclear bar has a leading twist that connects it to two bright @xmath41-band knots " of star formation .
this morphology was interpreted in terms of the expectations of gas orbits in the vicinity of an inner inner lindblad resonance ( iilr ; knapen et al .
1995b ) .
the final galaxy in figure [ spiraltypes ] is ngc 4921 , an example of an _ anemic _ spiral .
this is a type of spiral that is deficient in neutral atomic hydrogen gas , and as a consequence it has a lower amount of dust and star formation activity .
the arms of ngc 4921 resemble those of an sb or sbc galaxy in pitch angle and extensiveness , but are as smooth as those typically seen in sa galaxies .
anemic spirals were first recognized as galaxies with fuzzy " arms ( see van den bergh 1998 ) where star formation has been suppressed due to ram - pressure stripping in the cluster environment . in the case of ngc 4921 ,
the environment is the coma cluster .
the idea is that such galaxies will eventually turn into s0 galaxies ( van den bergh 2009a ) .
anemic spiral galaxies are further discussed in section 10.2 .
seigar et al .
( 2008 ) have demonstrated the existence of a correlation between spiral arm pitch angles and supermassive central black hole masses .
the sense of the correlation is such that black hole mass is highest for the most tightly wound spirals and lowest for the most open spirals .
the correlation is expected because black hole mass is tightly correlated to bulge mass and central mass concentration , and spiral arm pitch angle is tied to shear in galactic disks , which itself depends on mass concentration ( seigar et al .
2005 ) .
luminosity effects are evident in the morphology of galaxies through surface brightness differences between giants and dwarfs , and through the sophistication of structure such as spiral arms .
van den bergh ( 1998 ) describes his classification system which takes luminosity effects into account using a set of luminosity classes that are analogous to those used for stars .
the largest , most massive spirals have long and well - developed arms , while less massive spirals have less well - defined arms .
the nomenclature for the classes parallels that for stars : i ( supergiant galaxies ) , ii ( bright giant galaxies ) , iii ( giant galaxies ) , iv ( subgiant galaxies ) and v ( dwarf galaxies ) .
intermediate cases i - ii , ii - iii , iii - iv , and iv - v , are also recognized .
figure [ lclasses ] shows galaxies which van den bergh ( 1998 ) considers primary luminosity standards of his classification system .
the original van den bergh standards for these classes were based on the small scale paper prints of the palomar sky survey .
sandage and tammann ( 1981 ) adopted the precepts of the van den bergh classes but revised the standards based on large - scale plates .
in general , luminosity class i galaxies have the longest , most well - developed arms , luminosity class iii galaxies have short , patchy arms extending from the main body , while luminosity class v galaxies have very low surface brightness and only a hint of spiral structure .
the classes are separated by type in figure [ lclasses ] because among sb galaxies , few are of luminosity class iii or fainter , while among sc and later type galaxies , the full range of luminosity classes is found .
van den bergh does not use types like sd or sm for conventional de vaucouleurs late - types , but instead uses s@xmath5 and s@xmath4 to denote early " ( smoother ) and late " ( more patchy ) subgiant spirals .
similarly , van den bergh uses sb@xmath5 and sb@xmath4 to denote early " and late " sb spirals , respectively .
( some of these would be classified as sab and sbc by de vaucouleurs . ) according to the standards listed by van den bergh ( 1998 ) , an sb i galaxy is 2 - 3 mag more luminous than an sb iii galaxy , while an sc i galaxy is more than 4 mag more luminous than an s v galaxy .
bars are among the most common morphological features of disk - shaped galaxies . unlike spiral arms , bars cross the spiral - s0 divide " in the hubble sequence and are abundant among spirals ( at the 50 - 70% level ) when both sab and sb types are considered ( de vaucouleurs 1963 ; sellwood & wilkinson 1993 )
. the bar fraction has cosmological significance ( sheth et al .
2008 ) , and many estimates of the nearby galaxy bar fraction have been made from both optical and ir studies ( see buta et al .
2010a for a summary of recent work ) .
bars are fairly well - understood features of galaxy morphology that have been tied to a natural instability in a rotationally supported stellar disk ( see review by sellwood & wilkinson 1993 ) .
the long - term maintenance of a bar in a mostly isolated galaxy is thought to depend on how effectively it transfers angular momentum to other galaxy components , such as the halo ( athanassoula 2003 ) .
bars are thought to be transient features that , in spiral galaxies , may dissolve and regenerate several times over a hubble time ( bournaud & combes 2002 ) .
alternatively , bars may be long - lived density wave modes that drive secular evolution of both the stellar and gaseous distributions ( zhang & buta 2007 ; buta & zhang 2009 ) .
the possible secular evolution of bars in s0 galaxies is discussed by buta et al .
( 2010b ) .
bars are also thought to drive spiral density waves ( kormendy & norman 1979 ; buta et al .
2009 ; salo et al .
2010 ) .
the actual morphology of bars shows interesting variations that merit further study .
the family classifications sab and sb indicate some measure of bar strength , but do not allude to the varied appearances of bars even among those only classified as sb .
regular bars , such as those illustrated in figure [ tuning_fork ] , are the conventional types that define the sb class .
figure [ ansae ] shows ansae "- type bars , referring to bars which have handles " or bright enhancements at the ends .
martinez - valpuesta et al .
( 2007 ) carried out a statistical study and found that ansae are present in @xmath240% of early - type barred galaxies and are very rare for types later than stage sb . ansae are usually detectable in direct images , but their visibility can be enhanced using unsharp - masking ( all the right frames for each galaxy in figure [ ansae ] ) .
morphologically , ansae may be small round enhancements like those seen in ngc 5375 and 7020 , but in some cases , ansae are approximately linear enhancements , giving the bar a parallelogram appearance as in ngc 7098 , or curved arcs , giving the bar a partial ring appearance as in ngc 1079 .
color index maps in the dva show that ansae are generally as red as the rest of the bar , indicating the features are stellar dynamical in origin , rather than gas - dynamical .
nevertheless , ansae made of star - forming regions are known .
martinez - valpuesta et al .
( 2007 ) illustrate the case of ngc 4151 , a well - known seyfert 1 galaxy with a strong bar - like inner oval .
the appearance of this galaxy s ansae in the 1.65@xmath42 m @xmath43-band is shown in the lower right frames of figure [ ansae ] , where the ansae are seen to have irregular shapes compared to the others shown .
another subtlety about bars is their general boxy character .
athanassoula et al .
( 1990 ) showed that generalized ellipses fit the projected isophotes of bars better than do normal ellipses . for 11 or 12 sb0 galaxies examined in this study , a significant degree of boxiness
was found near the bar semi - major axis radius . the unsharp - masked image of ngc 7020 in figure [ ansae ] shows an x - shaped pattern in the inner regions that is the likely signature of a significantly three - dimensional bar .
ngc 1079 and 5375 shows hints of similar structure .
the x - pattern is expected to be especially evident in edge - on galaxies which show the extended vertical structure of bars .
many examples have been analyzed ( bureau et al .
2006 ; see also the dva ) .
buta et al .
( 2010a ) have suggested that edge - on bars recognized from the x - pattern be denoted sb@xmath44 . "
the cause of bar ansae is uncertain . in simulations ,
martinez - valpuesta et al .
( 2006 ) found that ansae form late , after a second bar - buckling episode in a disk model with a live axisymmetric halo , and appear as density enhancements in both the face - on and edge - on views .
figure [ ovals ] shows three examples of galaxies having oval inner disks .
these features are described in detail by kormendy and kennicutt ( 2004=kk04 ) , who present both photometric and kinematic criteria for recognizing them .
the images in figure [ ovals ] are optical and have been cleaned of foreground and background objects , and have also been deprojected based on the mean shape and major axis position angle of faint outer isophotes . in all three cases ,
the presence of an outer ring allows clear recognition of the oval shape , assuming that the inner and outer structures are in the same plane .
the upper panels of figure [ ovals ] show the full morpholopy with the outer rings , while the lower panels show the bright oval inner disks .
the shapes of the oval disks are varied and range from axis ratio 0.84 for ngc 4736 to 0.55 for ngc 1808 .
the most striking example is ngc 4941 , whose oval disk includes a bright , normal - looking spiral pattern with isophotal axis ratio 0.68 .
many other examples are provided by kk04 .
the ovals appear to play a bar - like role in these galaxies .
the outer rings may be resonant responses to the nonaxisymmetric potential of the ovals , which clearly harbor a great deal of mass in spite of the mildness of their departures from axisymmetry .
as noted by kk04 , oval disk galaxies are expected to evolve secularly in much the same manner as typical barred galaxies .
on the other hand , ovals themselves could be products of bar secular evolution .
laurikainen et al .
( 2009 ) found that the near - ir bar fraction in s0 galaxies is significantly less than that in s0/a or early - type spiral galaxies ( also found by aguerri et al .
2009 in the optical ) , while the oval / lens fraction is higher , suggesting that some ovals / lenses might be dissolved bars .
further evidence that bars might be dissolving in some galaxies is the detection of extremely weak bars in residual images of visually nonbarred s0 galaxies where a two - dimensional decomposition model has been subtracted .
such a bar is detected in the sa0@xmath24 galaxy ngc 3998 ( laurikainen et al .
aguerri et al .
( 2009 ) suggest that central concentration is a key factor in bar evolution , and that a unimodal distribution of bar strengths argues against the idea that bars dissolve and reform ( bournaud & combes 2002 ) .
regular barred galaxies also often include an oval bounded by an elongated inner ring .
the deprojected blue light images of two examples are shown in figure [ inner_rings ] .
ngc 1433 has a very strong normal bar and one of the most intrinsically elongated inner rings known .
the inner ring lies at the edge of an oval which is more conspicuous at longer wavelengths . in ngc 1433 , the inner ring ,
the oval , and the bar are all aligned parallel to each other .
the situation is different in eso 565@xmath3511 , whose bright oval is strongly misaligned with a prominent bar , but similar to ngc 1433 , the oval is bounded by an inner pseudoring .
the suggestion in this case is that the bar and the oval are independent patterns .
dust lanes are often the most prominent part of the interstellar medium detectable in an optical image of a galaxy .
figure [ dust_types ] shows different classes of dust lanes , using direct optical images on the left for each galaxy , and a color index map on the right .
the color index maps are coded such that blue features are dark while red features ( like dust lanes ) are light .
the bars of intermediate ( mainly sab to sbc ) spirals often show _ leading _ dust lanes , that is , well - defined lanes that lie on the leading edges of the bars , assuming the spiral arms trail .
the example shown in figure [ dust_types ] , ngc 1530 , has an exceptionally strong bar and the lanes are very straight , regular , and well - defined .
these dust lanes are a dynamical effect associated with the bar
. the lanes may be curved or straight .
athanassoula ( 1992 ) derived models of bar dust lanes and tied the curvature to the strength of the bar in the sense that models with stronger bars developed straighter dust lanes .
comern et al .
( 2009 ) recently tested this idea by measuring the curvature of actual dust lanes as well as quantitative values of the bar strengths for 55 galaxies .
they found that strong bars can only have straight dust lanes , while weaker bars can have straight or curved lanes . in the same manner as bars , a strong spiral often has dust lanes on the concave sides of the inner arms .
this is shown for ngc 1566 in the upper right panels of figure [ dust_types ] .
both bar and spiral dust lanes are face - on patterns .
another type of face - on pattern is the _ dust ring_. the inner dust ring of ngc 7217 is shown in the right , middle frames of figure [ dust_types ] , and it appears as the dark , inner edge of a stellar ring having the same shape .
dust rings can also be detected in more inclined galaxies .
an inclined galaxy with a significant bulge also can show another dust effect : in such a case , the bulge is viewed through the dust layer on the near side of the disk , while the dust is viewed through the bulge on the far side of the disk .
this leads to a reddening and extinction asymmetry across the minor axis such that the near side of the minor axis is more reddened and extinguished than the far side . in conjunction with rotation data ,
this near side / far side asymmetry was used by hubble ( 1943 ) and de vaucouleurs ( 1958 ) to show that most spirals trail the direction of rotation .
the lower frames of figure [ dust_types ] show the planar dust lanes seen in edge - on spiral galaxies .
the lane in ngc 7814 ( lower left frames ) is red which indicates that the galaxy is probably no later in type than sa .
this is consistent with the large bulge seen in the galaxy .
however , the planar dust lane seen in ngc 891 , type sb , has a thin blue section in the middle of a wider red section .
the blue color is likely due to outer star formation that suffers relatively low extinction .
some individual star forming regions can be seen along the dust lane . in spite of the blue color
, we are only seeing the outer edge of the disk in the plane .
also related to galactic dust distributions are observations of _ occulting galaxy pairs _ , where a foreground spiral galaxy partly occults a background galaxy , ideally an elliptical ( white & keel 1992 ) . with such pairs , one can estimate the optical depth of the foreground dust , often in areas where it might not be seen easily in an isolated spiral .
an excellent example is described by holwerda et al .
( 2009 ) , who are able to trace the dust distribution in an occulting galaxy to 1.5 times than the standard isophotal radius . another way of illustrating the dust distribution in galaxies is with _ spitzer space telescope _ infrared array camera ( irac ) images at 8.0@xmath42 m wavelength .
this is discussed further in section 12 .
a bulge is a very important component of a disk galaxy . in the context of structure formation in a cold dark matter ( cdm )
cosmology , bulges may form by hierarchical merging of disk galaxies , a process thought to lead to elliptical galaxies if the disks have approximately equal mass .
bulges formed in this way should , then , resemble elliptical galaxies , especially for early - type spirals . the bulges of later - type spirals , however , can be very different from the expectations of a merger - built bulge ( also known as a classical " bulge ) . in many cases
, the bulge appears to be made of material associated with the disk .
kk04 reviewed the concept of pseudobulges , " referring to galaxy bulges that may have formed by slow secular movement of disk gas to the central regions ( also known as disk - like bulges ; athanassoula 2005 ) .
the main driving agent for movement of the gas is thought to be bars , which are widespread among spiral galaxies and which exert gravity torques that can move material by redistributing the angular momentum . inside the corotation resonance , where the bar pattern speed equals the disk rotation rate ,
gas may be driven into the center to provide the raw material for building up a pseudobulge .
kk04 review the evidence for such processes and argue that pseudobulges are a strong indication that secular evolution is an important process in disk - shaped galaxies .
figure [ bulges ] shows the morphologies of both classical bulges and pseudobulges .
the four classical bulge galaxies shown in the lower row , m31 , ngc 2841 , m81 , and m104 , have bright smooth centers and no evidence for spiral structure or star formation .
classical bulges also tend to have rounder shapes than disks , and can have significant bulge - to - total luminosity ratios as illustrated by m104 .
classical bulges are also more supported by random motions than by rotation .
many references to classical bulge studies are given by kk04 .
formation mechanisms of such bulges are discussed in detail by athanassoula ( 2005 ) .
the two upper rows of figure [ bulges ] are all pseudobulges as recognized by kk04 .
the first row shows hst wide @xmath7-band ( filter f606w ) images of the inner 1 - 1.3 kpc of four galaxies , ngc 3177 , 4030 , 5377 , and 1353 , in the type range sa - sbc .
the areas shown account for much of the rise in surface brightness above the inward extrapolation of the outer disk light in these galaxies , and would be considered bulges just on this basis .
the hst images show , however , considerable spiral structure , dust , small rings , and likely star formation in these regions , characteristics not expected for a classical bulge .
kk04 argue that instead these are pseudobulges that are highly flattened , have a projected shape similar to the outer disk light , have approximately exponential brightness profiles ( sersic index @xmath16@xmath21 - 2 ) , and have a high ratio of ordered rotation to random motions .
kk04 argue that a low sersic index compared to @xmath16=4 appears to be the hallmark of these pseudobulges , and a signature of secular evolution .
the second row in figure [ bulges ] shows other kinds of pseudobulges discussed by kk04 .
ngc 6782 and 3081 ( two left frames ) have secondary bars , and kk04 considered that such features indicate the presence of a pseudobulge because bars are always disk features . in each case
, the secondary bar lies inside a nuclear ring .
the other two galaxies in the second row of figure [ bulges ] , ngc 128 and 1381 , are examples of boxy or box - peanut bulges .
these features have been linked to the vertical heating of bars , and if this is what they actually are , then kk04 argue that boxy and box - peanut bulges are also examples of pseudobulges .
however , boxy and box / peanut bulges would _ not _ necessarily be the result of slow movement of gas by bar torques , and subsequent star formation in the central regions , but instead would be related to the orbital structure of the bar itself ( athanassoula 2005 ) .
recent studies have shown that pseudobulges are the dominant type of central component in disk galaxies .
although originally thought to be important only for late - type galaxies , laurikainen et al .
( 2007 ) showed that pseudobulges are found throughout the hubble sequence , including among s0-s0/a galaxies , based on sophisticated two - dimensional photometric decompositions .
such galaxies frequently have nuclear bars , nuclear disks , or nuclear rings .
laurikainen et al . also found that bulge - to - total ( @xmath45 ) flux ratios are much less than indicated by earlier studies , especially for early hubble types , and that the sersic index averages @xmath46 2 across all types .
the lack of gas in s0 and s0/a galaxies complicates the interpretation of their pseudobulges in terms of bar - driven gas flow and subsequent star formation . instead , laurikainen et al .
link the pseudobulges in early - type galaxies to the evolution of bars .
laurikainen et al .
( 2010 ) also showed that s0s can have pseudobulges if they are stripped spirals , without invoking any bar - induced evolution .
galaxy morphology is replete with evidence for gravitational interactions , ranging from minor , distant encounters , to violent collisions and major / minor mergers .
many of the most puzzling and exotic morphologies can be explained by interactions , and even sublimely normal galaxies , like ordinary ellipticals , have been connected to catastrophic encounters . up to 4% of bright nearby galaxies
are involved in a major interaction ( knapen & james 2009 ) . in clusters ,
other types of interactions may occur , such as gas stripping and truncation of the star - forming disk . in this section , a variety of the types of morphologies that may be considered the results of external interactions are described and illustrated .
the three types of rings described so far , nuclear , inner , and outer rings , are aspects of the morphology of relatively normal , undisturbed galaxies .
inner rings and pseudorings are found in more than 50% of normal disk galaxies ( buta & combes 1996 ) , while outer rings and pseudorings are found at the 10% level .
the latter rings could be more frequent because their faintness may cause them to go undetected , which is less likely to occur for inner rings .
as has been noted , nuclear rings are found at the 20% level ( comern et al .
the high abundance of these types of ring features suggests that they are mainly products of _ internal _ dynamics , and in fact all three ring types have been interpreted in terms of internal processes in barred galaxies .
the main interpretation of these kinds of rings has been in terms of orbital resonances with the pattern speed of a bar , oval , or spiral density wave .
resonances are special places where a bar can secularly gather gas owing to the properties of periodic orbits ( buta & combes 1996 ) .
buta ( 1995 ) showed that the intrinsic shapes and relative bar orientations of inner and outer rings and pseudorings supports the resonance interpretation of the features .
schwarz ( 1981 ) suggested the outer lindblad resonance for outer rings and pseudorings , while schwarz ( 1984 ) suggested the inner 4:1 ultraharmonic resonance for inner rings and pseudorings , and the inner lindblad resonance for nuclear rings .
knapen et al .
( 1995b ) and buta & combes ( 1996 ) provide further insight into these interpretations .
the resonance idea may only be valid in the case of weak perturbations . in the presence of a strong perturbation , the concept of a specific _ resonance radius _ can break down , although the idea of a broad _ resonance region _ could still hold ( contopoulos 1996 ) .
regan and teuben ( 2003 , 2004 ) argue that nuclear rings and inner rings are better interpreted in terms of orbit transitions , that is , regions where periodic orbits transition from one major orbit family to another , as in the transition from the perpendicularly - aligned @xmath47 family to the bar - aligned @xmath48 family ( contopoulos & grosbol 1989 ) .
normal rings have also been interpreted in terms of invariant manifolds " which emanate from the unstable @xmath49 and @xmath50 lagrangian points in the bar potential ( romero - gmez et al .
2006 , 2007 ; athanassoula et al .
2009a , b ) .
this approach has also had some success in predicting the shapes and orientations of inner and outer rings , such as the r@xmath32 , r@xmath33 , r@xmath32r@xmath34 , and r@xmath34 morphologies shown in figure [ olr_schematic ] and figure [ olr_subclasses ] .
a morphology called rr@xmath32 " , which includes an oval inner ring and a figure eight - shaped r@xmath32 ring ( see ngc 1326 in figure [ olr_subclasses ] ) , is especially well - represented by this kind of model .
the manifolds are tubes which guide orbits escaping the @xmath49 and @xmath50 regions .
note that in this interpretation , outer rings are not necessarily associated with the olr ( romero - gmez et al .
2006 ) .
although the vast majority of the ring - like patterns seen in galaxies are probably of the resonance / orbit - transition / invariant - manifold type , other classes of rings are known that are likely the result of more catastrophic processes , such as galaxy collisions .
figure [ ringtypes ] shows resonance rings in comparison to three other types : accretion rings , polar rings , and collisional rings ( the latter commonly referred to as ring galaxies " ) . the three accretion rings shown , in hoag s object ( schweizer et al .
1987 ) , ic 2006 ( schweizer et al .
1989 ) , and ngc 7742 ( de zeeuw et al . 2002 ) are thought to be made of material from an accreted satellite galaxy . for ic 2006 and ngc 7742
, the evidence for this is found in _ counter - rotation _ :
the material in the rings counter - rotates with respect to the material in the rest of the galaxies . in hoag s object and ic 2006 ,
the accreting galaxy is a normal e system .
polar rings are also accreted features except that the accreting galaxy is usually a disk - shaped system , most often an s0 ( whitmore et al . 1990 ) . in these cases ,
the accreted material comes in at a high angle to the plane of the disk .
the configuration is most stable if the accretion angle is close to 90@xmath24 , or over the poles of the disk system .
this limits the ability of differential precession to cause the ring material to quickly settle into the main disk .
the polar feature can be a ring or simply an inclined and extended disk .
whitmore et al .
( 1990 ) presented an extensive catalogue of probable and possible polar ring galaxies .
the main example illustrated in figure [ ringtypes ] is ngc 4650a , where the inner disk component is an s0 .
galaxies like ngc 4650a have generated considerable research because polar rings probe the shape of the dark halo potential ( e. g. , sackett et al .
the galaxies are also special because the merging objects have retained their distinct identities , when most mergers lead to a single object .
brook et al .
( 2008 ) link the misaligned disks of polar ring galaxies to the process of hierarchical structure formation in a cold dark matter scenario . while polar rings are most easily recognizable when both disks are nearly edge - on to us , cases where one or the other disk is nearly face - on have also been recognized .
one example , eso 235@xmath3558 ( buta & crocker 1993 ) is shown in the middle - right panel of figure [ ringtypes ] . in this case , the inner component is almost exactly edge - on and shows a planar dust lane , and is likely a spiral rather than an s0 .
the ring component is inclined significantly to the plane of this inner disk but may not be polar .
the faint outer arms in this component caused eso 235@xmath3558 to be misclassified as a late - type barred spiral in rc3 .
spiral structure in polar disks has been shown to be excitable by the potential of the inner disk , which acts something like a bar ( theis , sparke , & gallagher 2006 ) .
an example where the main disk is seen nearly face - on is ngc 2655 ( sparke et al .
2008 ) . in this case , the polar ring material is seen as silhouetted dust lanes at an uncharacteristic angle to the inner isophotes .
ngc 2655 also shows evidence of faint shells / ripples , indicative of a recent merger ( section 10.3.3 ) .
silchenko & afanasiev ( 2004 ) have discussed ngc 2655 and other similar examples of inner polar rings in terms of the triaxiality of the potential . also illustrated in figure [ ringtypes ]
is ngc 660 , which was listed as a possible polar ring galaxy by whitmore et al .
( 1990 ) . like eso 235@xmath3558 ,
ngc 660 has an aligned dust lane in its inner disk component , which thus is likely to be a spiral , not an s0 .
the extraplanar disk is actually far from polar , being inclined only 55@xmath24 ( van driel et al .
1995 ) . a recent study of massive stars in the ring is given by karataeva et al .
( 2004 ) .
collisional ring galaxies ( arp 1966 ; appleton & struck - marcell 1996 ) are thought to be cases where a larger galaxy suffers a head - on collision with a smaller galaxy down its polar axis .
the collision causes an expanding density wave of massive star formation , and multiple rings are possible .
three examples are shown in figure [ ringtypes ] .
theys & spiegel ( 1976 ) have discussed various classes of ring galaxies .
arp 147 ( arp 1966 ) is an example of type re " , referring to a sharp elliptical ring with an empty interior .
the cartwheel ( higdon 1995 ) and the lindsay - shapley ring ( arp & madore 1987 ) are examples of type rn " , meaning an elliptical ring with an off - center nucleus .
not shown in figure [ ringtypes ] is a third category called rk " , where a single , large knot lies on one side of the ring , making the system very asymmetric .
madore , nelson , & petrillo ( 2009 ) have published a comprehensive atlas of all known likely collisional ring galaxies , many taken from the catalogue of arp & madore ( 1987 ) .
based on this study , only 1 in 1000 galaxies is a collisional ring galaxy . for entry , the madore et al .
atlas requires at least two objects in the immediate vicinity of the ring that might plausibly be the intruder galaxy .
most of the rings are not in cluster environments , however .
the atlas also brings attention to several double - ring collisional systems , which have been predicted by numerical simulations ( see struck 2010 for a review ) .
the unusual radial spokes " in the cartwheel , a feature not seen in any other collisional ring galaxy , could be related to interactive accretion streams ( struck et al .
1996 ) .
romano , mayya , & vorobyov ( 2008 ) present images of several ring galaxies that show the pre - collision stellar disk .
they also show that rings are generally delineated by blue knots and that the off - centered nuclei are usually more yellow in color .
in addition , some of the companion galaxies show diffuse asymmetric outer light suggesting that they are being stripped .
figure [ ringtypes ] shows that accretion rings can account for some of the rings seen in nonbarred galaxies .
buta & combes ( 1996 ) argue that a bar is an essential element in resonance ring formation .
eso 235@xmath3558 shows that a polar ring - related system can resemble a ringed , barred galaxy .
the three collisional rings are all very distinctive from the others .
galaxy clusters are excellent laboratories for detecting the effects of environment on galaxy morphology and structure .
frequent mergers and environmental conversion of spirals into s0s are thought to be at the heart of the morphology - density relation , where early - type galaxies dominate cluster cores , and spirals and irregulars are found mainly in the outer regions ( dressler 1980 ; van der wel et al .
2010 ) .
the issue of environmental effects has a direct bearing on how we might interpret the hubble sequence .
for example , the continuity of galaxy morphology certainly seems apparent from the discussions in previous sections of this review .
the hubble sequence e - s0-sa - sb - sc - i appears physically significant when total colors , mean surface brightnesses , and hi mass - to - blue light ratios are considered , and the way features are recognized in the classification systems also favors the continuity .
morphological continuity does _ not _ , however , automatically imply that the galaxy types are in fact ordered correctly .
for example , although hubble placed s0s as a transition type between elliptical galaxies and spirals , this placement has been questioned by van den bergh ( 1998 , 2009a ) .
based on a statistical analysis of types given in the rsa , van den bergh showed that s0 galaxies are typically 0.8 - 1.0 mag less luminous than e and sa galaxies , implying that s0 galaxies , on the whole , can not really be considered intermediate between e and sa galaxies .- band magnitudes of a well - defined sample of s0s are similar to those of early - type spirals in the osubsgs sample .
the sample was mostly drawn from rc3 and includes some galaxies classified as ellipticals in rc3 and as s0s in the rsa ( see section 12 ) . ]
the preponderance of s0 galaxies in clusters led to the early suggestion ( e.g. , spitzer & baade 1951 ; gunn & gott 1972 ; moore et al .
1996 ) that some type of external environmental interaction was responsible for stripping a spiral galaxy of its interstellar medium .
if this actually occurred , then , as suggested by van den bergh ( 2009a ) , this could imply that s0 galaxies have lost a substantial fraction of their spiral mass due to interactions .
alternatively , van den bergh ( 2009b ) argues that stripping of a lower luminosity , late - type spiral should be easier than stripping of a higher luminosity , early - type spiral , which could also account for the luminosity difference . in an examination of
the environment of s0 galaxies , van den bergh ( 2009b ) found no significant difference in the average luminosities , flattenings , or distribution of s0 subtypes in clusters , groups , or the field , indicating that some s0s develop as a result of internal effects , such as the influence of an active galactic nucleus .
barway et al .
( 2011 ) noted that lower luminosity s0s have a higher bar fraction than higher luminosity s0s ( 83% versus 17% ) , suggesting that the two groups form in different ways ( see also barway et al .
these authors suggest that faint s0s are stripped late - type spirals , which are known to have a high bar fraction ( barraza et al .
2008 ) .
environmental effects in clusters do not always have to involve drastic transformations in morphology .
sometimes the effects are more subtle .
figure [ virgotypes ] shows several spiral galaxies that are also members of the virgo cluster .
these galaxies highlight processes that affect the star - forming disk while leaving the older stellar disk relatively unaffected .
ngc 4580 and 4689 are galaxies having a patchy inner disk and a smooth outer disk , called virgo types " by van den bergh et al .
these objects suggest that the environment of such galaxies has somehow truncated the star - forming disk , with a greater concentration of truncated disks toward the cluster core .
similar results are obtained from observations of the hi gas disks of virgo cluster galaxies ( e.g. , giovanelli & haynes 1985 ; cayatte et al .
1994 ; chung et al .
2009 ) .
koopmann & kenney ( 2004 ) summarize the results of an extensive survey of h@xmath51 emission from virgo cluster galaxies , and identify different categories of environmentally - influenced star formation characteristics based on h@xmath51 imaging .
the blue light images of examples of each category are included in figure [ virgotypes ] , and show how the subtleties are manifested in regular morphology . using a sample of isolated spiral galaxies to define normal
" star formation , koopmann and kenney defined several categories of virgo spiral galaxy star - forming disks : category
n " refers to disks whose star formation is within a factor of three of the normal levels .
e " cases have star formation enhanced by more than a factor of three compared to normal .
a " cases are anemic " spirals ( section 6.4 ) having star formation reduced by more than a factor of 3 compared to normal .
t / n " refers to galaxies where the star - forming disk is sharply cut off , but inside the cutoff , the star formation levels are normal ( the [ s ] means truncation is severe ) .
one of these , ngc 4580 , is so unusual that sandage & bedke ( 1994 ) classify it as sc(s)/sa , where the sc part is the inner disk and the sa part is the outer disk . in
t / a galaxies " , the inner star - formation is at a low level , as in anemic cases , while in t / c " galaxies , most of the star formation is confined within the inner 1 kpc .
koopmann and kenney found that the majority of virgo cluster spiral galaxies have truncated star - forming disks .
the idea is that the interstellar medium ( ism ) of a cluster galaxy can interact with the intra - cluster medium ( icm ) , stripping the ism ( via ram pressure ; gunn and gott 1972 ) but leaving the stellar disk intact .
truncated gas and star - forming disks result because ram - pressure stripping is more severe in the outer parts of galaxies ( e. g. , book and benson 2010 and references therein ) . in virgo ,
most galaxies with truncated star - forming disks have relatively undisturbed stellar disks and normal to slightly enhanced inner disk star formation rates , suggesting that icm - ism stripping is the main mechanism in the reduction of their star formation rates .
the cases found to have relatively normal or enhanced star formation rates are preferentially located in the outer parts of the cluster and likely have never visited the core region .
only galaxies which go near the center get significantly stripped .
however , tidal effects also contribute to morphological changes .
several galaxies , including many of the t / c class , display peculiarities consistent with tidal effects , such as nonaxisymmetric circumnuclear star formation , shell features ( e.g. , ngc 4424 in figure [ virgotypes ] ) , and enhanced inner star formation rates .
a recent study by yagi et al . ( 2010 ) provides dramatic and clear evidence of disk gas stripping in galaxies thought to be relatively new arrivals to the core region of the coma cluster . using deep h@xmath51 imaging
, these authors detected ionized gas in clouds that are mostly outside the main disk of a dozen coma galaxies .
three distinct morphologies of the distributions of these clouds were found : ( 1 ) connected clouds that blend with disk star formation ; ( 2 ) long , connected lines of clouds that extend from a central gas knot but are not related to the disk light ; and ( 3 ) clouds completely detached from the main disk .
examples of these categories are illustrated in figure [ coma ] .
yagi et al . interpret them in terms of an evolutionary gas - stripping sequence where category ( 1 ) galaxies are in an earlier phase of stripping while the category ( 3 ) galaxies are in the most advanced phase .
it is likely that large disk galaxies in coma would be completely stripped eventually because of the cluster s high icm density and broad velocity distribution .
the same process seen in coma likely occurs in virgo but is only partial for the large spirals owing to the lower icm density and velocities in virgo ( koopmann & kenney 2004 ) .
it is perhaps fitting that the first major spiral galaxy discovered was in the interacting pair m51 ( section 2 ) .
numerical simulations ( salo & laurikainen 2000a , b ) have shown that both parabolic and bound passages of the companion , ngc 5195 , can explain the observed morphology and other characteristics of the system .
it turns out that m51 defines a class of interacting systems known as m51-type pairs .
figure [ pec_es ] shows an example in the pair ngc 2535 - 6 . in each case
, the larger component has a strong two - armed spiral , with one arm appearing drawn " to the smaller companion .
an extensive catalogue of m51-type pairs is provided by jokimaki et al .
( 2008 ) .
other distant encounters can produce tidal tails or bridges of material between galaxies ( top row , figure [ pec_es ] ) .
ngc 4676 , also known as the `` mice , '' is a pair of strongly interacting galaxies where a very extended tidal tail has formed in one component .
the strongly - interacting pair ngc 5216/18 has developed a bright connecting bridge of material , and each component shows tidal tails .
the evolution of this system , and the role of encounters on bar formation , is described by cullen et al .
( 2007 ) .
figure [ pec_es ] also shows several examples of morphologies that may result from minor mergers of a small galaxy with a more massive , pre - existing elliptical galaxy .
bertola ( 1987 ) brought attention to the unusual class of _ dust - lane ellipticals _ , where an otherwise normal elliptical galaxy shows peculiar lanes of obscuring dust .
it was de vaucouleurs s personal view that if an elliptical shows dust , then it s not an elliptical !
" however , bertola showed that an unusual case like the radio elliptical galaxy ngc 5128 , where a strong dust lane lies along the _ minor axis _ of the outer light distribution , is simply the nearest example of a distinct class of objects .
further study showed that dust - lane ellipticals come in several varieties .
the minor axis dust lane type appears most common , but cases of alignment along the major axis of the outer isophotes ( major axis dust lanes ) as well as cases of misalignment are also known ( see the upper left panels of figure [ pec_es ] ) .
the origin of these very regular dust lanes is thought to be a merger of a gas - rich companion ( e.g. , oosterloo et al .
the regularity of the dust lanes suggests that the mergers are in advanced states . the two lower left frames of figure [ pec_es ] show examples of galaxies having shells , " , or faint , arc - shaped brightness enhancements of varying morphology .
they were first discovered on deep photographs by malin ( malin & carter 1980 ) , and appeared to be associated mainly with elliptical galaxies .
in fact , the first examples , ngc 1344 and 3923 , are classified in catalogues as ordinary ellipticals because the shells are not detectable on photographs of average depth .
once the class was recognized , a detailed search led to other examples which were listed by malin & carter ( 1980 , 1983 ) .
the term shells " implies a particular three - dimensional geometry that schweizer & seitzer ( 1988 ) argued imposes a prejudice on the interpretation of the structures .
they proposed instead the alternate term ripples , " which implies less of a restrictive geometry .
the explanation of shell / ripple galaxies is one of the great success stories in galactic dynamics ( see review by athanassoula & bosma 1985 ) .
shells are thought to be remnants of a minor merger between a massive elliptical and a lower mass disk - like galaxy .
the main requirements are that the disk - shaped galaxy be cold " , or lack any random motions , and that the potential of the elliptical galaxy should be rigid , meaning the elliptical is much more massive than its companion .
the smaller galaxy s stars fall into the center of the galaxy and phase wrap , or form alternating outward - moving density waves made of the disk galaxy s particles near the maximum excursions of their largely radial orbits in the rigid potential .
many , but not all , of the main properties of shell es can be explained by this model .
other issues concerning shell galaxies are reviewed by kormendy and djorgovski ( 1989 ) .
taylor - mager et al .
( 2007 ; see their figure 2 ) have proposed a simple classification of interacting systems that highlights different interaction classes .
a pre - merger ( type pm ) includes two interacting galaxies that are sufficiently far apart to suffer little apparent distortion .
a minor merger ( mm ) is two galaxies showing evidence of merging , but one component is much smaller than the other . a major merger ( m ) has two comparable brightness galaxies in the process of merging , while a merger remnant ( mr ) is a state sufficiently advanced that the merging components are no longer distinct .
related to interacting systems are the infrared - bright galaxies first identified by rieke & low ( 1972 ) based on 10@xmath42 m photometry . from studies based on the infrared astronomical satellite ( iras ) , sanders & mirabel ( 1996 ) classified a galaxy as a luminous infrared galaxy " ( lirg ) if its luminosity in the 8 - 1000@xmath42 m range is between 10@xmath52 and 10@xmath53 l@xmath54 .
if the luminosity in the same wavelength range exceeds 10@xmath53 @xmath55 , then the object is called an ultra - luminous infrared galaxy " ( ulirg ) .
detailed studies have shown that at high redshifts , lirgs and ulirgs are a dominant population of objects ( see discussion in pereira - santaella et al .
2010 ) .
the morphologies of nine ulirgs were studied using hst @xmath21 and @xmath56-band images by surace et al .
their montage of six of these objects is shown in figure [ ulirgs ] . in every case
there are clear signs of interactions , and all are likely linked to mergers or mergers in progress .
several , like mk 231 , have bright seyfert nuclei .
arribas et al .
( 2004 ) obtained extensive imaging of local lirgs , and found a similar high proportion of strongly interacting and merging systems .
the merger rate is considered one of the most important parameters for understanding galaxy evolution .
it has been difficult to estimate , and issues connected with it are discussed by jogee et al .
( 2009 ; see also conselice 2009 ) . a merger is considered major if it involves a companion ranging from 1/4 to approximately equal mass to the main galaxy .
a major merger of two spiral galaxies can destroy both disks and lead to an @xmath57 law profile remnant through violent relaxation .
minor mergers involve companions having 1/10 to 1/4 the mass of the main galaxy . both types of mergers , while in progress , can lead to many specific morphological features such as highly distorted shapes , tidal tails and bridges , shells and ripples , and warps .
even some bars and spiral patterns are thought to be connected to interactions , and especially galaxies with a double nucleus are thought to be mergers .
as discussed in section 10.1 , mergers or collisions may also be at the heart of rare morphologies such as ring and polar ring galaxies . using visual classifications of merger types ,
jogee et al . (
2009 ) estimate that 16% of high mass galaxies have experienced a major merger , while 45% have experienced a minor merger , during the past 3 - 7gyr ( @xmath1=0.24 - 0.80 ) .
a warp is an apparent bend or slight twist in the shape of the disk of a spiral galaxy ( see sellwood 2010 and references therein ) . in a warp , stars and gas clouds move in roughly circular orbits , but the orientation of these orbits relative to the inner disk plane changes with increasing radius .
warps are most easily detected in edge - on galaxies because the bending of the outer orbits makes the galaxy look like an integral sign .
although often most pronounced in an hi map , warps can be seen in ordinary optical images of edge - on galaxies .
figure [ warps ] shows three galaxies having strong optical warping of the disk plane . in two of the galaxies ,
the bright inner disk is unwarped , while a fainter and thicker outer disk zone is twisted relative to the inner disk . in ugc 3697
, the warping is exceptionally visible . in general ,
optical warping is less severe than hi warping .
warping is a very common aspect of spiral galaxies ( e.g. , binney 1992 ) and has been interpreted in terms of perturbations ( gaseous infall or interactions ) that trigger bending instabilities ( e.g. , revaz & pfenniger 2007 ) .
garcia - ruiz , sancisi , & kuijken ( 2002 ) estimated hi warp angles , the angle between the inner disk plane and the assumed linear warping zone , to range from nearly 0@xmath24 to more than 30@xmath24 .
a useful summary of previous warp studies is provided by saha et al .
( 2009 ) , who examine warp onset radii in mid - ir images .
the theory of warps is reviewed by sellwood ( 2010 ) .
the morphology of active galaxies ( also called excited " galaxies by van den bergh 1998 ) is important to consider because of a possible link between morphological features and the fueling of the active nucleus .
early studies showed a preponderance of ring , pseudoring , and bar features in seyfert galaxies that suggested the link was bar - driven gas flow ( simkin , su , & schwarz 1980 ) .
several examples of the morphology of seyfert and other active galaxies are shown in figure [ actives ] .
the activity classifications are based mainly on spectroscopy , not on morphology , and are described by veron - cetty & veron ( 2006 ) .
a detailed study of active galaxy morphologies by hunt & malkan ( 1999 ) provided similar results to the early studies .
these authors examined the morphologies of a large sample of galaxies selected on the basis of their 12@xmath42 m emission , and found that outer rings and inner / outer ring combinations are 3 - 4 times higher in seyfert galaxies than in normal spirals .
in contrast , bars were found to occur with the same frequency ( @xmath269% ) in seyferts as in normal spirals , while for hii / starburst galaxies , the frequency was much higher ( @xmath380% ) .
although outer rings are found mostly in barred galaxies , bars do not promote the nuclear activity of seyfert galaxies .
hunt & malkan ( 1999 ) interpret this inconsistency in terms of timescales : it takes roughly 3@xmath5810@xmath59 yr for a closed outer ring to form , a timescale during which a bar may weaken or dissolve .
because of this , a high ring frequency in seyferts would indicate an advanced evolutionary state .
related to the same issue , comern et al .
( 2010 ) found that nuclear rings do not correlate with the presence of nuclear activity .
the study of hunt & malkan ( 1999 ) used mostly rc3 classifications to deduce the bar fraction in active galaxies .
these visual classifications are based on blue light images and hence dust could effectively obscure some bars .
knapen et al .
( 2000 ) used high resolution near - ir images of well - defined samples and quantitative bar detection methods to deduce that bars are more frequent in seyfert galaxies than in a control sample of non - active galaxies : 79% @xmath39 7.5% versus 59% @xmath39 9% ( see also laine et al .
laurikainen , salo , & buta ( 2004 ) came to a similar conclusion for 180 galaxies in the osubsgs , based on near - ir @xmath43-band images .
the former studies used ellipse fits to identify bars , while laurikainen et al . used fourier analysis .
mckernan , ford , & reynolds ( 2010 ) also consider outer rings and pseudorings as probes of models of agn fueling from interactions and mergers .
the idea is that a closed outer ring takes a long time to form and is very fragile , being sensitive to interactions and changes in the bar pattern speed ( e.g. , bagley et al . 2009 ) .
an interaction can change a closed outer ring into a pseudoring and could possibly destroy the ring .
thus , rings are probes of the interaction history of active galaxies .
mckernan et al . found no difference between the agn found in ringed galaxies and those found in galaxies without rings .
but in those with rings , recent interactions can be ruled out and activity may be tied to short - term internal secular evolutionary processes .
bahcall et al . ( 1997 ) presented hst images of 20 luminous , low redshift quasars observed with a wide @xmath7-band filter . figure [ quasars ] is reproduced from their paper and shows images of the host galaxies after removal of most of the quasar light .
the images show a variety of morphologies , including ellipticals , interacting pairs , systems with obvious tidal disturbances , and normal - looking spirals .
an example of the latter is pg1402 + 261 ( @xmath1=0.164 ) , which is type ( r@xmath33)sb(rs)a based partly also on the image shown in figure 7 of bahcall et al . based on the number of hosts showing signs of interactions , as well as the number of companions , bahcall et al .
conclude that interactions may trigger the quasar phenomenon .
matthews , morgan , & schmidt ( 1964 ) observed the optical morphologies of the radio sources identified with the brightest members of rich galaxy clusters .
they found that the most common form was what morgan ( 1958 ) called a d " galaxy , meaning a galaxy having an elliptical - like inner region surrounded by an extensive envelope ( see discussion in van den bergh 1998 ) .
although these superficially resembled hubble s s0s , none were found having a highly - elongated shape , implying that the galaxies are not as highly flattened as typical s0s .
another characteristic of the cluster d galaxies was their very large linear size and exceptional luminosity , much larger than a typical cluster member . to denote these extreme objects in the morgan system
, the prefix c was added as in the old classification of supergiant stars . even today
, morgan s notation cd " is used to describe these supergiant galaxies which are generally considered outside the scope of the hubble system .
the most detailed study of the photometric properties of brightest cluster members ( bcms ) was made by schombert ( 1986 , 1987 , 1988 ; see also various references therein ) .
the main bcm types schombert considered were ge ( giant ellipticals ) , d , and cd , distinguished mainly from the appearance of profile shape .
d galaxies are larger and more diffuse with shallower profiles than ge galaxies , while a cd galaxy is the same as a d galaxy but with a large extended envelope ( schombert 1987 ) .
cd envelopes can extend to 500 kpc or more .
kormendy & djorgovski ( 1989 ) argue that only cd galaxies are sufficiently physically distinct from ellipticals to merit being a separate class , and recommended that the d " class not be used .
two cd galaxies and two ge galaxies are shown in figure [ bcms ] .
to give an idea of the scale of these objects , the vertical dimension of the frames corresponds to 201 , 232 , and 132 kpc for ( left to right ) ugc 10143 ( a2152 ) , ngc 4874/89 ( a1656 ) , and ngc 6041 ( a2151 ) , respectively .
the cd classification of ngc 4874 is due to schombert ( 1988 ) , and one can see in figure [ bcms ] that it is much larger and has a shallower brightness profile than nearby ngc 4889 .
the cd envelope is detected as an excess of light in the outer regions relative to a generalized brightness profile , and may not even be the light that leads to the visual classification of cd .
based on structural deviations such as the large radii , shallow profile slopes , and bright inner regions , schombert ( 1987 ) concluded that bcms fit well with the predictions of merger simulations , including accretion and cannibalism " of smaller cluster members .
properties of cd envelopes ( as separated photometrically from the parent galaxy ) may suggest a stripping process for their formation ( schombert 1988 ) .
as noted in section 5.1 , many bcms in rc3 received the classification e@xmath4 , suggesting that the characteristic brightness profiles give a hint of an envelope interpreted as an incipient disk .
the distribution of axial ratios of cds actually is flatter on average than normal ellipticals ( schombert 1986 ) , but it is not clear that the perceived envelopes in bcm e@xmath4 galaxies are actually as flattened as a typical disk .
a local example of a ge galaxy is m87 , classified as type e@xmath40 - 1 by de vaucouleurs .
the standard waveband for galaxy classification , the @xmath21-band , is sensitive enough to the extreme population i component that the degree of resolution of spiral arms into star - forming complexes is part of the classification .
the @xmath21-band , however , also includes a substantial contribution from the older stellar background .
one way to isolate only the star - forming regions in a galaxy is imaging in h@xmath51 , which traces hii regions . apart from showing the distribution of star formation ( modified by extinction )
, h@xmath51 imaging also traces the rate of global photoionization , which in turn directly traces the rate of formation of stars more massive than about 10@xmath60 ( kennicutt , tamblyn , & congdon 1994 ) .
the initial mass function ( imf ) , either assumed or constrained in some way ( using , for example , broadband colors ) , allows h@xmath51 fluxes to be converted to the total star formation rate over all stellar masses .
h@xmath51 imaging often shows well - organized patterns of hii regions that follow structures like spiral arms and especially rings .
images of six early - to - intermediate - type ringed galaxies are shown in figure [ halpha ] ( crocker , baugus , & buta 1996 ) .
the way h@xmath51 imaging usually works is a galaxy is imaged in or near its redshifted h@xmath51 wavelength , and then in a nearby red continuum wavelength .
the net h@xmath51 image is the difference between the h@xmath51 image and the scaled red continuum image .
often , the h@xmath51 filter used is broad enough to include emission from [ nii ] 6548 and 6584 .
for each of the galaxies shown in figure [ halpha ] , the left image is the red continuum image , while the right image is the net h@xmath51 image .
of the four barred spirals shown ( ngc 1433 , 7329 , 6782 , and 7267 ) , three show no hii regions associated with the bar .
these three ( ngc 1433 , 7329 , and 6782 ) all have conspicuous inner rings which are lined with hii regions .
as shown by crocker , baugus , & buta ( 1996 ) , the distribution of hii regions around inner rings is sensitive to the intrinsic shape of the ring .
when the ring is highly elongated , the hii regions concentrate around the ends of the major axis ( which coincide with the bar axis ; ngc 6782 ) , while when the ring is circular , the hii regions are more uniformly spread around the ring ( ngc 7329 ) .
inner ring shapes do not correlate well with maximum relative bar torques in a galaxy ( dva ) , but grouchy et al .
( 2010 ) have shown that when local forcing is considered instead , ring shapes and bar strengths are well - correlated .
the case of ngc 7267 is different in that most of its h@xmath51 emission is concentrated in the bar .
martin & friedli ( 1997 ) have argued that star formation along galactic bars could provide clues to gas flow in the inner regions of galaxies and the fueling of starbursts and agn .
these authors present models which suggest an evolution from a pure h@xmath51 bar , to an h@xmath51 bar with ionized gas in the center , to a gas - poor bar with strong nuclear or circumnuclear star formation .
this suggests that the bar of ngc 7267 is younger than those in ngc 1433 , 7329 , and 6782 .
the two other galaxies shown in figure [ halpha ] are nonbarred or only weakly - barred .
ngc 6935 is an interesting case where a nonbarred galaxy has a strong ring of star formation .
grouchy et al .
( 2010 ) found that the star formation properties of inner rings , but not the distribution of hii regions , are independent of the ring shapes and bar strengths in a small sample .
the case of ngc 7702 is different from the others .
this is a late s0 ( type s0@xmath4 ) showing a very strong and largely stellar inner ring .
the ring shows little ionized gas and appears to be in a quiescent phase of evolution .
the best global imaging of nearby galaxies at ultraviolet wavelengths has been obtained with the _ galaxy evolution explorer _ ( galex , martin et al .
2005 ) , which provided detailed images of galaxies of all types at wavelengths of 0.225 and 0.152@xmath42 m .
these images reveal young stars generally less than @xmath2100 myr old but are affected by dust extinction .
there is a strong correlation between the uv morphology and the h@xmath51 morphology ( figure [ nuv - halpha ] ) . like h@xmath51 , uv fluxes from galaxies can be used to estimate star formation rates once extinction is estimated , and are particularly sensitive to the ratio of the present to the average past star formation rate .
uv imaging is an effective way of decoupling the recent star formation history of a galaxy from its overall , long - term star formation history .
figure [ nuv ] shows near - uv ( 0.225@xmath42 m ) images of eight galaxies over a range of hubble types .
the two earliest types , ngc 1317 and 4314 ( both sa ) show a near - uv morphology dominated by a bright circumnuclear ring of star formation .
the sb(r)b galaxy ngc 3351 shows a conspicuous inner ring of star formation and little emission from its bar region .
in contrast , the sbc galaxy ngc 7479 shows strong near - uv emission from its bar .
the late - type galaxies ngc 628 [ type sa(s)c ] and ngc 5474 [ type sa(s)m ] are typical of their types , but most interesting is ngc 4625 .
a key finding of galex was extended uv emission well beyond the optical extent of some galaxies .
ngc 4625 is an example where the main optical part [ type sabm ] is only a small fraction of the extent of the uv disk ( gil de paz et al .
2005 ) .
the final object shown in figure [ nuv ] is ngc 5253 , a basic example of what de vaucouleurs classified as i0 ( section 5.3 ) .
the inner region is a bright boxy zone of star formation , and even the extended disk is prominent .
related to star formation morphologies are the distributions of atomic and molecular gas . far from being randomly - distributed clouds of interstellar material , atomic and molecular gas morphologies
can be highly organized , well - defined patterns .
recent high quality surveys have provided some of the best maps of these distributions in normal galaxies .
atomic hydrogen is mapped with the 21 cm fine structure emission line , which has the advantage of not being affected by extinction and for being sufficiently optically thin in general to allow total hi masses to be derived directly from hi surface brightness maps .
in addition , hi maps provide information on the kinematics and dynamics of the ism , as well on the existence and distribution of dark matter in galaxies .
molecular hydrogen is generally mapped using the @xmath53co j=1 - 0 rotational transition at a wavelength of 2.6 mm , under the assumption that co and hydrogen mix in a roughly fixed proportion .
although numerous maps have been made of the hi distribution in nearby galaxies , the most sophisticated and detailed survey made to date is the the hi nearby galaxies survey " ( things , walter et al .
the earliest surveys had shown that hi is a tracer of spiral structure in galaxies , and the things provides some of the highest quality maps revealing this correlation as well as other characteristics . from a morphological point of view , hi maps tend to reveal : ( 1 ) enhanced surface brightness in star - forming features such as spiral arms , rings , and pseudorings ; ( 2 ) extended gaseous disks , such that the hi extent can exceed the optical extent by several times ; and ( 3 ) supernova - driven and star - formation driven , wind - blown holes .
figure [ himaps ] shows the hi morphologies of eight things galaxies ranging from the sab galaxy ngc 4736 to the sm galaxy ddo 50 ( holmberg ii ) .
bright sc galaxies like ngc 628 ( m74 ) and ngc 5236 ( m83 ) show hi distributions that extend well beyond the optical disks .
these extended patterns can include large spirals as in ngc 628 .
in m81 and m83 , the hi traces the optical spiral structure well .
large rings or pseudorings are seen in ngc 2841 and ngc 4258 , while m81 shows an intermediate - scale ring of gas that is much less evident optically .
the bright star - forming inner ring in ngc 4736 is well - defined and easily distinguished in hi , but the galaxy s diffuse stellar outer ring is more of an asymmetric spiral zone .
most interesting in hi maps are the obvious holes where there appears to be a deficiency of neutral gas compared to surrounding regions .
especially large holes are seen in the hi morphology of the late - type dwarf ddo 50 .
these holes are thought to be regions cleared by the stellar winds and explosions of massive stars contained or once contained within them .
the holes are 100pc to 2kpc in size , and have different systematic properties in early and late - type galaxies in the sense that holes may last longer in late - type dwarfs owing to the lack of serious shear due to strong differential rotation ( bagetakos et al . 2009 ) .
holes may also be found outside the standard isophotal optical angular diameter .
the dwarf galaxy ddo 154 shown in figure [ himaps ] has one of the largest ratios of hi to optical diameter , a factor of 6 at least according to carignan & purton ( 1998 ) , who also estimated that 90% of the mass of the galaxy is in the form of dark matter . the hi and optical morphologies bear little resemblance to one another .
another example of strongly uncorrelated hi and optical morphologies is ngc 2915 , a very gas - rich galaxy whose optical morphology is a blue compact dwarf while its at least 5 times larger hi morphology includes a prominent outer spiral with no optical counterpart ( meurer et al .
1996 ) , leading to the concept of a purely dark spiral . "
bertin & amorisco ( 2010 ) consider a general intepretation of such large outer hi spirals , especially those seen in galaxies like ngc 628 and ngc 6946 : the spirals represent short trailing waves that carry angular momentum outwards from corotation and , at least in the gaseous component , penetrate a normal barrier at the olr of the stellar disk pattern and go far out into the hi disk .
the short trailing waves are thought to excite the global spiral arms seen in the main optical body of the galaxy ( where the prominent optical spirals of ngc 628 and ngc 6946 are found ) .
how ngc 2915 fits into this picture is unclear since the main stellar body of this galaxy is not a grand - design spiral .
galaxies whose hi disks do not extend much beyond the optical light distribution are also of interest .
ngc 4736 in figure [ himaps ] is an example .
the large , nearby ringed barred spiral ngc 1433 was found by ryder et al .
( 1996 ) to have neutral hydrogen gas concentrated in its central area , its inner ring , and in its outer pseudoring , with no gas outside the visible disk light and a lower amount of gas in the bar region compared to the ring regions .
higdon et al .
( 1998 ) showed that in the ringed , barred spiral ngc 5850 , neutral gas is concentrated in the inner ring and in the asymmetric outer arm pattern , with little or no emission detected outside this pattern .
the asymmetry led higdon et al . to propose that ngc 5850 has possibly experienced a high speed collision with nearby ngc 5846 . in galaxy clusters ,
it is well - known that environmental effects can truncate an hi disk so that it is _ smaller _ than the optical disk light .
this is dramatically illustrated in the high resolution vla hi maps of virgo cluster galaxies by chung et al .
( 2009 ) , who found that galaxies within 0.5 mpc of the cluster core have severely truncated hi disks typically less than half the size of the optical standard isohotal diameter , @xmath61 .
a variety of earlier studies had already shown these galaxies to be hi - deficient compared to field galaxies of similar types .
as noted in section 10.2 , an interaction between a cluster galaxy s ism and the intra - cluster medium can account for these unsual modifications of hi morphology .
chung et al .
also provide evidence for this interaction in some morphologies that appear to be gas stripping in progress .
the berkeley - illinois - maryland survey of nearby galaxies ( bima song ; regan et al . 2001 ; helfer et al .
2003 ) provided some of the highest quality co maps of normal galaxies .
co emission is often seen in intermediate ( sab - sd ) galaxies , which have a high enough gas content and metallicity to allow the @xmath53co j=1 - 0 2.6 mm emission line to be detectable .
the co distributions of eight such galaxies from this survey are shown in figure [ comaps ] .
these display some of the range of molecular gas morphologies seen .
co traces the inner spiral arms of ngc 628 , 1068 , and 4535 , and is seen along the bar of ngc 2903 .
a common co morphology is a large - diameter ring of giant molecular clouds ( gmcs ) , without a central co concentration , as seen in ngc 2841 and 7331 .
the rings are the peaks of exponentially - declining distributions .
the coherent inner molecular gas ring in ngc 7331 appears more like a typical resonance ring , and has an estimated molecular gas mass of 3.4@xmath5810@xmath59 @xmath60 ( regan et al .
the co distribution in ngc 2403 appears to be concentrated in individual gmcs , with little diffuse emission , while that in ngc 3351 is characterized by a small central bar aligned nearly perpendicular to the galaxy s primary bar .
helfer et al . ( 2003 ) show that the milky way , m31 , and m33 have co morphologies that are consistent with the range of morphologies found by bima song .
infrared observations have considerable advantages over optical observations of galaxies . while traditional @xmath21-band images are sensitive to dust and extinction ( both internal and external ) , the effects of extinction in the near- and mid - ir are much less , and become virtually negligible at 3.6@xmath42 m .
spiral galaxies imaged at wavelengths successively longer than @xmath21-band become progressively smoother - looking , not only due to the reduced effect of extinction , but also to the de - emphasis of the young blue stellar component .
the combination of these effects has led to the popular idea that ir imaging reveals the stellar backbone " of galaxies , i. e. , the distribution of actual stellar mass ( e. g. , rix & rieke 1993 ; block et al .
thus , infrared imaging has become a staple for studies of the gravitational potential and stellar mass distribution in galaxies ( e. g. , quillen , frogel , & gonzlez 1994 ) , and for the quantification of bar strength from maximum relative gravitational torques ( e.g. , buta & block 2001 ; laurikainen & salo 2002 ) . infrared imaging has also revealed interesting outer structures such as the large outer red arcs seen in m33 , which have been interpreted by block et al .
( 2004b ) to be swaths of extremely luminous carbon stars formed from external accretion of low metallicity gas .
power spectrum analysis of the ir structures in classical spirals has been used to detect azimuthal star streams " and to evaluate the role of turbulence on star formation and spiral structure ( block et al .
2009 and references therein ) . near - infrared imaging from 0.8 - 2.2@xmath42 m can be successfully obtained from groundbased observatories but with the serious drawback that the brightness of the sky background increases substantially over this range . as a result
, it has not been possible to achieve a depth of exposure at , for example , 2.2@xmath42 m comparable to the kinds of depths achievable at optical wavelengths without excessive amounts of observing time . the first major near - ir survey designed for large - scale morphological studies was the ohio state university bright spiral galaxy survey ( osubsgs , eskridge et al .
2002 ) , which included optical @xmath62 and near - ir @xmath63 images of 205 bright galaxies of types s0/a to sm in a statistically well - defined sample selected to have total blue magnitude @[email protected] and isophotal diameter @xmath61@xmath656@xmath315 .
this survey allowed a direct demonstration of how galaxy morphology actually changes from optical to near - ir wavelengths , not merely for a small , selected sample of galaxies , but for a large sample covering all spiral subtypes .
the main near - ir filter used in this survey was the @xmath43-band at 1.65@xmath42 m .
the osubsgs was later complemented by the _ near - infrared s0 survey _
( nirs0s , laurikainen et al .
2005 , 2006 , 2007 , 2009 , 2010 ; buta et al .
2006 ) , a @xmath66-band imaging survey of 174 early - type galaxies in the type range s0@xmath5 to sa , but mostly including s0s , some of which were misclassified as ellipticals in rc3 .
nirs0s images are deeper than osubsgs near - ir images owing to the use of larger telescopes and longer on - source times .
although s0 galaxies are dominated by old stars and are usually smooth even in blue light images , the @xmath66 band was chosen to complement the osubsgs sample of spirals in order to make a fair comparison between bar strengths and bulge properties of s0s and spirals .
also , s0 galaxies are not necessarily dust - free , and near - ir imaging is still necessary to penetrate what dust they have .
nirs0s has led to several important findings about s0 galaxies : ( 1 ) a class of s0s , not previously recognized , having prominent lenses but very small bulges that are more typical of sc galaxies than of earlier type spirals ( example : ngc 1411 , laurikainen et al .
2006 ) ; ( 2 ) considerable evidence that s0 galaxies have pseudobulges just as in many spirals ( laurikainen et al .
while the bulges of the latter are likely to be made of rearranged disk material in many cases ( section 9 ) , those in s0s are likely to be related to the evolution of bars .
s0 bulges tend to be nearly exponential ( sersic index @xmath16 @xmath65 2 ) , are supported against gravity by rotation rather than random motions , and often include clear inner disks ; ( 3 ) good correlations between bulge effective radii , @xmath10 , and disk radial scalelength , @xmath67 , as well as between the @xmath66-band absolute magnitudes of the bulge and disk , suggest that s0 bulges are not formed from hierarchical mergers , implying that s0s could be stripped spirals , although the lower bar fraction in s0s suggest that this is in conjunction with evolution due to bars and ovals ; ( 4 ) 70% of s0-s0/a galaxies have ovals or lenses , suggesting that bars have been weakened in such galaxies over time ; and ( 5 ) not only bulges , but also disks of s0s are similar to those in s0/a - scd spirals . the _ two - micron all - sky survey _
( 2mass , skrutskie et al . 2006 ) provided near - ir @xmath68 images of a much larger galaxy sample than either the osubsgs or nirs0s , although these images lack the depth of the osubsgs and nirs0s images in general .
2mass provided considerable information on near - infrared galaxy morphology , which led to the extensive _ 2mass large galaxy atlas _
( jarrett et al . 2003 ) .
the best imaging of galaxies at mid - ir wavelengths has been obtained with the _
spitzer space telescope _ using the infrared array camera ( irac , fazio et al .
2004 ) and 3.6 , 4.5 , 5.8 , and 8.0@xmath42 m filters .
the 3.6 and 4.5@xmath42 m filters provide the most extinction - free views of the stellar mass distribution in galaxies , while the 5.8 and 8.0@xmath42 m filters reveal the interstellar medium ( ism ) ( pahre et al .
the loss of coolant in 2008 prevented further observations with the 5.8 and 8.0@xmath42 m filters , but the 3.6 and 4.5@xmath42 m filters could still be used .
this led to the _ spitzer survey of stellar structure in galaxies _ ( s@xmath69 g , sheth et al .
2010 ) , a 3.6 and 4.5@xmath42 m survey of 2,331 galaxies of all types closer than 40mpc .
these wavelengths sample the rayleigh - jeans decline of the stellar spectral energy distribution of all stars hotter than 2000k .
s@xmath69 g images shown here are from buta et al .
( 2010a ) and are based on pre - survey archival images processed in the same manner as survey images . _
spitzer _ observations have a very low background compared to groundbased near - ir observations , and thus irac images are the deepest galaxy images ever obtained in the ir
. figure [ m51_frames ] compares images of m51 at four wavelengths : the galex 0.15@xmath42 m band , the @xmath21-band ( 0.44@xmath42 m ) , the near - infrared @xmath66 band ( 2.2@xmath42 m ) , and the irac 3.6@xmath42 m band .
only the @xmath21 and @xmath66-band images are from groundbased observations .
the galex image reveals the extensive star formation in the spiral arms , and the complete absence of star formation in the companion ngc 5195 as well as in the complex tidal material north of the companion .
the star formation in the arms is more subdued in the @xmath21-band , and almost completely subdued in the @xmath66-band .
the arms are so smooth in the @xmath66 band that the galaxy resembles an sa or sab system .
( the @xmath21-band type is sbc . )
surprisingly , this is not the case in the 3.6@xmath42 m image whose considerably greater depth compared to the @xmath66-band image is evident .
the spiral arms in the 3.6@xmath42 m band are lined by numerous resolved objects , many of which are correlated with the star - forming regions seen in the @xmath21-band .
this is dramatically seen also in the sb(s)cd galaxy ngc 1559 ( figure [ ngc1559 ] ) , where an irac 3.6@xmath42 m image is compared with a @xmath21-band image .
these show that resolved features in the deep 3.6@xmath42 m image are well - correlated with @xmath21-band star - forming complexes .
thus , mid - ir 3.6@xmath42 m images are _ not _ completely free of the effects of the extreme population i stellar component ( see discussions in block et al .
2009 and buta et al .
2010a ) .
a sampling of s@xmath69 g images as compared to @xmath21-band images for the same galaxies from the dva is shown in figure [ nir01 ] .
the four galaxies shown , ngc 584 , 1097 , 628 , and 428 , have dva types of s0@xmath5 , sbb , sc , and sm , respectively , thus covering almost the entire hubble - de vaucouleurs sequence .
although the very dusty interacting system ngc 1097 looks slightly earlier " at 3.6@xmath42 m , these images show again that on the whole the morphology in the two wavebands is very similar .
the same is seen for other galaxies described by buta et al .
( 2010a ) , who found that 3.6@xmath42 m types , judged using the same precepts described in the dva for blue light images , are well - correlated with blue - light types . on average mid - ir classifications for rc3
s0/a - sc galaxies are about 1 stage interval earlier than @xmath21-band classifications , with little difference for types outside this range .
the correlation is much better than what was expected from previous near - ir studies ( eskridge et al .
3.6@xmath42 m galaxy morphology is sufficiently contaminated by recent star formation to allow the same criteria defined for blue light images to be used for galaxy classification , a surprising result .
drastic differences between 3.6@xmath42 m and @xmath21-band morphology are seen only for the most dusty galaxies .
one example is ngc 5195 , shown also in figure [ m51_frames ] .
this galaxy , classified as irr ii in the hubble atlas and as i0 by de vaucouleurs , appears as a regular early - type galaxy of type sab(r)0/a ( see also block et al .
1994 ) . other galaxies that can look very different are flocculent spirals such as ngc 5055 ( figure [ nir03 ] ) .
the flocculence largely disappears at 3.6@xmath42 m , and a more global pattern is seen ( see also thornley 1996 ) .
the type of ngc 5055 , sbc , remains largely unchanged from @xmath21 to 3.6 @xmath42 m . as noted by helou et al .
( 2004 ) , the mid - ir wavelength domain marks the transition from emission dominated by starlight to emission dominated by interstellar dust . while images at 3.6@xmath42 m show the stellar bulge and disk almost completely free of dust extinction , an image at 8@xmath42 m shows very little starlight but considerable emission from the ism in the form of glowing dust .
figure [ m81frames ] shows an 8@xmath42 m image of the nearby spiral galaxy m81 as compared to a @xmath70 color index map coded such that blue star - forming features are dark while red dust lanes are light .
the 8@xmath42 m image of m81 shows that its ism is closely associated with its spiral arms .
comparison with the @xmath70 color index shows that both the star - forming arms as well as near - side dust lanes can be seen at 8@xmath42 m .
even the far - side lanes in the bulge region are clear at 8@xmath42 m , where no tilt asymmetry is manifested .
willner et al . (
2004 ) argue that the dust emission from m81 s ism is likely dominated by polycyclic aromatic hydrocarbons ( pahs ; gillett et al .
1973 ) which have a prominent emission feature at 7.7@xmath42 m .
willner et al . also showed good correspondence between the nonstellar dust emission in m81 and the distribution of near - ultraviolet ( nuv ) emission .
some regions with bright dust emission and little nuv emission were attributed to excessive uv extinction , while areas with bright nuv and little dust emission were attributed to the effects of supernovae .
the @xmath70 color index map of m81 shows an additional set of dust lanes that have no counterpart in the 8@xmath42 m map .
these lanes are oriented roughly perpendicularly to the major axis about halfway between the center and the northern arm .
sandage & bedke ( 1994 ) interpret these as foreground dust associated with high galactic latitude nebulosities in the halo of our galaxy .
the key to detecting observable evidence for galaxy evolution , that is , to actually see morphological differences that are likely attributable to evolution , is to observe galaxies at high redshift with sufficient resolution to reveal significant details of morphology .
butcher & oemler ( 1978 ) had already found strong evidence for morphological evolution in the excess number of blue galaxies in very distant ( @xmath1=0.4 - 0.5 ) rich galaxy clusters .
these authors suggested that the blue galaxies are spiral galaxies , and that by the present epoch , these are the galaxies that become the s0s that dominate nearby rich , relaxed clusters ( like the coma cluster , abell 1656 ) .
an excellent summary of the issues connected with high redshift morphological studies is provided by van den bergh ( 1998 ) .
progress on galaxy morphology at high redshift could only be achieved with the resolution and depth of the hubble space telescope .
the hubble deep field north ( hdf - n , williams et al . 1996 ) , south ( hdfs , volonteri et al . 2000 ) , and ultra - deep field ( hudf , beckwith et al . 2006 ) , and the goods ( great observatories origins deep survey ; giavalisco et al . 2004 ) , gems ( galaxy evolution from morphology and seds ; rix et al . 2004 ) , cosmos ( cosmic evolution survey ; scoville et al . 2007 ) , and other surveys ( e.g. , cowie et al .
1995 ) , have provided a large body of information to work with .
for example , studies of galaxies in the redshift range 0.3 @xmath71 0.9 show that the proportion of irregular - shaped galaxies dramatically increases ( e.g. , abraham et al .
this means that the hubble sequence as we know it did not always exist but was built up over time via mergers or secular evolution or both .
observations of submillimeter sources ( chapman et al .
2003 ) suggest some of these irregulars are extended major mergers .
interpretation of evolution in the various deep surveys depends on knowledge of redshifts , which can be difficult to measure spectroscopically .
a very useful technique for isolating galaxies in high redshift ranges is the uv - drop out method ( steidel & hamilton 1992 ) .
galaxies are compared in different filters , such as @xmath72 , @xmath73 , @xmath74 , and @xmath75 .
if the redshift is high enough to move the lyman limit at 0.0912@xmath42 m out of any of the first three filters , there will be a significant drop in flux owing to absorption by hydrogen , and the galaxy is said to drop out .
" galaxies found in this way are called lyman break " galaxies because the effect is partly caused by the spectral characteristics of hot stars , which show such a break .
( another cause is uv self - absorption . )
steidel et al .
( 1996 ) used the uv drop - out approach to identify high @xmath1 galaxies in the hdf - n , and also used direct spectroscopy to confirm that the method works .
beckwith et al .
( 2006 ) utilized the method to identify galaxies in the hudf having @xmath1 from 3.5 to 7 . if a galaxy is seen to drop out of a @xmath76-band filter such as f300w and seen in a @xmath21-band filter such as f450w ( both used for the hdf - n ; williams et al .
1996 ) , then the redshift range selected is @xmath1=2.4 to 3.4 . van den bergh ( 1998 )
argues that most lyman break galaxies are young ellipticals or bulges .
van den bergh et al .
( 2000 ) describe the issues connected with morphological classifications of galaxies to redshifts of @xmath1@xmath21 .
resolution , band - shifting , and selection effects due to the magnitude - limited nature of surveys all enter into the interpretation of intermediate to high redshift galaxy morphology .
resolution is important because , typically , a nearby galaxy will have 100 times or more pixels in the image than a high @xmath1 galaxy will have for classification .
this is fewer pixels than sky survey images of nearby galaxies would have .
the bandpass effect is important because the @xmath21-band , the standard wavelength for historical galaxy classification , is not sampling the same part of the spectrum as it would for nearby galaxies . for example , at @xmath1=1 , a @xmath21-band filter samples mid - ultraviolet light ( @xmath20.22@xmath42 m ) and would be much more sensitive to young star - forming regions than it would be for nearby galaxies . ideally , then
, for comparison with nearby galaxies we would like to choose a redshifted band as close as possible to the _ rest - frame _ @xmath21-band .
even accounting for all these effects , significant differences between nearby and distant morphologies do exist .
for example , van den bergh et al .
( 2000 ) discuss the paucity of grand - design spirals and barred galaxies in the hdf - n , and use artifically redshifted images of nearby galaxies to demonstrate that the deficiencies are likely to be real .
figure [ highz ] shows several of the different categories of intermediate and high redshift galaxy morphologies , based on @xmath7 and @xmath77-band images from the gems and hubble udf .
the redshifts range from 0.42 to 3.35 and provide a wide range of look - back times .
first , in such a range , some galaxies look relatively normal , as shown by the spiral and elliptical galaxies in the two upper left frames of figure [ highz ] .
the @xmath1=0.59 spiral is classifiable as type sa(s)bc and the elliptical as type e3 .
the @xmath1=0.99 spiral shown in the middle right frame of figure [ highz ] has larger clumps , no clear central object , and more asymmetry than the @xmath1=0.59 spiral , but is still recognizable as a spiral .
however , other less familiar categories are found . in general , high redshift galaxies are smaller than nearby galaxies on average ( e.g. , elmegreen et al .
2007a = eefm07 ) .
chain galaxies " were first identified by cowie et al .
( 1995 ) and are linear structures with superposed bright knots that have sizes and blue colors similar to normal late - type galaxies and relatively flat major axis luminosity profiles .
they have the shapes of edge - on disk galaxies but lack clear bulges or nuclei .
a recent study by elmegreen , elmegreen , & sheets ( 2004=ees04 ) of faint galaxy morphologies ( redshifts 0.5 - 2 ) in the advanced camera for surveys ( acs)tadpole " galaxy ( ugc10214 ) field showed that chain galaxies are the most common linear morphology at magnitudes fainter than @xmath56=22 , accounting for more than 40% of the sample .
their dominance ( also found by cowie et al .
1995 ) is interpreted by ees04 as a selection effect because relatively optically thin edge - on galaxies are more favored to be seen near the limit owing to a higher projected surface brightness than for face - on versions of the same galaxies .
ees04 suggest that chain galaxies are edge - on irregular galaxies that will evolve to late - type disk galaxies .
chains are the most flattened linear morphology at faint magnitudes .
clump clusters " ( ees04 ) are somewhat irregular collections of blue knots or clumps with very faint emission between clumps .
the clumps have sizes of @xmath2500pc and masses of @xmath210@xmath7810@xmath59 @xmath60 .
both of the examples shown in figure [ highz ] have @xmath1@xmath31 .
elmegreen , elmegreen , & hirst ( 2004=eeh04 ) identified clump clusters as the face - on counterparts of the linear chain galaxies , based on the similarities of the properties of the clumps with those seen in chain galaxies , and on the distribution of axis ratios of the systems as compared with normal disk galaxies .
the lack of a bulge clump is also consistent with this conclusion .
nevertheless , analysis of nicmos ir images in the hudf led elmegreen et al .
( 2009a ) to conclude that 30% of clump clusters and 50% of chain galaxies show evidence of young bulges , implying that at least half of these galaxies might be genuinely bulgeless . in a related study ,
elmegreen et al . (
2009b ) show that the best local analogues of clump clusters are dwarf irregular galaxies like ho ii , scaled up by a factor of 10 - 100 in mass .
this study also brought attention to clump clusters with faint red background disks , as opposed to blue clump clusters which lack such a feature .
elmegreen et al . argue that the red background clump clusters are part of an evolutionary sequence leading from the blue clump clusters to spirals with a classical " bulge ( e.g. , kk04 ) .
the clumps , formed by gravitational instabilities in a turbulent disk , are large and few in number , and thus will eventually coalesce near the galaxy center if they survive the effects of supernova explosions ( e. g. , elmegreen , bournaud , & elmegreen 2008 ) . tadpoles " ( van den bergh et al .
1996 ) are asymmetrically - shaped head - tail " morphologies with a bright off - centered nucleus and a tail , like a tadpole .
a rare local example is ngc 3991 .
tadpoles were recognized in 3% of the galaxies in hdf - n and were found to be very blue in color .
usually both the head and the tail are blue , but van den bergh et al .
( 1996 ) show one example where the head is red and the tail is blue .
eeh04 showed that tadpoles have neither exponential major axis profiles nor clear bulges , and in their sample of linear objects , tadpoles are the least frequently seen .
the bottom frames of figure [ highz ] show bent chains and rings or partial rings ( elmegreen & elmegreen 2006=ee06 ) .
the rings and partial rings are thought to be mostly collisional in nature ( i.e. , like the conventional ring galaxies shown in figure [ ringtypes ] ) and show the different morphologies expected when small companions plunge through a larger disk galaxy in different ways ( appleton & struck - marcell 1996 ) .
although bent chains resemble the partial rings , they lack offset nuclei and any evidence of a background , more face - on disk .
ee06 suggest that bent chains are simply warped versions of the more common linear chains that have suffered an interaction .
the ages of the bent chain clumps are younger than those found in rings and partial rings , and ee06 argue that relative separations and sizes of the clumps indicate they form by gravitational instabilities .
elmegreen et al .
( 2005 ) show that approximately 1/3 of the ellipticals catalogued in the hudf have prominent blue clumps in their centers ( see also menanteau et al .
2001 , 2004 ) .
they argued that these clumps probably imply accretion events based on comparison of their magnitudes and colors with local field objects .
menanteau et al .
( 2001 ) were able to reproduce the color distributions with a model having a starburst superposed on a pre - existing older stellar population .
galaxy morphology at intermediate and high redshifts also includes obvious interacting cases as well as possible merger morphologies .
bridges , tidal tails , plumes , and even m51 analogues are seen as in nearby galaxies , but are smaller in scale than for nearby objects ( eefm07 ) .
the middle frame of the bottom row shows a possible merger in progress of two bent chains ( or , alternatively , two interacting spirals ) , called an assembly galaxy " by eefm07 because they appear to be assembling from smaller objects . eeh04 and
eefm07 also discuss the double systems , considered another category of the linear systems .
the double systems like the @xmath1=3.35 one shown in figure [ highz ] are probably merging ellipticals .
eefm07 also describe shrimp galaxies " , which appear to be interacting galaxies with a single curved arm or tail , curling at one end into a body . "
other studies of high redshift galaxy morphology have focussed on the specific redshift ranges that are selected by the uv drop - out technique .
conselice & arnold ( 2009 ) examined the morphologies of galaxies in the @xmath1=4 - 6 range from the hudf , and measured quantitatve parameters such as the concentration - asymmetry - clumpiness ( star formation ) parameters ( cas ; conselice 2003 ) and other related parameters that are useful for distinguishing mergers from normal galaxies .
the cas system is based on simple global parameters that are easily derived automatically for large numbers of galaxies .
conselice ( 2003 ) tied the @xmath79 parameter to the past evolutionary history of galaxies while parameters @xmath80 and @xmath81 measure more active evolution from mergers and star formation .
conselice & arnold found that half of the hudf drop - out galaxies they studied have significant asymmetries and may be undergoing merging , while the other half is mainly smooth symmetric systems that may have collapsed quickly into a temporary , quiescent state .
other quantitative approaches to these issues include the sersic @xmath16 index that characterizes radial luminosity profiles ( ravindranath et al .
2006 ; elmegreen et al .
2007b ) and the gini coefficient ( abraham et al . 2003
; lotz et al . 2006 ) .
the gini coefficient provides a way of quantifying high redshift morphology that does not depend on galaxy shape or the existence of a well - defined center , and is well - suited to the kinds of objects shown in figure [ highz ] .
lotz et al .
( 2006 ) found in a sample of 82 lyman break galaxies that 10 - 25% are likely mergers , 30% are relatively undisturbed spheroids , and the remainder are disks , minor mergers , or post - mergers .
given the rise in peculiar and irregular - shaped galaxies with increasing redshift , the question naturally arises : when did the hubble sequence and all its accompanying details fall into place ?
this question is considered by conselice et al .
( 2004 ) , who quantitatively analyzed a well - defined high redshift sample using the cas system .
conselice et al .
identify luminous diffuse objects " ( ldos ) as galaxies having @xmath79 less than 1@xmath8 below the average , and luminous asymmetric objects " ( laos ) as galaxies having @xmath82 .
some of both classes of objects are covered by the elmegreen et al .
categories described above .
all of the ldos and laos have @xmath83 @xmath18 @xmath3519 , and conselice et al .
suggest such objects might be the precursors of modern disk and elliptical galaxies .
these are found in the redshift range 0.2 @xmath84 2 , suggesting the present day hubble sequence began taking shape in this interval .
conselice et al .
( 2008 ) consider the morphologies of galaxies more massive than 10@xmath85 @xmath60 and in the range 1.2@xmath18 @xmath1 @xmath18 3 . to a @xmath75 magnitude of 27 , the majority of these galaxies are peculiar .
they conclude that such galaxies undergo [email protected] mergers to @xmath1=3 .
the van den bergh luminosity classes highlight how luminosity and surface brightness generally go together .
low surface brightness usually means low luminosity and small size , hence a dwarf classification .
however , the discovery of rare giant low surface brighness ( glsb ) galaxies by bothun et al .
( 1987 ) shows that morphology can sometimes be misleading for judging absolute luminosity .
the hallmarks of these objects are a relatively normal bulge and an extremely low surface brightness , very large disk .
disk radial scalelengths and luminosities are unusually large , and extrapolated disk central surface brightnesses are unusually faint , compared to more normal spirals .
the disks tend to be relatively smooth with a few large , isolated hii regions .
bothun et al .
( 1987 ) point to a model whereby the disks of these galaxies have such a low gas surface density that they are largely unevolved , due to the inefficiency of star formation .
glsb galaxies can be classified within the hubble - sandage and de vaucouleurs classification systems although , as noted by mcgaugh , schombert , and bothun ( 1995 ) , the majority are classified later than stage sc .
bulges , bars , rings , and spiral patterns are evident in some examples , in spite of the low disk surface brightness .
figure [ glsbs ] shows three of the originally recognized glsb examples ( mcgaugh , schombert , & bothun 1995 ) : malin 2 ( also known as f568@xmath356 ) , ugc 1230 , and ugc 6614 . in the images , the length of a side is 131 , 38 , and 77 kpc , respectively .
these can compared to the giant normal spiral ngc 7531 in the far right frame , where the length of a side is also 38 kpc .
malin 2 and ugc 6614 are especially enormous physical objects .
ugc 1230 is also very large for such a late - type morphology .
van den bergh ( 1998 ) likens the size of malin 2 to the core of a cluster of galaxies .
he considers monsters " like malin 1 and malin 2 to be only one of three types of lsb galaxies .
some lsb galaxies are as big as normal galaxies , like ugc 1230 .
most lsb galaxies , however , are dwarfs : ellipticals , irregulars and , less frequently , spirals .
these are described further in section 15.2 .
an example of another object that could be considered a large lsb galaxy , but which lacks a bulge or any evident recent star formation , is sgc 2311.8@xmath354353 , the mysterious ghost - like companion close to the right of ngc 7531 in figure [ glsbs ] .
this peculiar object is 2/3 the size of ngc 7531 ( buta 1987 ) but has unknown redshift .
if it is associated with ngc 7531 , it would be as much as 30 kpc in diameter at the faintest detectable isophote level and would clearly not be a dwarf . a recent study of three glsbs ( malin 1 , ugc 6614 , and ugc 9024 ) by rahman et al .
( 2007 ) showed that ir emission from such objects is consistent with their optically - determined low star formation rates , with the diffuse optical disks being undetected from two of the three .
a dynamical study of two glsbs ( malin 1 and ngc 7589 ) by lelli , fraternali , & sancisi ( 2010 ) led the authors to conclude that at least in these cases , the glsb galaxy can be thought of as an inner high surface brightness galaxy having a very extended lsb disk .
this is based on the steeply rising rotation curves found for these galaxies , which is very much like what is seen in early - type high surface brightness galaxies .
impey & bothun ( 1997 ) argue that lsb galaxies brighter than @xmath83=@xmath3514 contribute significantly to the luminosity density of the local universe , are dark - matter - dominated at almost all radii , and have an evolutionary history involving late collapse of a low amplitude perturbation , a low star formation rate , and very slow changes .
large lsbs are greatly underrepresented in galaxy catalogues but are clearly an important class of objects .
the sloan digital sky survey ( york et al . 2000 ; gunn et al . 1998 ) provides the largest body of information on the colors of galaxies .
the multi - wavelength imaging in @xmath86 filters has allowed the production of high quality color images for thousands of galaxies .
although the hubble - sandage - de vaucouleurs classification systems were based on blue light images alone , it is still possible to reliably classify galaxies with sdss color images , which are based on combined @xmath87 images ( lupton et al . 2004 ) . in such images
one can directly see the stellar population differences that characterize different galaxy types .
figure [ colortf ] shows the color hubble sequence from e to sm .
once galaxy colors were systematically measured using photoelectric photometry ( de vaucouleurs , de vaucouleurs , & corwin 1976 ) , it was noted that integrated colors vary smoothly with advancing stage along the hubble sequence .
the latest stages have corrected total color index @[email protected] - 0.4 while e and s0 galaxies have @[email protected] - 1.0 ( e.g. , buta et al . 1994 ) .
the latter colors correspond to yellow - orange while the former are bluish - white .
the colors begin to change at s0/a and sa and become progressively bluer .
figure [ colortf ] shows the reason for the change .
galaxies earlier than sa are dominated by old stars having the colors of k giants .
as stage advances from sa to sm , the spiral structure becomes progressively more important compared to the bulge .
since the arms are dominated by complexes of massive young stars , this makes the integrated colors of the galaxies become progressively bluer until by the end of the sequence , the bluer colors of these stars have overcome the yellowish light of the background old disk stars .
figure [ colortf ] also shows that the intermediate colors of intermediate types such as sb and sc are due to the yellowish - orange colors of bulges and bars as combined with the bluer colors of spiral arms .
the analysis of integrated sdss galaxy colors for more than 100,000 galaxies led to one of the most dramatic findings of the survey : a clear bimodality in the distribution of color that correlates with morphology : a red peak that includes mainly e , s0 , and sa galaxies , and a blue peak that includes mainly sb , sc , and irr galaxies ( strateva et al .
although the correlation of galaxy color with types had been known for a long time from photoelectric measurements ( e. g. , de vaucouleurs 1961 ; buta et al .
1994 ) , the large sample provided by sdss allowed the bimodality to be demonstrated to a high degree of significance . in plots of @xmath89 color index versus absolute @xmath90 magnitude , the galactic equivalent of a stellar h - r diagram , nearby early - type galaxies follow a narrow band called the red sequence , while nearby later - type , mostly star - forming galaxies appear as a broad blue sequence ( also sometimes called the blue cloud " ) .
baldry et al . ( 2004 ) showed that the bimodality ( in the form of a double gaussian number distribution over all types ) is detectable from @xmath90 @xmath2 @xmath3515.5 to @xmath90 @xmath2 @xmath3523 , being undetectable only for the most luminous galaxies .
wyder et al .
( 2007 ) showed that use of galex near - ultraviolet magnitudes and optical @xmath91-band magnitudes provides even greater discrimination between the gaussian components .
bell et al .
( 2004 ) showed that the bimodality is detectable in faint galaxies to @xmath1@xmath21 , indicating that this characteristic of the galaxy population extends to a lookback time of at least 9gyr .
it is thought that galaxies evolve from the blue sequence to the red sequence as their star formation is quenched , perhaps through mergers , gas depletion , or agn feedback ( e.g. , martin et al .
the possibility of evolution has engendered great interest in the galaxies lying near the minimum of the bimodal distribution ( the so - called green valley " ; thilker et al .
2010 and references therein ) .
virtually all the galaxies shown in figure [ colortf ] are of relatively high luminosity , with absolute blue magnitudes @xmath92 averaging about @xmath3520 . when physical parameters such as @xmath92 are considered , it becomes clear that the peculiar shape of the de vaucouleurs classification volume shown in figure [ classification_volume ] only highlights the morphological diversity of families and varieties at each stage , but does not tell us about the _ physical parameter space _ at each stage , which expands considerably at each end of the volume ( mcgaugh , schombert , and bothun 1995 ) .
most known dwarf galaxies are either early or late - type , but not intermediate .
the most extensive study of dwarf galaxy morphologies was made by binggeli , sandage , & tammann ( 1985=bst ) , who used deep photographs to probe the low luminosity population of the virgo cluster , using mostly morphology to deduce cluster membership .
examples of several categories of virgo cluster dwarf galaxies are shown in figure [ dwarfs ] using sdss color images and the classifications of bst .
the most common type is the dwarf elliptical , or de type , which accounts for 80% of the galaxies in the bst catalogue .
de galaxies range from @xmath83 = @xmath3518 to @xmath358 ( ferguson & binggeli 1994 ) .
many des have an unresolved , star - like nucleus whose presence is indicated by an n attached to the type , as in de0,n .
the top right panels of figure [ dwarfs ] show three fairly typical examples .
possibly related to these normal de systems are the larger , lower surface brightness ellipticals ( large de " ) shown in the two lower right frames of figure [ dwarfs ] .
the second row of figure [ dwarfs ] shows examples of the interesting class of dwarf s0 galaxies .
all of the examples shown are distinct from des in showing a smooth structure but with additional features such as a lens or a weak bar .
ds0 galaxies can also be nucleated and are called ds0,n .
in addition to the low surface brightness des and ds0s , the virgo cluster includes two high surface brightness classes of dwarfs . the ce category refers to compact ellipticals that resemble m32 .
the blue compact dwarf ( bcd ) galaxies are a special class of star - forming dwarf irregulars characterized by a few bright knots imbedded in a stellar background of low surface brightness ( sandage and binggeli 1984 ) .
the most extreme cases are nearly stellar ( thuan & martin 1981 ) .
the knots are often super star clusters associated with 30-doradus - like hii regions , and the faint background can be very blue ( thuan et al . 1997 ) .
spectroscopically , bcds have narrow emission lines superposed on a blue continuum , and the lines indicate a low metallicity .
figure [ dwarfs ] shows three examples from the bst virgo cluster catalogue .
gil de paz et al .
( 2003 ) present an atlas of more than 100 bcds that highlight their structure .
as described in section 5.1 , the de galaxies shown in figure [ dwarfs ] are _ not _ the low luminosity extension of more luminous ellipticals .
these together with the ds0 class are labeled spheroidal " galaxies by kormendy et al .
( 2009 ; see figure [ cores ] ) , who confirmed the finding by kormendy ( 1985 ) that these galaxies are more related to evolved low luminosity spirals and irregulars than to genuine ellipticals . based on correlations of well - defined photometric parameters ( e.g. , central surface brightness or velocity dispersion versus core radius or absolute magnitude )
, these authors link ellipticals like m32 to the actual low luminosity end of the e galaxy sequence ( see also wirth & gallagher 1984 ) .
thus , the ce galaxies shown in figure [ dwarfs ] are in a sense truer dwarf ellipticals " than the de galaxies shown .
this does not negate the value of the bst classifications , since these are purely morphological interpretations .
the lowest luminosity galaxies that we can study in detail are in the local group .
fortunately , a few are in the area covered by the sdss so that they can be illustrated in color . these objects , all fainter than @xmath93=@xmath3512 , are shown in figure [ lgroup ] .
leo i and leo ii are usually called dwarf spheroidal , or dsph , galaxies ( e.g. , ferguson & binggeli 1994 ) .
dsph galaxies tend to have to have absolute visual magnitudes @xmath93 @xmath3 @xmath3515 and a low degree of flattening ; they are believed to be the most abundant type of galaxy in the universe .
leo i ( @[email protected] ) and leo ii ( @[email protected] ) look different in part because of their different star formation histories .
dsph and dwarf irregular ( di ) galaxies are now known to have complex and varied star formation histories that may involve multiple episodes of star formation and effects of interactions ( mateo 1998 and references therein ) . the other two galaxies in figure [ lgroup ] , leo a and ddo 155 , are dwarf irregulars having @xmath93@[email protected]
. a detailed hst study of the stellar content of leo a ( cole et al . 2007 ) showed that 90% of the star formation in the galaxy occurred less than 8 gyr ago , with a peak at 1.5 - 3 gyr ago .
a useful summary of the properties of dsph galaxies is provided by van den bergh ( 1998 ) and , most recently , by tolstoy , hill , & tosi ( 2009 ) , who also discuss the star formation histories of local group di galaxies .
the sdss has facilitated the discovery of many new local group dwarf galaxies ( belokurov et al .
for example , the sdss led to the discovery of one of the faintest known dsph galaxies , a new dwarf in ursa major ( called the uma dsph ) having @xmath93 @xmath2 @xmath357 ( willman et al .
most interesting is leo t , which is an @xmath93 = @xmath357.1 dsph galaxy with some recent star formation , providing one of the most dramatic illustrations of the link between dsph and di galaxies , and the least luminous galaxy known to have recent star formation ( irwin et al .
2007 ) .
sandage & binggeli ( 1984 ) described the classification of dwarf galaxies based on the virgo cluster , and concluded that there are no real dwarf spirals . "
this refers mainly to dwarf spirals that might be classified as types sa , sb , or sc , i.e. , having both a bulge and a disk .
dwarf late - type spirals are already built into de vaucouleurs s modified hubble sequence as sd - sm types and connect directly to magellanic irregulars , as shown in figure 1 of sandage & binggeli ( 1984 ) .
thus , any genuine examples of dwarf sa , sb , or sc spirals would be of great interest as they would challenge the idea that for a galaxy to be able to make well - defined spiral arms , it would have to be more massive than some lower limit ( estimated as 5@xmath5810@xmath59 @xmath60 by sandage & binggeli ) .
four of the best cases of genuine dwarf spirals are illustrated in figure [ dwarfspirals ] .
the two rightmost frames show ic 783 ( bst type ds0,n ) and ic 3328 ( bst type de1,n ) , both virgo cluster members having absolute magnitudes @xmath92 @xmath2 @xmath3516 to @xmath3517 , and found to have subtle spiral structure by barazza , binggeli , & jerjen ( 2002 ) and jerjen , kalnajs , & binggeli ( 2000 ) , respectively .
the patterns are hard to see in the direct sdss color images shown in figure [ dwarfspirals ] , but these authors use photometric models , fourier decomposition , and unsharp - masking to verify the reality of the patterns .
barraza et al .
conclude that many of the bright early - type dwarfs in the virgo cluster have disks .
an example with a bar _ and _ spiral arms is ngc 4431 ( shown as ds0 in figure [ dwarfs ] ) .
the leftmost panel of figure [ dwarfspirals ] shows an hst wide @xmath7-band image ( carollo et al .
1997 ) of ngc 3928 , an absolute magnitude @xmath92 = @xmath3518 galaxy which on small - scale , overexposed images looks like an e0 , but which harbors a miniature ( 2kpc diameter ) , low - luminosity sb spiral ( van den bergh 1980b ) . based on spectroscopic analysis , taniguchi & watanabe ( 1987 ) have suggested that ngc 3928 is a spheroidal galaxy which experienced an accretion event that supplied the gas for star formation in the miniature disk .
schombert et al .
( 1995 ) brought attention to possible dwarf field spirals .
one of their examples , d563@xmath354 , is shown in the second frame from the left in figure [ dwarfspirals ] .
this galaxy has @xmath92@xmath2@xmath3517 .
a few other examples are given in the paper , and schombert et al .
find that they are not in general grand design spirals , are physically small , and have low hi masses .
however , van den bergh ( 1998 ) considers all of schombert et al.s examples as subgiant spirals rather then true dwarf spirals . a possible true dwarf spiral given by van den bergh
is ddo 122 ( type s v ) .
the galaxy zoo project ( lintott et al . 2008 ) has made extensive use of sdss color images .
the project uses a website to enlist the help of citizen scientists worldwide to classify a million galaxies as well as note interesting and unusual cases in various forum threads .
with such a large database to work from , and the potential for discovery being real , the project has attracted many competent amateur galaxy morphologists .
one such discovery was a new class of galaxies called green peas , " which are star - like objects that appear green in the sdss composite color images ( figure [ zoofig ] , left ) .
cardemone et al .
( 2009 ) used auxiliary sdss data to show that peas are galaxies that are green because of a high equivalent width of [ oiii ] 5007 emission .
they are sufficiently distinct from normal galaxies and quasars in a two - color @xmath94 vs. @xmath95 plot that such a plot can be used to identify more examples .
other characteristics noted are that peas are rare , no bigger than 5 kpc in radius , lie in lower density environments than normal galaxies , but may still have morphological characteristics driven by mergers , are relatively low in mass and metallicity , and have a high star formation rate .
cardemone et al .
conclude that peas are a distinct class of galaxies that share some properties with luminous blue compact galaxies and uv - luminous high redshift galaxies .
another prominent colorful galactic - sized object identified by galaxy zoo is
hanny s voorwerp " ( figure [ zoofig ] ) , a peculiar collection of blue clumps just south of ic 2497 .
lintott et al .
( 2009 ) found that the voorwerp is mostly ionized gas , and after ruling out some possible sources of ionizing radiation , concluded that the object could be the first identified case of a quasar light echo .
the implication is that the companion galaxy underwent a temporary quasar phase .
masters et al .
( 2010 ) examine the properties of face - on late - type spiral galaxies whose colors are much redder than is typical ( figure [ zoofig ] , right panels ) , suggesting that they are passive objects where star formation has largely ceased or is lower than normal .
these authors showed that red spirals are not excessively dusty and tend to be near the high end of the mass spectrum .
a range of environmental densities was found , implying that environment alone is not sufficient to make a spiral red .
a significantly higher bar fraction was found for red spirals as compared to blue spirals , which masters et al .
suggest could mean the bars themselves may have acted to shut down the star formation in these galaxies .
a major philosophical aspect of galaxy zoo is the value of human visual interpretation of galaxy morphology .
that is , the human eye can integrate the detail in an image more reliably than a computer program can , in spite of the latter s ability to classify numbers of galaxies well beyond the capability of a single individual .
this philosophy was used to compile a major catalogue of galaxy merger pairs described by darg et al .
( 2010 ) . in the initial set - up of galaxy zoo ,
a single button allowed a classifier to select whether an object was a merger " based simply on the appearance of peculiarities . for each galaxy , a weighted average number , @xmath96 , was derived that characterized the fraction of classifiers who interpreted a pair of galaxies as a merger , essentially
morphology by vote " and in a way reminiscent of rc3 where morphological types were in many cases based on a weighted average from a small number of classifiers ( buta et al .
taking @xmath96 @xmath3 0.4 , darg et al .
( 2004 ) identified 3003 pairs and groups of merging galaxies , and also showed that the spiral to elliptical galaxy ratio in merger pairs is a factor of two higher than for the global galaxy population , suggesting that mergers involving spirals are detectable for a longer period than those that do not involve spirals .
a similar philosophy to galaxy zoo was used by buta ( 1995 ) to compile the catalogue of southern ringed galaxies , and also by schawinski et al .
( 2007 ) , who visually classified 48,023 sdss galaxies to identify a significant - sized sample of early - type galaxies for a study of the connection between nuclear activity and star formation . in the latter study , visual interpretation was argued to be needed to avoid bias against star - forming early - type galaxies which would be excluded from color - selected samples .
a major result of this study was the identification of a time sequence whereby an early - type galaxy has its star formation suppressed by nuclear activity , a manifestation of agn feedback .
if interactions and mergers can have profound effects on galaxy morphology , then the morphology of _ isolated _ galaxies clearly is of great interest .
such galaxies allow us to see how internal evolution alone affects morphology , i. e. , what pure nature " morphologies look like .
karachentseva ( 1973 ) compiled a large catalogue ( the catalogue of isolated galaxies , or cig ) of 1050 isolated galaxies that has proven very useful for examining this issue . a galaxy of diameter @xmath97 is sugggested to be isolated if it has no comparable - sized companions of diameter @xmath98 between @xmath97/4 and 4@xmath97 within a distance of 20@xmath98 .
verdes - montenegro et al .
( 2005 ) show that this means that an isolated galaxy 25 kpc in diameter , in the presence of a typical field galaxy velocity of 150 km s@xmath99 , has not been visited by a comparable mass companion during the past 3gyr .
these authors discuss the limitations of the cig ( e. g. , the isolation criteria do not always work ) , but in general it is the best source of isolated galaxies available .
sulentic et al .
( 2006 ) examined all 1050 cig galaxies on palomar ii sky suvey charts in order to refine the sample and found that isolated galaxies cover all hubble types .
of these , 14% were found to be e / s0 types , while 63% were sb - sc types , with the spiral population more luminous than the e / s0 population . over the type range sa - sd
, the proportion rises to 82% .
thus , an isolated galaxy sample is very spiral - rich .
nevertheless , the presence of early - type galaxies in the sample implies that these are not likely to be nurture " formed , as such galaxies might be in denser environments .
the refinement of the cig sample forms the basis of the analysis of the interstellar medium of isolated galaxies ( amiga ) project ( verdes - montenegro et al .
2005 ) . figure [ isolateds ] shows six examples of isolated sb - sc galaxies from the amiga sample , based on sdss color images .
all of these look relatively normal , but it is interesting how nonbarred galaxies like ngc 2649 and 5622 ( upper left frames in figure [ isolateds ] ) show such conspicuous global spirals , which argues that the spirals in these galaxies have not been excited by an interaction .
durbala et al .
( 2008 ) analyzed the photometric properties of 100 isolated sb - sc amiga galaxies , and found that a majority have pseudobulges rather than classical bulges . in comparing the properties of isolated galaxies with a sample of sb - sc galaxies selected without an isolation criterion , durbala et al .
found that isolated spirals have longer bars and , using cas parameters , also less asymmetry , central concentration , and clumpiness , durbala et al .
( 2009 ) analyzed the fourier properties of the same set of 100 isolated spirals , and estimated bar lengths and strengths .
earlier types in the sample were found to have longer and higher contrast bars than later types .
spiral arm multiplicities were investigated also , and it was shown that cases having an inner @xmath40=2 pattern and an outer @xmath40=3 pattern occurred in 28% of the sample .
elmegreen , elmegreen , & montenegro ( 1992 ) argued that in such morphologies , the @xmath40=3 pattern is driven internally by the @xmath40=2 pattern , and that three - armed patterns measure the time elapsed since an interaction .
particularly interesting in the domain of color galaxy morphology are the various deep field surveys that have been at the heart of high redshift studies .
the most recent is the hst wide field camera 3 early release science ( ers ) data ( windhorst et al .
2010 ) , which provides very deep panchromatic images based on 10 filters ranging from 0.2@xmath42 m to 2@xmath42 m in wavlength .
figure [ ers ] shows three subsections of the main ers field , which coincides with the goods south field ( giavalisco et al .
the galaxies seen range in redshift from @xmath1=0.08 to at least @xmath1=3 .
windhorst et al .
( 2010 ) argue that images like the ers field are deep enough to allow probing of galaxy evolution in the crucial redshift range @xmath1@xmath21 - 3 where the galaxies assembled into their massive shapes .
by @xmath1@xmath21 , the hubble sequence was largely in place .
figure [ ers ] shows a variety of interesting nearby as well as high @xmath1 morphologies , including some of those illustrated in figure [ highz ] .
images like the ers field allow us to connect the local and distant galaxy populations in unprecedented detail .
the process of galaxy classification thus far described is a manual exercise where an observer attempts to sort a galaxy into its appropriate stage , family , variety , outer ring classification , etc .
, by visual inspection alone . for a small number of observers ,
this has been done for as many as 14,000 galaxies by nair & abraham ( 2010 ) and 48,000 galaxies by schawinski et al .
( 2007 ) , while for a large number of observers working in concert ( e.g. , the galaxy zoo project ) it has been done for a million galaxies . but
critical to such ventures is the preparation of images for classification , and a need for a homogeneous , objective approach to very large numbers of galaxies .
this has led to extensive application of automated methods for classifying galaxies .
for example , nair & abraham ( 2010 ) visually classified galaxies into coded bins for the purpose of training an automatic classification algorithm .
( the remainder of this part will be found in the printed publication . )
the large number of illustrations in this article attests to the richness of the diversity of galaxy morphology .
it is , of course , not possible to illustrate all aspects of morphology that might be worth discussing , but most interesting is how far _ physical galaxy morphology _ has come in the 35 years since allan sandage wrote his review of galaxy morphology in volume ix of _ stars and stellar systems_. galaxy morphology is no longer the purely descriptive subject it once used to be .
internal perturbations such as bars are apparently capable of generating a great deal of the interesting structure we see in disk galaxies , and more theoretical and observational studies should elucidate this further .
the impact of bars on morphology seems well understood , as summarized in detail by kk04 in their monumental review article on pseudobulges .
bars redistribute angular momentum and reorganize gas clouds to flow into resonance regions and fuel star formation .
the gathering of gas into resonance regions can drive the formation of rings , and indeed can also build up the central mass concentration to the point of bar destruction . even failing this , the pile - up of gas into the nuclear region can lead to the formation of a pseudobulge .
the richness of barred galaxy morphology attests to the strong role secular evolution plays in structuring galaxies .
progress in understanding the role of mergers and interactions on galaxy morphology has also proceeded at a rapid pace .
great success in numerical simulations and the theory of interacting galaxies has made it possible to link a specific type of interaction to a specific morphology ( e. g. , collisional ring galaxies ) .
the complex structure of early - type galaxies , with boxy and disky isophote shapes , shells and ripples , and other features , shows that interactions and mergers play an important role in molding galaxies ( see the excellent review by schweizer 1998 ) . with the advent of the hubble space telescope , this role has been elucidated even more clearly because the merger rate was higher in the past . in spite of the theoretical progress , it is interesting that classical morphology has not lost its relevance or usefulness even after more than 80 years since hubble published his famous 1926 paper .
no matter how much progress in understanding the physical basis for morphology is made , there is still a need for the ordering and insights provided by classical hubble - sandage - de vaucouleurs galaxy classifications .
morphology went through a low phase in the 1980s and 90s when it was perceived that galaxy classification placed too much emphasis on unimportant details and was too descriptive to be useful .
it was thought that the hubble classification had gone as far as it can go , and that another approach needed to be tried to build a more physical picture of galaxies . at that time
, there was a sense that the focus should be more on the component building blocks " of galaxies , or what might be called galactic subsystems ( e.g. , djorgovski 1992 ) .
quantification of morphology became more possible as advanced instrumentation allowed more detailed physical measurements to be made . in the end , as morphology became better understood , it also became clear what a type such as
( r)sb(r)ab " might really mean , which enhanced the value of classification ( kk04 ) .
in addition , numerical simulations became sophisticated enough to make predictions about morphology ( e.g. , the r@xmath32 and r@xmath100 subclasses of outer rings and pseudorings ) .
these types of things , as well as the movement of morphology from the photographic domain to the digital imaging domain , the broadening of the wavelength coverage available to morphological studies from the optical to the ultraviolet and infrared domains , the sloan digital sky survey , and the accessiblity of high redshift galaxies to unprecedentedly detailed morphological study , all played a role in bringing galaxy morphology to the forefront of extragalactic research .
even so , the writing of this article has shown that many important galaxies and classes of galaxies have not been studied well enough to have much modern data available .
for example , in spite of the considerable interest in collisional ring galaxies the past 20 years or so , struck ( 2010 ) was forced to lament that ring galaxies are underobserved . "
the same can be said for resonance ring galaxies , giant low surface brightness galaxies , dwarf spirals , magellanic barred spirals , counter - winding spirals , and other classes of interacting galaxies .
the most that can be said about this is that further studies will likely be made , especially if instrumentation facilitates the objects in question .
rotation and dynamics are far short of photometry for most classes of galaxies , but would add a great deal of insight if obtainable . at the other extreme ,
early - type ( e and s0 galaxies ) continue to be the focus of major photometric , kinematic , and theoretical research projects .
important clues to the formation and evolution of such galaxies are contained in their intrinsic shapes ( oblate , prolate , triaxial ) , in the ages , metallicities , and radial mass - to - light ratios of their stellar populations , in their three - dimensional orbital structure , and in the kinematic peculiarities often found in such systems ( de zeeuw et al .
2002 ) . among the most recent studies
are the massive photometric analysis of early - types in the virgo cluster by kormendy et al .
( 2009 ) , and the atlas@xmath101 project described by cappellari et al .
atlas@xmath101 is the largest kinematic database of high - quality two - dimensional velocity field information ever obtained for early - type galaxies , including 260 such galaxies in a well - defined and complete sample .
this survey is simply the latest part of the long - term effort by many researchers , beginning in the 1980s , to understand early - type galaxies in terms of quantitative parameters that can be tied to theoretical models .
early - types have been a persistent enigma in morphological studies , and considerable evidence suggests that the e , s0 sequence as defined by hubble , sandage , and de vaucouleurs hides a great deal of important physics associated with these objects . the atlas@xmath101 project was designed to exploit the @xmath25 parameter described by emsellem et al .
( 2007 ; see section 5.1 ) , which separates early - types into fast and slow rotators and discriminates galaxies along the red color sequence ( section 15.1 ) . these advances for
early - type galaxies do not mean that quantitative analyses of later - type galaxies are lacking . as codes for two - dimensional photometric decomposition become ever more sophisticated ( e. g. , peng et al .
2010 ; laurikainen et al .
2010 and references therein ) , parameters that characterize the bulges , disks , bars , lenses , rings , and spiral patterns are being derived for large numbers of galaxies ( especially barred galaxies ) that were not reliably decomposable with earlier one - dimensional approaches . for the future
, it is to be hoped that the sloan digital sky survey will be extended to cover the whole sky , and provide access to high quality morphological studies of several million more galaxies , some of which might have new and exotic structures . the _ james webb space telescope _ should be able to carry the hst s torch to greater depths and resolutions of high redshift galaxies , to further enhance our understanding of galaxy evolution .
this article is dedicated to allan sandage ( 1926 - 2010 ) , one of the 20th century s greatest astronomers , who helped set the stage for galaxy morphology to be one of the most active fields of modern extragalactic research .
it was dr .
sandage s efforts that firmly cemented hubble s ideas on morphology into astronomy .
the author is grateful to dr .
sandage for the inspiration he provided for this article and for his standard of excellence in astronomy .
the author also gratefully acknowledges the helpful comments and suggestions from the following people that considerably improved this article : martin bureau , adriana durbala , debra elmegreen , william keel , jeffrey kenney , johan h. knapen , rebecca koopmann , john kormendy , eija laurikainen , barry madore , karen l. masters , patrick m. treuthardt , sidney van den bergh , and xiaolei zhang .
the author also thanks debra elmegreen for providing the images of high redshift galaxies shown in figure [ highz ] , masfumi yagi for the illustrations in figure [ coma ] , and john kormendy , jason surace , donald p. schneider , and rogier windhorst for the use of published illustrations from specific papers .
this article uses images from many sources too numerous to acknowledge here , but mainly drawn from the dva , the nasa / ipac extragalactic database , the ohio state university bright spiral galaxy survey ( osubsgs ) , the sloan digital sky survey , and several published papers by other authors .
the nasa / ipac extragalactic database ( ned ) is operated by the jet propulsion laboratory , california institute of technology , under contract with the national aeronautics and space administration .
funding for the osubsgs was provided by grants from the nsf ( grants ast 92 - 17716 and ast 96 - 17006 ) , with additional funding from the ohio state university .
funding for the creation and distribution of the sdss archive has been provided by the alfred p. sloan foundation , the participating institutions , nasa , nsf , the u.s .
department of energy , the japanese monbukagakusho , and max planck society .
observations with the nasa / esa _ hubble space telescope _ were obtained at the space telescope science institute , which is operated by the association of universities for research in astronomy , inc .
, under contract nas 5 - 26555 .
spitzer space telescope _ is operated by the jet propulsion laboratory , california institute of technology , under nasa contract 1407 .
the _ two micron all - sky survey _ is a joint project of the university of massachusetts and the infrared processing and analysis center / california institute of technology , funded by the national aeronautics and space administration and the national science foundation .
galex is a nasa mission operated by the jet propulsion laboratory .
galex data is from the multimission archive at the space telescope science institute ( mast ) .
support for mast for non - hst data is provided by the nasa office of space science via grant nnx09af08 g and by other grants and contracts .
this article has also made use of things , the hi nearby galaxy survey " ( walter et al .
2008 ) , and bima - song , the berkeley - illinois - maryland survey of nearby galaxies ( helfer et al .
2003 ) .
bagetakos , i. brinks , e. , walter , f. , de blok , w. j. g. , rich , j , w. , usero , a. , & kennicutt , r. c. 2009 , in the evolving ism in the milky way and other galaxies , k. sheth , a. noriega - crespo , j. ingalls , and r. paladini , eds . , http://ssc.spitzer.caltech.edu/mtgs/ismevol , e17 block , d. , freeman , k. c. , puerari , i. , combes , f. , buta , r. , jarrett , t. , & worthey , g. 2004a , in penetrating bars through masks of cosmic dust , eds .
d. l. block , i. puerari , k. c. freeman , r. groess , & e. block , springer , kluwer , p. 15 | galaxy morphology is a product of how galaxies formed , how they interacted with their environment , how they were influenced by internal perturbations , agn , and dark matter , and of their varied star formation histories .
this article reviews the phenomenology of galaxy morphology and classification with a view to delineating as many types as possible and how they relate to physical interpretations .
the old classification systems are refined , and new types introduced , as the explosion in available morphological data has modified our views on the structure and evolution of galaxies .
_ index terms _ : hubble - sandage - de vaucouleurs classification ; spiral galaxies ; elliptical galaxies ; s0 galaxies ; irregular galaxies ; resonance ring galaxies ; outer rings ; inner rings ; nuclear rings ; secondary bars ; nuclear bars ; pseudorings ; grand design spirals ; flocculent spirals ; merger morphologies ; tidal tails ; collisional ring galaxies ; polar ring galaxies ; barred galaxies ; lens galaxies ; spiral arm classes ; counter - winding spiral galaxies ; anemic spiral galaxies ; spiral arm multiplicity ; dust lanes ; pseudobulges ; classical bulges ; dust - lane elliptical galaxies ; shell / ripple galaxies ; ultra - luminous infrared galaxies ; bar ansae ; brightest cluster members ; low surface brightness galaxies ; dwarf galaxies ; dwarf spiral galaxies ; dwarf spheroidal galaxies ; galaxy zoo project ; isolated galaxies ; blue compact dwarf galaxies ; star - forming galaxy morphologies ; warped disks ; stellar mass morphology ; luminosity classes ; boxy ellipticals ; disky ellipticals ; late - type galaxies ; early - type galaxies ; magellanic barred galaxies ; color galaxy morphology ; atomic and molecular gas galaxy morphology |
low - mass stars form from the gravitational collapse of dense molecular cloud cores . in the earliest stages ,
the young protostar is deeply embedded in a cold envelope of infalling gas ( the so - called class 0 stage ; * ? ? ?
* ; * ? ? ?
studies of the deeply embedded stages reveal the properties of the youngest stellar objects during or just after formation of the star .
an interesting interplay takes place in these earliest stages between material falling toward the central star - disk system and that being dispersed by a protostellar outflow while simultaneously being heated by radiation from the central star .
this paper presents a large survey of 8 deeply embedded protostars with 13 ( @xmath9 au ) resolution in a wide range of line and continuum emission at submillimeter wavelengths using the submillimeter array ( sma ) .
these observations are ideally suited to probe the inner warm and dense material surrounding low - mass protostars , in contrast with lower frequency data which are more sensitive to the cold outer envelope .
combined studies of the physical and chemical structure of low - mass protostars are important for a complete picture of the evolution of young stellar objects , for understanding their structure on few hundred au scales and for addressing a number of questions about low - mass star formation .
for example , do all deeply embedded protostars have disks ?
how massive are these disks , both compared to their envelopes and to the disks around the more evolved t tauri stars ?
is there evidence that their physical properties and composition ( e.g. , the properties of the dust of which they are made ) are different from those of their more evolved counterparts ?
what are the time scales for the dissipation of the protostellar envelopes ?
how important are protostellar outflows for the dissipation , e.g. , compared to the amount of material accreted onto the central star+disk system ?
likewise , what is the importance of shocks associated with the outflows in regulating the chemistry in the protostellar envelopes ?
are complex organic molecules common in these envelopes and do they originate in shocks or due to heating by the central protostar ? observations at ( sub)millimeter wavelengths can address a number of these issues : thermal dust continuum emission can be used to constrain the envelope density and temperature structure whereas line observations give information about the dynamical structure ( e.g. , of the infalling material ) , of excitation conditions and of the chemistry . submillimeter dust emission and high - excitation lines constrain the properties of the actual protostellar envelope and are less sensitive to the surrounding cloud material than , e.g. , observations from high - resolution 3 mm studies ( e.g. , * ? ? ? * ; * ? ? ? * ; * ? ? ? * ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
previously such submillimeter observations were only feasible with single - dish telescopes with resolutions of a few thousand au for nearby star forming regions ( see , e.g. , * ? ? ? * for a recent review ) . with the submillimeter array @xcite ,
however , a new window has opened up , allowing for detailed studies of the radial variations of both the physical / dynamical and chemical structure of protostellar cores .
the sma is ideal for studies of these inner regions for a number of reasons : first , previous interferometric studies based on lower excitation lines did not probe deeply inside the envelope since these lines are sensitive to the chemistry of the outer , cold regions and become optically thick . in the 325365 ghz window
a wealth of molecular transitions constrain the chemistry in the denser ( @xmath10 @xmath11 ) and warmer ( @xmath12 k ) material of the envelope .
second , since the dust continuum flux scales with frequency as @xmath13 or steeper , submillimeter observations are well suited to probe the dust in protostellar disks . at the same time , the innermost regions of the envelopes where the temperature increases above 100 k are heavily diluted in a single - dish beam ( @xmath14 size compared to typical single - dish beam sizes of 10 - 20@xmath3 ) .
third , interpretation of the line emission from these regions relies on extrapolation of the density and temperature distribution from observations on larger scales .
typical sma observations resolve the emission down to these scales and make it possible to disentangle the emission from the envelope and circumstellar disk .
a number of recent papers have presented studies of individual protostars with the sma that focused on different aspects of the physical and chemical structure of protostellar envelopes , disks and outflows @xcite . to make statistical comparisons and general statements about the deeply embedded protostars as a whole , however , the systematic effort described here is warranted .
the purpose of this paper is to describe the details of the observations and present an overview of the results with a few pointers to important implications .
the observational details , including the sample , observed settings , reduction and calibration strategy , are presented and discussed in [ obs ] . in
[ contresults ] , the results of the continuum observations are discussed with simple estimates of the compact emission from the central disks , their masses and the slope of the dust opacity law .
the line observations are discussed in [ lineresults ] , showing all observed maps and addressing which molecular species are particularly useful for probing different aspects of the protostellar cores , and focusing on the dynamical and chemical impact of the protostellar outflows .
finally , [ summary ] concludes the paper .
this paper serves as a starting point for a series of focused papers that will describe more specific topics in detail and will serve as an important reference for planning future sma ( and eventual alma ) observations of low - mass protostars .
the sources observed in this program were selected from a large single - dish survey of submillimeter continuum and line emission toward low - mass protostars @xcite .
that sample consisted of nearby ( @xmath15 pc ) class 0 objects with luminosities less than @xmath16 - all visible from mauna kea ( i.e. , predominantly northern sources ) .
for this study we selected a subset of 7 objects that , in addition to the above single - dish survey , have been studied extensively through aperture synthesis observations at 3 millimeter wavelengths . to this we added the class 0 object b335 which likewise has been studied in great detail using ( sub)millimeter single - dish telescopes @xcite , through longer wavelength aperture synthesis observations @xcite , at near - ir wavelengths @xcite , and using detailed dust and line radiative transfer @xcite
. detailed line and continuum radiative transfer models exist for each of the objects which can be used to address the extended emission associated with their envelopes ( e.g. , * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) and in addition single - dish maps have been obtained for a subset of the lines which will be incorporated in forthcoming papers . the full sample of objects is summarized in table [ sample ] . in appendix
[ sampledescription ] , we provide a brief description of the characteristics of each object from previous observations .
.sample of sources . [ cols= " < ,
< , < , < , < , < , < , < " , ] @xmath17systemic velocity : statistical uncertainties from gaussian fits of 0.050.1 km s@xmath18 .
@xmath19line width ( fwhm ) : statistical uncertainties from gaussian fits of 0.10.3 km s@xmath18 .
all sources show detections in the 21 transition of the co isotopic species at 230 ghz and most sources are detected in the expected strong lines of the main isotopologues of cs 76 and h@xmath1co @xmath20 lines at 342.8 and 351.7 ghz , with l1527 being the exception . for the other lines iras2a shows strong centrally condensed emission close to the central protostar in all species , whereas the other sources in the sample have detections in only some of the lines with extended emission in some cases .
lcccccccc + @xmath21co 21 & + & + & + & + & + & + & + & + + @xmath8co 21 & + & + & + & + & + & + & + & + + so @xmath22 & ( o ) & + & + & + & + & + * & + & ( o ) + c@xmath23o 21 & + & + & + & + & + & + & + & + + ch@xmath2oh & - & - & - & ( o ) & - & - & - & - + + c@xmath24o 32 & ( + ) & ( + ) & + & + & - & - & + & + + c@xmath25s 76 & - & + & ( + ) & - & - & - & - & - + ch@xmath2oh @xmath26 & ( + ) & + & o & o & - & + & - & - + ch@xmath2oh @xmath27 & ( + ) & + & o & o & - & + & - & - + ch@xmath2oh @xmath28 & - & + & o & o & - & - & - & - + ch@xmath2oh @xmath29 & - & + & - & - & - & - & - & - + sio 87 & - & - & ( o ) & o & - & - & - & - + hn@xmath8c 43 & - & - & - & - & - & - & - & - + h@xmath8co@xmath0 43 & + & - & - & + * & - & + * & - & + + + cs 76 & + & + & + /o&+/o & - & + & + & + + h@xmath1co @xmath20 & + & + & + /o & o & - & ( + ) & + & + + so@xmath1 & ( + ) & - & - & - & - & - & - & - + as with the continuum observations , a significant fraction of the line emission is resolved out - and only the brightest , most compact regions are detected . for most species and sources , this occurs close to the continuum positions - as one would expect from a centrally condensed envelope .
it is also seen that the sources which are observed with the shortest possible baselines in the compact configuration show stronger emission in some of the prominent lines : for example , strong c@xmath24o 32 emission is detected in l1157 whereas it does not show particularly strong c@xmath24o 32 lines compared to the other sources in the survey of @xcite .
again , this likely reflects that most of the c@xmath24o 32 emission seen in the single - dish observations comes from the extended envelope and thus is very sensitive to the differences in the lengths of the shortest baselines .
the amount of resolving out for a somewhat extended molecular species is estimated by comparison to the single - dish observations of c@xmath23o 21 by @xcite ( for b335 , cso observations presented by @xcite were used ) .
the integrated line intensities from the single - dish observations are converted from antenna temperature scales to jy beam@xmath18 using the standard expressions and compared to the central spectra from either the interferometer datasets restored with a beam size comparable to the single - dish beam size ( in cases where the signal - to - noise is sufficient ) or the flux integrated over a smaller region where the emission was detected . for c@xmath23o 21
the amount of flux recovered by the interferometer estimated in this way varies from 13% for l1448 , iras2a , iras4b , l1527 and l483 , about 6% for iras4a to 16% and 22% for l1157 and b335 , respectively .
the slightly higher amount of recovered emission in iras4a ( compared to the other sources observed in the compact - north configuration ) and vice versa the slightly lower amount of recovered emission in l483 ( compared to the sources observed in the compact configuration ) are nice examples of the importance of source structure .
iras4a has significant structure within the single - dish beam , in particular by the two binary components well separated by the interferometer and each contributing to the total flux , whereas l483 typically shows a more smooth distribution of its emission over larger scales .
still , since only 1520% of the emission is recovered even in the best cases of b335 and l1157 , it is clear that care must be taken when interpreting the interferometric data .
it is , for example , not possible to deduce quantitative information from the integrated line fluxes without compensating for the lack of short spacing data .
two sources , iras4a and iras4b , show characteristic `` inverse p cygni '' line profiles where the red - shifted part of the lines is seen in absorption against the continuum while the blue - shifted part is seen in emission .
such profiles can be taken as the least ambiguous evidence for infall in protostellar cores where the foreground material falling toward the central source ( red - shifted ) is absorbing the continuum emission .
inverse p cygni profiles have been seen previously in lines of h@xmath1co and cs toward iras4a and iras4b in iram plateau de bure observations by @xcite . with the sma operating at submillimeter wavelengths ( and therefore measuring stronger continuum emission and high excitation lines )
, it should be ideally suited for detecting such line profiles .
of all the lines in the sample , only the @xmath8co 21 lines toward iras4a and iras4b show clear inverse p cygni profiles ( figure [ pcyg_spectra ] ) .
together with iras 16293 - 2422 @xcite , these sources provide the only known examples of inverse p cygni profiles toward low - mass protostars .
it could suggest that the observed inverse p cygni profiles are related to much larger scale infalling motions such as those seen in single - dish observations of iras4 by @xcite rather than collapse onto the central protostar .
the 230 ghz continuum emission toward iras4a is marginally resolved between the two components of the binary and no significant difference is seen between the widths of the absorption across the continuum possibly arguing in favor of the presence of more global collapsing motions .
of all the lines in the survey , the co 21 lines provide the most striking features , showing extended outflow emission over most of the sma primary beam ( figure [ co_maps ] ) .
the co emission has significant differences , from the highly collimated iras4b outflow to the somewhat more diffuse l483 , l1527 and b335 outflows .
the close binarity of iras2a and iras4a is also seen in the co maps causing more confusing pictures of those outflows .
iras2a for example shows hints of both a prominent north - south cone - shaped outflow and a more collimated east - west jet ( see also discussion in * ? ? ?
* ; * ? ? ?
for l1448 , iras2a and l1157 , the co emission is seen to delineate the edges of the outflow cavities .
it should be emphasized that some of the differences between outflows in this sample could be due to inclination effects .
a highly collimated outflow directed in the plane of the sky will have small relative co velocities over the extent of the outflow and its co emission will be more susceptible to resolving out as it merges with the surrounding cloud material .
l1527 shows a `` butterfly '' shape in lower-@xmath30 co and hco@xmath0 lines @xcite , suggesting that inclination is significant there .
likewise , an outflow with a large opening angle viewed at an inclination close to the opening of the outflow cones will show both red- and blue - shifted emission overlapping in both lobes , which has been shown to be the case for b335 @xcite .
recently , @xcite imaged a sample of nine class 0 , i and ii young stellar objects in the @xmath21co 1 - 0 line observed with the owens valley radio observatory millimeter array .
they found that outflow opening angles are a good discriminant between the different evolutionary stages of the objects and suggested that it reflects that the material in the protostellar envelope is cleared as the outflow opens up an increasingly wide cavity .
a similar suggestion has been put forward by @xcite .
all of the outflows in our sample are found to have small opening angles ( @xmath31 ) , further supporting the conclusion of @xcite that the outflows with the narrowest opening angles belong to the class 0 objects . the individual outflows in our sample share many of the features with the three different class 0 outflows in the sample of @xcite , ranging from the jet - like iras4b , iras4a2 and iras2a ( east - west ) outflows ( similar to the hh114-mms outflow from the sample of @xcite ) , to the more cone - like l1448 , iras2a ( north - south ) , b335 and l1157 outflows ( similar to the iras 03282 + 3035 outflow in the @xcite sample ) .
in addition the diffuse l483 and l1527 outflows appear more similar to the class i outflows in the sample of @xcite .
it is interesting to note that no significant changes in the shape of the cone is seen for the observed outflows all the way to the few hundred au scales probed by the sma observations , in contrast with what is expected if the outflow angle increases over the duration of the class 0 stage .
that is , the outflows do not show a parabolic shape reflecting a larger opening angle close to the central protostar compared to material further away - in contrast to what is seen in the perhaps best example of this scenario in b5-irs1 @xcite .
it is also interesting to note that the most collimated outflows ( iras4a and iras4b ) are related to the objects with the highest ratios of disk to envelope masses from the continuum studies while the objects with the lowest disk masses compared to envelope masses ( l1527 and l483 ) show the most diffuse outflows .
this may seem a disparity ; in a simple picture of envelope material falling into a circumstellar disk and accreting onto the central protostar , the mass of the disk is expected to increase until the envelope has dissipated ( e.g. , * ? ? ?
* ; * ? ? ?
one would therefore expect that the more evolved ( still embedded ) objects would have the highest disk - to - envelope mass ratios in contrast to what is observed here .
the question is whether l483 and l1527 still will accrete significant material from the currently observed dust envelope . as pointed out in
[ cont_morph ] , both show very flat density profiles from the scuba maps , which could reflect the importance of heating of the protostellar envelopes through their outflow cavities .
indeed , both sources show near - infrared scattering nebulosities ( e.g. , * ? ?
* ; * ? ? ?
* ; * ? ? ?
* ) , supporting the view that their envelopes are being heated at larger scales . the high angular resolution n@xmath1h@xmath0 observations of l483 by @xcite furthermore indicate a velocity gradient in the larger scale quiescent envelope material in the direction of the outflow , likely reflecting that the envelope indeed is being dispersed by the action of the outflow .
it is therefore plausible that l1527 and l483 in particular , are in a phase where the accretion has more or less come to a stop - and the protostar just is `` waiting '' to disperse its remaining envelope through the action of the outflow .
the relatively large disk masses of the remaining objects could imply that they also are close to the maximum disk masses - but comparisons to both disk and envelope structures of more evolved class i objects are needed to confirm or refute this . also , firmer constraints on both disk structures and outflow parameters for these sources will be important to obtain a more quantitative picture of the dynamical properties of the envelope and disk material .
another example of the importance of the outflows is most clearly seen in the chemical differentiation toward some of the sources . for iras4b , for example ,
a number of lines show emission separated into clumps about 6 ( 1500 au ) north / south of the central protostar , prominently seen in cs , ch@xmath2oh and h@xmath1co . these species are found to be located on the tips of the co outflows and likely probe shocks where the outflow impacts the envelope . in iras4a , characteristic shocks
( although less prominent than in iras4b ) are seen north and south in the outflow associated with the northwestern binary component , iras4a - nw ( iras4a2 ) .
fainter compact emission is also seen toward the center of this source ; interestingly , this emission appears to be red - shifted with respect to the systemic velocity , also suggesting an outflow impact on the smallest scales .
the shock - tracing sio and so emission also peak away from the central protostars in iras4a and iras4b , although only a very tentative sio emission is observed toward iras4a . for three other sources in the sample ,
compact ch@xmath2oh emission is seen , most prominently in iras2a .
in contrast to the two other sources in ngc 1333 , a number of high excitation transitions are observed in the ch@xmath2oh @xmath29 band toward this source .
this is indicative of the presence of hot gas on small scales in iras2a , consistent with the conclusion of @xcite who imaged even higher excitation ch@xmath2oh transitions also with the sma .
the three sources in ngc 1333 ( iras2a , iras4a and iras4b ) have all been suggested to have `` hot corinos '' ( e.g. * ? ? ?
* ) , i.e. , inner regions where the temperature increases above 100 k due to passive heating by the protostar and where complex organic molecules are produced into the gas - phase as ice mantles evaporate .
this was in particular inferred from multi - transition observations of h@xmath1co and ch@xmath2oh which in some cases could not be fit by constant abundances .
@xcite discussed these single - dish observations in context of the hot corino scenario and found that these sources required an abundance enhancement by up to four orders of magnitude in the innermost regions .
@xcite on the other hand demonstrated that those abundance enhancements in most cases were a result of specific assumptions , e.g. , about the ratio between the abundances ortho- and para - h@xmath1co .
furthermore , shocks caused by the outflows can enhance the abundances of these species on larger scales which further complicates the interpretation of hot corinos : @xcite , for example , found that the h@xmath1co and ch@xmath2oh abundance variations in iras4b were best explained by a scenario in which the molecules were released in the outflow driven shocks , further supported by the line profiles of these species and high excitation cs 109 lines . to properly address these issues high angular resolution observations are clearly important , e.g. , to directly image where molecular species have their origin .
@xcite for example demonstrated that the emission of so toward the inner few arcseconds of the class 0 source iras 16293 - 2422 was enhanced by the action of the outflow also traced by centimeter continuum emission and h@xmath1o maser activity .
more recently , @xcite resolved the emission of complex organic species , ch@xmath2cn and hcooch@xmath2 , toward this source and derived a size of @xmath32 au , comparable to the expected size of the hot corino .
@xcite on the other hand found that the emission of high excitation transitions and complex organic molecules toward this source was shifted by a few tenths of an arcsecond ( @xmath33 au ) coinciding with the shock seen in longer wavelength continuum observations .
the sma observations presented in this paper indicates that the peaks in the ch@xmath2oh and h@xmath1co emission toward iras4a - nw and iras4b are related to the shocks and that no trace of compact emission from these species is seen toward the central continuum sources , in contrast to what should be expected in the case of hot corinos .
we note that also for these transitions the sma resolve out a significant fraction ( 5075% ) of the observed single - dish flux indicating that h@xmath1co and ch@xmath2oh are present in the large scale envelopes .
iras2a still stands out with compact hot gas as witnessed , e.g. , by the detection of many of the higher excitation hyperfine components of the @xmath34 line ( fig .
[ ch3oh_spectra ] and higher @xmath30 ch@xmath2oh lines @xcite that appear unresolved . of the well - studied low - mass protostars ,
iras2 therefore appears to be the best hot corino candidate .
given the importance of outflows in the other low - mass sources where complex organic molecules have been observed , however , it is a valid question to ask whether the protostellar outflow is important on these small scales in iras2a , where it is still unresolved by the sma observations ( similar to the case of iras 16293 - 2422 ) .
the so emission do show evidence of a velocity gradient which could suggest that the shocks are impacting the material on small scales toward iras2a .
also , it is possible that accretion shocks in the disk could temporarily heat the gas in the surface layer to high temperatures and likewise inject complex molecules from the ice mantles into the gas - phase . in iras2a in particular
, this seems to be a feasible scenario , given the extent of the disk from submillimeter continuum data ( see discussion in * ? ? ?
again , further high - resolution line observations could potentially disentangle the velocity field of the disk and further address its physical properties ( temperature , density ) and chemical structure .
this discussion directly illustrates the importance of high - angular resolution imaging for the interpretation of chemical variations in protostellar environments . with the high - resolution sma observations , it becomes possible to image directly these physical and chemical variations in the immediate vicinity of the low - mass protostars
a large line and continuum survey with the submillimeter array of 8 deeply embedded , low - mass protostellar cores has been presented .
each source has been observed in three different spectral settings that included lines of the most common molecular species , co , hco@xmath0 , cs , so , h@xmath1co , ch@xmath2oh and sio . emission from ( sub)millimeter lines from 11 molecular species ( including isotopologues ) have been imaged at high angular resolution ( 1@xmath33@xmath3 ; typically corresponding to 200600 au ) together with continuum emission at 230 ( 1.3 mm ) and 345 ghz ( 0.8 mm ) .
a number of conclusions can be inferred from simple considerations of this dataset : * compact continuum emission is observed for all sources . comparing the 230 ghz and 345 ghz data ,
the emission is found to be well described as @xmath35 with an average value of @xmath36 of 2.4 .
@xmath37 was only found for one source , l483 . for this source ,
the compact continuum emission is weak at 230 ghz and the emission comes from the envelope rather than the circumstellar disk . for the remaining sources , the continuum emission likely originates in the disks being marginally optically thick and having dust opacities , @xmath38 with @xmath7 .
lower limits to the total disk masses ( gas+dust ) are found to be typically 0.1 @xmath5 for a dust temperature of 30 k ( 110% of their envelope masses ) . *
inverse p cygni profiles indicative of infalling motions are seen in the @xmath8co 21 lines toward iras4a and iras4b .
these are not observed in other lines or toward other sources ( besides iras 16293 - 2422 ) , however , and it is possible that these profiles have their origin in the ambient cloud / outer envelope rather than infall in the inner envelope themselves .
* prominent outflows are present in all sources extending over most of the interferometer field of view .
most of the outflows show small opening angles @xmath31 comparable to other previously studied class 0 objects - and in agreement with the suggestion that the outflow opening angle increases with time @xcite .
still , significant differences are seen in their morphologies with some outflows showing more jet - like structure and others seemingly tracing material in the outflow cavity walls .
the most diffuse outflows are found for the objects with the lowest ratios of disk - to - envelope masses , in particular l1527 and l483 , which are thought to be transitionary between the class 0 and i stages .
it is suggested that these sources have reached the end of their main accretion and that the remaining envelope material will be dispersed by the outflows .
* shocks are present on all scales in the protostellar environments , most clearly traced by extended emission of ch@xmath2oh in iras4a and iras4b .
compact red - shifted ch@xmath2oh emission is seen toward one of the components of the iras4a binary , iras4a - nw ( iras4a2 ) .
these observations suggest that the emission of h@xmath1co and ch@xmath2oh , often taken as signposts of low - mass `` hot corinos '' , are in fact related to the shocks caused by the protostellar outflows . in this context ,
iras2a stands out as the only source with clear evidence for hot , compact gas ( as also seen in previous sma observations by @xcite ) .
further high resolution observations will be necessary to ascertain whether this emission is related to a passively heated hot corino , a shock in the circumstellar disk or perhaps just the action of the outflow on the smallest scales .
it is clear that these high - resolution observations pushes our understanding of embedded protostars as simple and quiescent structures toward a more dynamic picture .
outflows play an important role in all aspects of the structure of these low - mass protostellar systems , from the larger scale swept up material clearly probed by the co outflows down to the smallest ( arcsecond ) scales where shocks are present .
it is important to realize that the larger scale envelope carrying the bulk of the material probed by single - dish observations is almost completely resolved out by the sma observations . using the sma data in conjunction with single - dish data and/or detailed models for the source structures
, it will become possible to make quantitative statements , e.g. , about the physical structure of the envelopes or the chemical abundances from scales of thousands of au down to just the few hundred au probed by the interferometers .
more detailed discussions of aspects of star formation based on the data presented in this paper , in addition to such single - dish data and detailed radiative transfer models , will be presented in a series of upcoming specialized papers .
it is a pleasure to thank everybody involved with the submillimeter array for the continued development of this instrument .
we are also grateful to the observers who have contributed by observing one or more tracks in this program .
the research of j.k.j . was supported by nasa origins grant nag5 - 13050 .
f.l.s . acknowledges support from the swedish research council .
astrochemistry in leiden is supported by a nwo spinoza grant and a nova grant .
l1448-c(n ) is one of @xmath3910 young stellar objects in the l1448 complex in the perseus molecular cloud .
recent spitzer observations @xcite have resolved it into two components with a separation of about 8 ( 2000 au ) .
the northern of these two is known from millimeter aperture synthesis observations ( e.g. , * ? ? ?
* ; * ? ? ?
* ) which is taken as the pointing center for the observations in this paper .
ngc 1333-iras2a is part of a protostellar system also encompassing the protostar iras2b and likely pre - stellar core iras2c seen in scuba maps @xcite .
both iras2a and iras2b have been imaged at high angular resolution at millimeter wavelengths @xcite whereas iras2c does not show any compact continuum emission .
iras2b is located outside the primary beam field of view at a separation of 30@xmath3 ( 7500 au ) , but is detected in the 1.3 mm maps presented here .
iras2a itself has been speculated to be a protostellar binary , resulting in its characteristic quadrupolar outflow ( see also [ cooutflow ] ) . at 0.8 mm
, it appears to be resolved into two separate components with a fainter companion northeast of the main peak , but as for l1448 above , it is not clear whether this is simply a result of resolving out differing amounts of envelope material .
@xcite examined sma continuum observations from iras2a and showed that the emission could be explained by a combination of an extended envelope and a 300 au diameter disk .
ngc 1333-iras4a ( together with its companion iras4b ; both well - separated in scuba maps ) is one of the most well - studied protostellar systems in ngc 1333 .
@xcite detected inverse p cygni profiles indicative of infalling motions in lines of h@xmath1co toward these sources .
iras4a is resolved into two binary components , ngc 1333-iras4a - nw and -iras4a - se with a separation of 1.8@xmath3 ( 400 au ) @xcite .
ngc 1333-iras4b is located about 30 southeast of ngc 1333-iras4a and also shows signs of inverse p cygni profiles @xcite .
a fainter companion ngc 1333-iras4b@xmath40 ( also sometimes referred to as ngc 1333-iras4c or ngc 1333-iras4b2 ) is located another 10 to the east @xcite .
the outflow in iras4b appears to play an important role of 510 scales affecting the emission of a number of molecular species .
@xcite for example found that the abundance variations of h@xmath1co and ch@xmath2oh deduced from single - dish observations best were explained by the impact of this outflow - also accounting for the line profiles of these species and high excitation cs 109 emission detected toward this source .
l1527-irs ( iras 04368 + 2557 ) is one of the possibly more evolved protostars in the sample located in a dense core in taurus .
l1527-irs was previously studied at high angular resolution with the nobeyama millimeter array @xcite and the owens valley radio observatory millimeter array ( ovro ; * ? ? ?
* ; * ? ? ?
* ) and found to show a characteristic x - shaped structure in lines of @xmath8co 10 and hco@xmath41 that likely delineates the walls of an outflow cavity . it was suggested to be a binary based on 800 @xmath42 m continuum observations @xcite with the secondary at about 21from the main protostar .
no evidence is found of this secondary in the previous high resolution 3 mm maps or in the maps presented here , however ( it would be located outside the primary beam field of view at 345 ghz , but inside at 230 ghz ) .
the scuba maps show a very extended core with little central concentration ( a flat envelope density profile @xmath43 with @xmath44 is found for this source @xcite whereas most of the other protostars have steeper profiles with @xmath45 ) .
l483-mm is a strong infrared source ( iras 18148 - 0440 ) associated with an isolated core .
@xcite suggested it was transitionary between the class 0 and i stages .
it has been imaged at high resolution in continuum and lines at 3 mm with the berkeley maryland illinois association ( bima ) millimeter array @xcite and ovro mma @xcite . in the analysis of the continuum emission @xcite found that all the detected continuum emission could be attributed to the extended envelope , contrasting l483-fir from l1448-c and ngc 1333-iras2a described above where a compact continuum source was present .
@xcite recently mapped l483-mm in the hcn 43 transition at submillimeter wavelengths using the aste single - dish telescope .
they found that the hcn emission was slightly extended , which they argued is reflecting the heating of the envelope material to 4000 au scales through the outflow cavity .
b335 has often been considered one of the best protostellar collapse candidates .
it has been imaged in continuum at subarcsecond resolution @xcite and in lines @xcite .
@xcite found that the continuum emission could be well described by a @xmath46 envelope down to scales of @xmath32 au within which a disk with a size @xmath47 100 au is seen .
@xcite showed that the line emission thought to trace the collapse were in fact dominated by a number of clumps extending along the outflow cavity into the low density medium .
l1157-mm is perhaps best known for its prominent outflow @xcite , but also its small - scale envelope / disk structure has gotten a fair amount of attention . @xcite
observed l1157 in lines of co ( and isotopologues ) and continuum using the iram plateau de bure interferometer ( pdbi ) .
they found that the continuum emission probed a combination of a large disk and an extended envelope with a clear outflow cavity whereas the co isotopologues probed a progression from the envelope , throughout its cavities to the outflow itself . @xcite and @xcite observed transitions of ch@xmath2oh at 3 mm and 1.3 mm using ovro , suggesting that these ch@xmath2oh transitions had their origin in a warm layer of the disk heated by an accretion shock . , e. f. , jrgensen , j. k. , maret , s. , ceccarelli , c. , caux , e. , schier , f. l. , castets , a. , & tielens , a. g. g. m. 2005 , in the dusty and molecular universe : a prelude to herschel and alma , ed
a. wilson , 191 | this paper presents a large spectral line and continuum survey of 8 deeply embedded , low - mass protostellar cores using the submillimeter array .
each source was observed with three different spectral settings , to include high excitation lines of some of the most common molecular species , co , hco@xmath0 , cs , so , h@xmath1co , ch@xmath2oh and sio . line emission from 11 molecular species ( including isotopologues ) originating from warm and dense gas have been imaged at high angular resolution ( 13@xmath3 ; typically corresponding to 200600 au scales ) together with continuum emission at 230 ghz ( 1.3 mm ) and 345 ghz ( 0.8 mm ) .
compact continuum emission is observed for all sources which likely originates in marginally optically thick circumstellar disks , with typical lower limits to their masses of @xmath4 @xmath5 ( 110% of the masses of their envelopes ) and having a dust opacity law , @xmath6 , with @xmath7 .
prominent outflows are present in co 21 observations in all sources , extending over most of the interferometer field of view .
most outflows are highly collimated .
significant differences are seen in their morphologies , however , with some showing more jet - like structure and others seemingly tracing material in the outflow cavity walls .
the most diffuse outflows are found in the sources with the lowest ratios of disk - to - envelope mass , and it is suggested that these sources are in a phase where accretion of matter from the envelope has almost finished and the remainder of the envelope material is being dispersed by the outflows .
other characteristic dynamical signatures are found with inverse p cygni profiles indicative of infalling motions seen in the @xmath8co 21 lines toward ngc 1333-iras4a and ngc 1333-iras4b .
outflow - induced shocks are present on all scales in the protostellar environments and are most clearly traced by the emission of ch@xmath2oh in ngc 1333-iras4a and ngc 1333-iras4b .
these observations suggest that the emission of ch@xmath2oh and h@xmath1co from these proposed `` hot corinos '' are related to the shocks caused by the protostellar outflows .
one source , ngc 1333-iras2a , stands out as the only one remaining with evidence for hot , compact ch@xmath2oh emission coincident with the embedded protostar . |
the search for higgs bosons has been one of the most important lines of research during the lep era .
final lep-1 results from data taken at the z - resonance @xcite and preliminary lep-2 results for complete data up to 209 gev center - of - mass energy @xcite were recently reviewed .
the mssm is well motivated and the most discussed extension of the sm .
it predicts three neutral(h , h , and a ) and two charged higgs bosons . of lep operation in 1989 , no radiative corrections to the neutral higgs boson masses were calculated and @xmath0 was predicted .
second - order corrections increased the upper mass bound to about 130 gev .
the neutral higgs boson masses and production cross sections strongly depend on various model parameters . at lep
two ways were chosen to present the mass limits : benchmark limits for a certain set of mssm parameters and mass limits based on mssm parameter scans .
details of the benchmark and scan parameters are given for example in refs .
the importance of a mssm parameter scan to set mass limits was already realized at lep-1 .
the region marked by the thick black line in fig .
[ fig : lep1 ] shows that the h mass limit is reduced from 41 to 25 gev in a parameter scan compared to benchmark results @xcite .
an early lep-2 study showed that a benchmark limit of 52 gev on the a mass @xcite disappeared completely in a mssm parameter scan @xcite ( fig .
[ fig:172 ha ] ) .
result.,scaledwidth=43.0% ] result .
only the dark region is excluded in addition to lep-1.,scaledwidth=40.0% ]
the results from very different parameter scan methods of different lep experiments agree well as shown in figs .
[ fig:189opalha ] and [ fig:189delphiha ] for 189 gev data @xcite .
the shaded region in fig .
[ fig:189opalha ] is excluded by charge- and color - breaking ( ccb ) criteria @xcite .
the dotted line indicates the corresponding benchmark limit.,scaledwidth=47.0% ] result .
the dotted line indicates the corresponding benchmark limit.,scaledwidth=47.0% ] limits from charmless b - decays , the electroweak parameter @xmath1 and direct searches for supersymmetric particles do not change the h and a mass limits @xcite .
the parameter scan reduced the benchmark limits by up to 7 gev @xcite .
the constraints on @xmath2 are given in figs .
[ fig:189delphiatgb ] and [ fig:189delphihtgb ] .
result.,scaledwidth=47.0% ] result.,scaledwidth=47.0% ]
with increasing center - of - mass energy , benchmark and parameter scan limits agree within 1 gev for 202 gev data @xcite .
the mass and @xmath2 limits are given in figs .
[ fig:202delphiha ] , [ fig:202delphiatgb ] and [ fig:202delphihtgb ] .
result.,scaledwidth=50.0% ] result.,scaledwidth=50.0% ] result.,scaledwidth=50.0% ]
the mass limits from benchmark and parameter scan agree within about 2 gev in the combination of the 202 gev data from all lep experiments @xcite .
the scan limits are given in figs .
[ fig : lep202delphiha ] , [ fig : lep202delphiatgb ] and [ fig : lep202delphihtgb ] .
result.,scaledwidth=47.0% ] result.,scaledwidth=47.0% ] result.,scaledwidth=47.0% ]
the latest benchmark and scan results including 209 gev data agree within 2 gev @xcite and give mass limits on h and a of 89 gev ( figs .
[ fig:209delphiha ] , [ fig:209delphiatgb ] and [ fig:209delphihtgb ] ) .
the importance of a mssm parameter scan is underlined in fig .
[ fig:209inv ] , which shows that large parameter regions exist where a higgs boson could decay invisibly into a pair of neutralinos .
thus , only in conjunction with a dedicated search @xcite the h and a mass limits are set .
result.,scaledwidth=47.0% ] result.,scaledwidth=47.0% ] result.,scaledwidth=47.0% ] result from searches for invisibly decaying higgs bosons.,scaledwidth=47.0% ]
for 202 gev data , taken in 1999 , and first 2000 data , h and a mass limits were 2 gev below expectation @xcite . this tendency is enhanced by including complete 209 gev data , in which case the limits are 3.1 to 3.6 gev below the expectations of about 95 gev @xcite .
a possible explanation is that the hz excess at about 115 gev is due to the heavier scalar h and that , in addition , the production of ha with masses between 90 and 100 gev occurs @xcite .
figure [ mssm - bbbb ] ( from @xcite ) shows a data excess above @xmath3 for @xmath4 gev in the channel .
the same data excess is also expressed by the confidence level @xmath5 for a signal observation as shown in fig .
[ mssm - diag ] ( from @xcite ) .
the hypothesis of the production of three mssm higgs bosons is supported by the data excess seen in fig .
[ sm - mass ] ( from @xcite ) at 100 gev which could result from hz production in addition to hz production . for the reported mssm parameters @xcite @xmath6
; therefore @xmath7 .
the @xmath8 limit in the 100 gev mass region shows a deviation of about 2@xmath9 between the expected and observed limit , as seen in fig .
[ sm - xi-1 ] ( from @xcite ) .
figure [ sm - xi-2 ] shows that this new support is only observed in the complete lep data . at 95% cl ( @xmath10 ) for the mssm processes ha
this corresponds also to limits on @xmath11 in the general extension of the sm with two higgs boson doublets .
the data of the four lep experiments collected at energies from 88 to 209 gev are combined .
the solid curve is the observed result and the dashed curve shows the expected median .
shaded areas indicate the @xmath12 and @xmath3 probability bands .
[ mssm - bbbb],scaledwidth=47.0% ] as a function of @xmath13 for the case @xmath10 ( where only the ha process contributes since @xmath14 ) .
the straight line at 0.5 and the shaded @xmath12 and @xmath3 probability bands represent the expected background - only result .
the solid curve is the observed result and the dashed curve shows the expected median for a signal .
[ mssm - diag],scaledwidth=47.0% ] , scaledwidth=47.0% ] as a function of , where @xmath15 is the hzz coupling relative to the sm coupling . about @xmath3 deviations from the expectation
are observed at @xmath16 gev and @xmath17 gev . in the mssm ,
hz production at the lower mass and hz production at the higher mass are possible .
[ sm - xi-1],scaledwidth=47.0% ] region including 209 gev data is compared with the results from combined lep-1 data @xcite , taken around 91 gev center - of - mass energy , and previous lep-2 limits @xcite up to 183 , 189 , 196 and 202 gev .
the @xmath8 limit below 100 gev does not become significantly stronger when the 209 gev data , taken in 2000 , are included .
[ sm - xi-2],scaledwidth=47.0% ]
the lep-1 and lep-2 data are consistent with a background - only hypothesis and give stringent mass limits on the neutral higgs bosons h and a of the mssm . during the lep era , mssm parameter scans reduced or for early lep-2 data even removed completely the benchmark mass limits .
the importance of parameter scans is stressed by large parameter regions where the higgs boson decays invisibly , which is not considered in benchmark results . at the highest center - of - mass energies ,
benchmark limits are only slightly reduced by a general parameter scan when the results from invisible higgs boson searches are included .
table [ tab : comp ] compares benchmark and scan mass limits in the mssm .
the combined lep data show a preference for the sm higgs boson of 115.6 gev , which can also be interpreted as a preference for a higgs boson of that mass in the mssm .
further small data excesses for higgs boson pair - production and bremsstrahlung between 90 and 100 gev allow the hypothesis that h , a and h of the mssm all have masses below 116 gev .
previously reported mssm parameter combinations from a general parameter scan for this scenario are supported by the complete data set .
i would like to thank the organizers of the conference for their kind hospitality .
9 a. sopczak , phys .
rep . , higgs physics at , in press .
a. sopczak , ` complete lep data : status of higgs boson searches ' , nanp-1 , moscow , june 2001 , proc . in press , hep - ph/0112082 .
aleph , delphi , l3 and opal coll . and
the lep working group for higgs boson searches , lhwg note/2001 - 04 .
delphi coll.,delphi 2001 - 81conf-509 , eps hep 2001(budapest ) and lp01(rome ) .
a. sopczak , ph.d .
thesis , univ . of california ,
san diego ( 1992 ) ; + l3 coll .
* b 294 * ( 1992 ) 457 ; + z. phys .
* c 57 * ( 1993 ) 355 .
j. rosiek and a. sopczak , phys . lett .
* b 341 * ( 1995 ) 419 .
delphi coll .
j. * c 2 * ( 1998 ) 1 .
a. sopczak , eur .
j. * c 9 * ( 1999 ) 107 .
opal coll .
j. * c 12 * ( 2000 ) 567 .
delphi coll .
, eur.phys.j.*c 17*(2000)549 .
delphi coll .
, eur.phys.j.*c 17*(2000)187 .
delphi coll .
j. * c * , cern - ep/2001 - 87 .
aleph , delphi , l3 and opal coll . and
the lep working group for higgs boson searches , aleph 2000 - 074 conf 2000 - 051 , delphi 2000 - 148 conf-447 , l3 note 2600 , opal tn 661 .
a. sopczak , hep - ph/0004015 , pascos-99 , ( world scientific , singapore , 2000 ) , p. 511 .
delphi coll.,delphi 2001 - 78conf-506 , eps hep 2001(budapest ) and lp01(rome ) .
delphi coll.,delphi 2001 - 79conf-507 , eps hep 2001(budapest ) and lp01(rome ) .
a. sopczak , hep - ph/0011285 , proc .
dpf-2000 , columbus , ohio , aug .
912 , 2000 , http://www.dpf2000.org/bsm1.htm#ses4 .
aleph , delphi , l3 and opal coll . and the lep working group for higgs boson searches , cern - ep/2001 - 55 .
a. sopczak , desy 97 - 129 , indian j. phys . * a 72 * ( 1998 ) 495 .
delphi coll .
, delphi 98 - 95 conf-163 , ichep 1998 ( vancouver ) .
delphi coll .
, delphi 98 - 124 conf-185 , ichep 1998 ( vancouver ) . | the lep era has brought immense progress in searches for higgs bosons over the last 12 years which will guide searches at future colliders .
the evolution of the higgs boson mass limits is reviewed with the focus on results from general parameter scans in the minimal supersymmetric extension of the standard model ( mssm ) in contrast to the so - called benchmark limits .
the hint for a standard model ( sm ) higgs boson of 115.6 gev can also be interpreted as a preference for a higgs boson of that mass in the mssm .
further small data excesses allow the hypothesis that the neutral higgs bosons of the mssm all have masses between 90 and 116 gev .
-1.0 cm + 3.0 cm * mssm constraints from higgs boson searches * + lancaster university 2.5 cm * abstract * 3.cm ( 0.001,0.001)(0,0 ) ( , 0 ) the lep era has brought immense progress in searches for higgs bosons over the last 12 years which will guide searches at future colliders .
the evolution of the higgs boson mass limits is reviewed with the focus on results from general parameter scans in the minimal supersymmetric extension of the standard model ( mssm ) in contrast to the so - called benchmark limits .
the hint for a standard model ( sm ) higgs boson of 115.6 gev can also be interpreted as a preference for a higgs boson of that mass in the mssm .
further small data excesses allow the hypothesis that the neutral higgs bosons of the mssm all have masses between 90 and 116 gev . _ presented at the seventh topical seminar on the legacy of lep and slc , siena , italy , october 2001 _ |
Subsets and Splits
No saved queries yet
Save your SQL queries to embed, download, and access them later. Queries will appear here once saved.